Method for accessing at least one decentral network

Abstract

This disclosure relates to the field of sustainability, in particular to the field of sustainable industrialization. The disclosure relates to methods, apparatuses, systems, and computer elements for accessing at least one decentral network.

Description

METHOD FOR ACCESSING AT LEAST ONE DECENTRAL NETWORKTECHNICAL FIELD

[0001] The invention relates to the field of sustainability, in particular to the field of sustainable industrialization. The following disclosure relates to methods, apparatuses, systems, and computer elements for accessing at least one decentral network.TECHNICAL BACKGROUND

[0002] In the supply and pcroduction of products multiple regulatory requirements need to be met, which differ depending on the product. To fulfil such regulatory requirements, data on such products may need to be exchanged between different participants involved in the production and use of such products. Such data may be exchanged in a secure and controlled manner within a decentral network connecting different participants involved in the production and / or recycling of the product. Due to the high complexity accessing these decentral networks may be cumbersome, especially for various supply chain participants. Further, output materials may be used in different product ecosystems, e.g. for production of different (end) products, rendering it challenging e.g. for smaller supply chain participants to provide the data on their output product to the further participants of the respective product ecosystems via the correct decentral network.SUMMARY

[0003] In a first aspect disclosed is a method for accessing at least one decentral network to provide and / or obtain (e.g. consume) a product data set associated with an output product, wherein the output product is produced or producible from one or more input materials by a production environment, the method comprising the steps: obtaining, from an operator of the production environment, a user instruction including (e.g. natural language and) at least data indicative of the production environment, the output product, and at least one other production environment, the at least one other production environment configured to produce a product based on the output product and / or the at least one other production environment providing input material for producing the output product; determining, based on the user instruction, a tool sequence, wherein a tool sequence comprises an indication of one or more tool(s), wherein the one or more tools comprises at least one agent(s) configured to generate at least one data provision instruction for accessing the at least one decentral network, wherein determining the tool sequence comprises providing a task instruction, based on the user instruction, to at least one generative data-driven model, the at least one generative data-driven model beingconfigured to generate the tool sequence in response to obtaining the task instruction; carrying out or causing to carry out the one or more tool(s) comprising the at least one agent(s) according to the tool sequence; receiving output data from the one or more one or more tool(s), the output data comprising at least a part of the at least one data provision instruction for accessing the at least one decentral network; providing the at least one data provision instruction; receiving, based on the at least one data provision instruction, an operation data set related to the at least one data provision instruction; determining, based on receiving the operation data set, at least one of an identification associated with the production environment or security credential(s) for accessing the at least one decentral network for obtaining and / or providing the product data set; and providing, to the operator, at least one of the identification associated with the production environment or the security credential(s).

[0004] In a second aspect disclosed is a method for generating a digital product passport associated with a product, the method comprising: accessing a decentral network according to a method and obtaining, via the decentral network, a product data set associated with an output product, wherein the output product is input material for producing the product); providing product data associated with the product; providing a decentral identifier associated with the product data; generating the digital product passport including the decentral identifier and at least part of the product data; providing the digital product passport for access by a decentral network node associated with a product consumer under control or controlled by a decentral network node associated with a data owner of the product passport.

[0005] In a third aspect disclosed is a method for accessing a digital product passport associated with a product, the method comprising: accessing a decentral network according to a method of the first aspect; providing a request to access the digital product passport, the request including a decentral identifier associated with the passport; and receiving the digital product passport based on providing the request.

[0006] In a fourth aspect disclosed is an apparatus comprising respective means for carrying out or performing the steps of the method according to the first, second and / or third aspect or comprising at least one processor and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus at least to carry out the steps of the method according to the first, second and / or third aspect.

[0007] According to further aspects, respective apparatus, system, and use are disclosed.EMBODIMENTS

[0008] Any disclosure, embodiments and examples described herein relate to the method, the aspects, e.g. system, apparatus, product (e.g. chemical product) and computer element lined out above and below. Advantageously, the benefits provided by any of the embodiments and examples may equally apply to all other embodiments and examples. In the following, embodiments of the present disclosure will be outlined by ways of embodiments and / or example. It is to be understood that the present disclosure is not limited to said embodiments and / or examples. All terms and definitions used herein are understood broadly and have their general meaning if not specified otherwise.

[0009] The methods, the apparatuses, the systems, the uses, and the computer elements disclosed herein may enable or improve the exchange of output product data associated with output product(s) used as input material to produce one or more product(s) e.g. to generate product passports for the one or more product(s).

[0010] An efficient, secure and robust way for sharing or exchanging product or property data across different participant nodes in value chains may be provided by allowing operators to quickly and reliably access a decentral network. In particular, by providing supply chain product specific data via the supply chain product specific passport or digital asset, properties like environmental impact can be shared and made transparent from the material to the product produced from such material. A generated passport or digital asset may enable secure data exchange, since data access can be controlled by the supply chain product provider. The exchanges of data assets can be specific to the supply chain product as produced and tailored to the needs of the supply chain product user. This way an improved tracking and tracing of supply chain products can be achieved by securely providing product or property data in diverse and highly complex value chains. The properties can hence be tracked leading to simpler, more efficient and sustainable handling of supply chain products by value chain participants.

[0011] The disclosure particularly relates to the use of generative artificial intelligence in the context of decentral networks. As more and more supply chains move towards decentral data transfers multiple decentral network protocols emerge. The decentral network protocols may be specific for the supply chain, the tier of the supply chain, the industry, the supply chain products, the end products of the supply chain or any intermediate product of specific supply chains. In essence these protocols define communication standards that may differ depending on the decentral network. One part of the communication standard may relate to data models for describing the digital twin of the respective supply chain product. The supply chain product may relate to physical products produced by the supply chain players. The supply chain product may also relate to intangible products such as energy that is distributed over the grid and whereenergy provided to the grid may be tracked in a book and claim manner. Products may be used in different value chains and hence associated product data sets may need to be provided in different decentral network, which may require different data provided by e.g. an operator of a production environment. For instance, a different technical infrastructure may be needed for accessing these networks. Product data sets may be differently structured for instance. Using a generative data-driven model together with a template database may allow an operator to identify the correct decentral network to provide the product data to and may help to identify which data needs to be provided and may validate the provided data.

[0012] To orchestrate or manage multiple protocol standards associated with multiple decentral networks and to reliably track and trace products across their respective supply chain, multiple communication protocols need to be fulfilled by the digital assets or product data sets generated for such supply chain products. The diversity and regular changes make reliable operation according to such protocol standards cumbersome. Owing to the regulatory need to provide supply chain products with their unique digital assets, reliability to orchestrate or manage the multiple protocol standards becomes even more important. A Large Language Model (LLM) based access may enhance the flexibility of the system, since the selection from the configuration base can be executed even if the user of the system cannot accurately specify the decentral network or the data points needed to successfully onboard to the decentral network. A stepwise enabled onboarding process makes operating in multiple decentral networks with diverse protocols simpler and hence product tracking in multiple supply chains more reliable.

[0013] Through simple and convenient interaction with a natural language enabled system, operations data of the distributed chemical production environment can be made accessible in a targeted manner to operators for generating product passports. Embodiments of this disclosure may enable robust and reliable onboarding of supply chain participants by utilizing a large language model to assist operators through the onboarding process. This may significantly reduce the complexity associated with accessing decentral networks, thereby facilitating secure and controlled data exchange between different participants involved in the production and / or recycling of products. The performance monitoring unit, including a data logging component, may provide stable and continuous tracking of data exchanges, ensuring that all regulatory requirements are met efficiently. This may enhance the monitoring and controlling of production processes, leading to improved compliance with regulatory standards. By integrating a large language model, the embodiment may offer a user-friendly interface that guides operators through the intricate steps required to connect to the appropriate decentral networks. This maybe particularly advantageous for smaller supply chain participants, who may find it challenging to navigate the complexities of multiple product ecosystems. Furthermore, the secure and controlled manner of data exchange facilitated by the embodiment may contribute to a reduction in environmental impact by ensuring that data related to the recycling and reuse of materials is accurately and efficiently shared. This may promote sustainable practices within the supply chain. Overall, the embodiment may provide a robust and stable solution for managing the complexities of regulatory compliance and data exchange in the production and recycling of products, thereby enhancing the reliability and efficiency of supply chain operations.

[0014] According to an example embodiment, determining the tool sequence further comprises at least the following steps: providing a task instruction for generating the tool sequence to a generative data-driven model, wherein the task instruction is generated based on the user instruction and the set of tool data sets, the generative data-driven model (e.g. by having been trained on general purpose training data sets) being configured to generate a tool generated data set related to the tool sequence, in response to receiving the task instruction; providing the tool generated data set as the tool sequence or in case the tool generated data set is not of the same format of type of a tool sequence: parsing or causing to parse the tool generated data set into the tool sequence and providing the tool sequence. A task instruction for generating the tool sequence may be generated based on the user instruction and the set of tool data sets. Parsing the tool generated data set may be based on using a rule-based engine, that is configured e.g. to structure the tool generated data set in specific order or format based e.g. on certain keywords. Parsing the tool generated data set may be based on instructing a generative data-driven model to generate an output of a certain type or format based on the tool generated data set. e.g. by generating a task instruction based on the user instruction, and providing the task instruction to a generative data-driven model that has been trained on general-purpose data comprising natural language, wherein the generative data-driven model generates and provides the tool sequence. The tool generated data set related to the tool sequence may be the tool sequence, for instance the respective task instruction for generating the tool generated data set may comprise an instruction to generate a tool sequence e.g. according to a certain format or type and the task instruction may further comprise e.g. at least one example of at least a part of a tool sequence. The generative data-driven model for generating the tool generated data set or tool sequence may e.g. be GPT4o (e.g. gpt-4o-2024-08-06 or later), GPT5, Gemini 3 or a large language model having at least similar capabilities. Such a generative data-driven model may e.g. be configured to provide structured outputs so that the model adheres to a given output structure. A generative data-driven model may have a response format parameter, which maybe set so that the output of the model matches a provided format or schema (e.g. a JSON schema).

[0015] According to an example embodiment, determining the tool sequence comprises:

[0016] retrieving a set of tool data sets, wherein each tool data set of the set of tool data sets is associated with a tool of the at least one tool; wherein the determining the tool sequence is further based on the set of tool data sets.

[0017] According to an example embodiment, determining the tool sequence comprises: determining whether input data required for carrying out the at least one tool is provided by the user instruction; upon determining that the input data required for carrying out the at least one tool is provided by the user instruction: including the at least one tool in the tool sequence to be carried out based on the input data.

[0018] According to an example embodiment, determining the tool sequence comprises: providing a task instruction for generating the tool sequence to a generative data-driven model, wherein the task instruction is generated based on the user instruction and a set of tool data sets, the generative data-driven model trained or having been trained in particular on general purpose training data sets.

[0019] According to an example embodiment, upon determining that the input data required for carrying out the at least one tool is not provided by the user instruction: determining the tool sequence, the determining comprising: retrieving a set of tool data sets, wherein each tool data set of the set of tool data sets is associated with a tool of the at least one tool; identifying, based on the set of tool data sets, a tool providing the input data used by the at least one tool; introducing said tool in the tool sequence before the at least one tool.

[0020] According to an example embodiment, a tool sequence is determined or generated using a generative data-driven model in particular a generative data-driven model trained on general-purpose training data sets or a generative data-driven model based on (e.g. distilled from) a generative data-driven model trained on general-purpose training data sets. Preferably, determining the tool sequence comprises: providing a task instruction for generating the tool sequence to a generative data-driven model, wherein the task instruction is generated based on the user instruction and the set of tool data sets, the generative data-driven model having been trained on general purpose training data sets and being configured to generate a tool sequence or a tool output data set related to the tool sequence, in response to receiving the task instruction.

[0021] According to an example embodiment, upon receiving output data from the at least one tool, at least a part of the output data is provided to a / the generative data-driven model.Preferably a decision task instruction, based on or comprising the user instruction, at least a part of the tool sequence and at least a part of the output data, is provided to the generative data-driven model for determining whether the output data is suitable to determine a data provision instruction.

[0022] According to an example embodiment, upon determining that the putput data is not suitable to determine a data provision instruction: determining or re-determining the tool sequence, the (re-)determining comprising: retrieving a set of tool data sets, wherein each tool data set of the set of tool data sets is associated with a tool of the at least one tool; identifying, based on the set of tool data sets, a tool providing the input data (e.g. according to the input data format, input data type and / or input data structure) used by the at least one tool; introducing said tool in the tool sequence before the at least one tool.

[0023] The method may further comprise: Carrying out or causing to carry out the at least one tool according to the re-determined tool sequence or carrying out the at least one further tool; and Receiving output data from the at least one tool of the re-determined tool sequence or the at least one further tool; and Determining the at least one product data set based on at least a part of the output data from the at least one tool of the re-determined tool sequence or the at least one further tool.

[0024] According to an example embodiment, the method further comprises or the step of carrying out or causing to carry out the one or more tool(s) comprises, carrying out or causing to carry out a first tool according to the tool sequence; receiving output data from a first tool indicated in the tool sequence; upon receiving output data from the first tool, determining an operation instruction to a second tool indicated in the tool sequence, e.g. using a / the generative data-driven model, for instance by providing a task instruction for generating the operation instruction to the generative data-driven model, this task instruction comprising at least a part of the output data from the first tool; Carrying out or causing to carry out the second tool according to the tool sequence based on the operation user instruction. This process may be repeated for some or all tools according to the tool sequence, taking for instance into account at least part of the output data from previous tools for carrying out subsequent tools according to the tool sequence.

[0025] According to an example embodiment, at least a part of the at least one data provision instruction is determined or generated using a generative data-driven model in particular a generative data-driven model trained on general-purpose training data sets or a generative data- driven model based on (e.g. distilled from) a generative data-driven model trained on general- purpose training data sets.

[0026] A computational agent or short agent may be a self-operating computational unit, e.g. a software component or system that can independently, i.e. without further user interaction, execute tasks or operations. This execution may involve processing input data, applying predefined algorithms, and making decisions based on the results. It may also include the ability to adjust or modify its operations in response to changes in input data or outcomes of its operations. The self-operating computational unit may carry out these tasks without the need for continuous human supervision or intervention, although human input may be incorporated as part of its decision-making processes or to change its operational parameters. It should be understood that the specific functionalities, operations, and level of independence of the selfoperating computational unit can vary based on the design and requirements of the specific system it is implemented in. A tool is in particular an electronic tool or operating engine and may be a computational agent or function, e.g. configured to calculating a weighted average of some quantities or any other certain form of aggregation. A computational agent or function may e.g. be included in the tool sequence if the user instruction related to post-processing.

[0027] A tool sequence may be an ordered list of one or more tool(s) that should be executed in the given order using certain input data that may be provided by the user instruction, another tool that is carried out earlier according to the tool sequence, or a / the generative data-driven model based on the user instruction or another tool that is carried out earlier according to the tool sequence.

[0028] In one embodiment, the agent configured to generate at least one data provision instruction carries out at least the following steps: obtaining (e.g. from a routing or orchestrator agent) the user instruction and optionally routing information; obtaining (e.g. from a database) one or more instruction templates relating to accessing at least one decentral network; selecting at least one instruction template of the one or more instruction template(s) based on the user instruction; generating one or more instruction(s) including at least one task instruction related to accessing at least one decentral network, based on the at least one instruction template, the user instruction and optionally the routing information; generating at least one data provision instruction for accessing the at least one decentral network by providing the one or more generated instruction(s) to a generative data-driven model, wherein the generative data- driven model has been trained on (e.g. unstructured) natural language data and is configured to process natural language.

[0029] In one embodiment, depending on the user instruction multiple instruction templates may be selected and processed sequentially, wherein the instruction templates relate to to identification of at least one decentral network, setting up at least one edge data connector,preparing and testing end-to-end integration, deploying access infrastructure to the production environment, wherein optionally generating instructions includes providing at least part of the data provision instruction generated by the generative data-driven model from one or more preceding instruction(s) to a subsequent instruction template.

[0030] In one embodiment, routing information comprises data for identifying the at least one decentral network among a set of decentral networks associated with different products and / or other production environments, based on the data indicative of the production environment, the output product, and at least one other production environment; and the at least one agent(s) being configured to generate at least one data provision instruction specific for the at least one decentral network, wherein optionally the data provision instruction comprises network identification data indicative of the at least one decentral network. For instance, a routing agent may determine, e.g. using a the generative data-driven model, routing information e.g. by extracting relevant information from the user instruction, based on training data of the generative data driven model, etc. Such routing information may help a tool or agent of the tool sequence to more reliably determine at least a part of the at least one data provision instruction for accessing the at least one decentral network.

[0031] In one embodiment, the pre-defined routing information is obtained from a knowledge database, in particular a knowledge database comprising data sets for (e.g. each) decentral networks in a set of decentral networks comprising the at least one decentral network, wherein the data sets are descriptive of a respective decentral network (e.g. data descriptive of a set of decentral networks may comprise descriptions of production environments associated with (each of) the decentral network(s), products or output products associated with (e.g. each of) the decentral network(s)). So the routing or orchestrator agent or network selection agent may determine based on the knowledge data base which decentral networks product data set needs to be accessed. A knowledge database my provide description of data spaces such as the decentral networks, e.g. descriptions of verticals, industry, products, production environments per decentral network. Based on the description (as an example of routing information) being closest to the user instruction according to a similarity measure as described above, e.g. cosine similarity, the description may be handed to the generative data-driven model as context e.g. in terms of few shot learning to enhance the generated instructions. In an embodiment, steps carried out by the generative data-driven model may be carried out by a graphical processor unit (GPU). Determining or generating at least a part of a data provision instruction may involve matrix operations performed by tensor cores of a / the GPU, utilizing the parallel processing capabilities of a GPU, which may allow enhances efficiency. For instance, any embedding, e.g.of user instructions into an embedding space, may be managed by the GPU, utilizing high throughput and memory bandwidth to handle large datasets and complex computations concurrently. The routing information, e.g. may at least in part be included in a task instruction or respective instruction template.

[0032] A knowledge data base may be specific for a certain agent or data specific for a certain agent may be retrieved from the knowledge database, e.g. based on metadata associated with data provided in the knowledge data base. For instance, data sets descriptive of a respective decentral network may be used as routing information for a routing agent e.g. for identification of at least one decentral network product data set(s) should be provided on or obtained from. For instance, at least one data set descriptive of a respective decentral network may be obtained from a knowledge database based on a similarity measure, e.g. by obtaining a plurality of embeddings of respective data sets descriptive of a respective decentral network, wherein an embedding comprises an embedding of the respective data set descriptive of a respective decentral network; Embedding the user instruction into an embedding space comprising said plurality of embeddings; Determining a similarity score for the embedding of the at least one data set descriptive of a respective decentral network of the plurality of operation data embeddings, wherein the similarity score is based on the distance between the at least embedding and the embedded user instruction in the embedding space with respect to a similarity measure; receiving the at least one data set descriptive of a respective decentral network associated with the at least one embedding having a similarity score above or below or threshold similarity or the data set being associated with an embedding having the highest similarity score. A task instruction related to accessing at least one decentral network may comprise at least a part of the received at least one data set descriptive of a respective decentral network. The task instruction related to accessing at least one decentral network may be provided to a generative data-driven model, wherein the generative data-driven model has been trained on unstructured natural language data and is configured to process natural language to generate at least one data provision instruction for accessing the at least one decentral network.

[0033] Data sets related to setting up at least one edge data connector and / or deploying access infrastructure to the production environment may be used by an integration facilitation agent e.g. to configure the setup of an edge data connector and / or manage deployment of access infrastructure to a production environment. Said data sets related to setting up at least one edge data connector and / or deploying access infrastructure may be provided by the knowledge database. For instance, at least one data set related to setting up at least one edge data connector and / or deploying access infrastructure to the production environment may beobtained from a knowledge database based on a similarity measure as described above for the at least one data set descriptive of a respective decentral network. A task instruction related to accessing at least one decentral network or in particular related to setting up at least one edge data connector and / or deploying access infrastructure may comprise at least a part of the received at least one data set descriptive of a respective decentral network and may e.g. be based on a integration facilitation template. The task instruction related to accessing at least one decentral network or in particular related to setting up at least one edge data connector and / or deploying access infrastructure may be provided to a generative data-driven model, wherein the generative data-driven model has been trained on unstructured natural language data and is configured to process natural language to generate at least one data provision instruction for accessing the at least one decentral network.

[0034] In one embodiment, the at least one agent(s) configured to generate at least one data provision instruction for accessing the decentral network provides the one or more instruction(s) to a generative data-driven model, the generative data-driven model having been trained to generate the at least one data provision instruction in response to obtaining the one or more instruction(s).

[0035] In another embodiment, the agent configured to generate at least one data provision instruction carries out the following steps: providing at least one user instruction and one or more instruction templates relating to accessing at least one decentral network; selecting one or more instruction templates based on the user instruction; generating one or more instruction(s) including at least one task instruction related to accessing at least one decentral network; generating at least one data provision instruction for accessing the at least one decentral network by providing the one or more generated instruction(s) to a generative data- driven model, wherein the generative data-driven model has been trained on unstructured natural language data and is configured to process natural language; and providing the data provision instruction. Selecting at least one task instruction template(s) of the one or more task instruction template(s) may be based on a similarity measure (e.g. Cosine similarity or Euclidian distance) in a common embedding space of the user instruction and the one or more task instruction template(s). For instance, the user instruction may be embedded. For instance selection may be based on a distance between embeddings (e.g. wherein an embedding is a numerical representation) ) of the user instruction and embedding (e.g. numerical representation(s)) of respective task instruction template(s). For example a similarity measure may be calculated between the embedded user instruction and embeddings of the one or more task instruction template(s). For instance, the task instruction template relating to the embeddedtask instruction template closest to embedded user instruction in relation to the similarity measure may be selected. In an example, to avoid mismatches, task instruction template relating to the embedded task instruction template closest to embedded user instruction in relation to the similarity measure may be selected in case the similarity measure is below or above a predefined threshold value.

[0036] In one embodiment, the at least one agent configured to generate at least one data provision instruction for accessing the at least one decentral network is or comprises a network agent configured for a specific decentral network of the at least one decentral network(s). In an example embodiment the network agent carries out at least the following steps: obtaining predefined network access routing information (e.g. being a type of routing information); generating a network access routing class for the user instruction at least based on similarity between the routing information and the user instruction; routing, based on the network access routing class, the user instruction to at least one integration facilitation agent(s) configured to generate the at least one data provision instruction for accessing the at least one decentral network, wherein the at least one data provision instruction is specific to the integration facilitation; providing the at least one data provision instruction (e.g. to the operator or the routing agent). In an example embodiment the network agent is configured in a similar manner to the orchestrator or routing agent, wherein the the network agent analogously as described for the orchestrator agent determines at least on further agent to be carried out.

[0037] A routing agent may route the user instruction or part of the user instruction to an integration facilitation agent, e.g. the at least one agent configured to generate at least one data provision instruction for accessing the at least one decentral network is or comprises an integration facilitation agent. An integration facilitation agent may e.g. be configured to manage the setup of an edge data connector and / or deploy access infrastructure to a production environment, wherein the integration facilitation agent may coordinate data exchange and communication between distributed nodes in the decentral network. The integration facilitation agent may be an agent specific for deploying access infrastructure to the production environment or an agent specific for setting up an edge data connector. An integration facilitation may e.g. relate to or be setting up an edge data connector, deploying access infrastructure to the production environment; the data provision instruction may comprise instructions to provide network configuration details, such as IP addresses and gateway settings, to establish a connection with the decentralized network. The data provision instruction may comprise instructions to identify the types of data sources that the edge connector will interface with, including sensors, loT devices, databases, and other systems. The data provision instructionmay comprise instructions to provide specifics about the devices that will connect to the edge connector, such as device types, operating systems, and communication protocols.

[0038] In one embodiment, the integration facilitation agent carries out the at least the following steps: obtaining (e.g. from a database) one or more integration facilitation template(s) specific for the integration facilitation for accessing the specific decentral network and optionally the predefined network access routing information; selecting at least one integration facilitation templates of the one or more integration facilitation template(s) based on the user instruction; generating one or more instruction(s) including at least one task instruction related to accessing at least one decentral network, based on the at least one integration facilitation templates, the user instruction and optionally the pre-defined network access routing information; generating the at least one data provision instruction for accessing the at least one decentral network by providing the one or more generated instruction(s) to a generative data-driven model, wherein the generative data-driven model has been trained on unstructured natural language data and is configured to process natural language; and optionally providing the at least one data provision instruction (e.g. to the operator or a computer system associated with the production environment). In an example embodiment the integration facilitation agent is configured in a similar manner to the orchestrator or routing agent, wherein the the network agent analogously as described for the orchestrator agent determines at least one further agent to be carried out.

[0039] In an embodiment, generating one or more instruction(s) including at least one task instruction related to accessing at least one decentral network comprises contextualizing the at least one instruction template based on the user instruction.

[0040] In an embodiment of the second aspect, the method further comprises: obtaining, via the decentral network, a product data set associated with an output product, wherein the output product is input material for producing the product;

[0041] In an embodiment of the second aspect, providing the product data includes gathering via the decentral network input material data associated with input materials used to produce the product from decentral network node(s) associated with input material producer(s) producing the input material(s).

[0042] In an embodiment, the method further includes a step of generating group access data associated with a generated output product data set and optionally one or more authorization rule(s) defining access to and / or usage of the generated output product data set. The group access data may identify access control groups including decentral participant identifier(s) associated with decentral network participants permitted to access at the generated output product data set. This access group data may allow to control access to such data sets, henceavoiding access of unauthorized data consumers to such data. This may ensure that the output product data sets are only shared with such data consumers that require such data, for example for the generation of product passports.

[0043] The decentral network may include participant nodes configured to perform data transactions or data exchanges between participant nodes, e.g. via peer-to-peer communication between participant nodes. The participant nodes may be associated with producers or user of supply chain products, such as chemical product producers, intermediate product producers, end product producers or end product users. The data transactions or exchanges may be based on a decentral network protocol establishing peer-to-peer communication between participant nodes. The participant nodes of the decentral network may be configured to provide data consuming component(s) and / or data providing component(s). A data providing component may be configured to provide or send product data to another participant node of the decentral network. A data consuming component may be configured to ingest or receive product data from another participant node of the decentral network. Providing product data to the data providing component for access by the data consuming component may include indirect or direct access of the data consuming component to the product data.

[0044] The decentral network protocol may relate to decentral identifier(s) identifying the at least one supply chain product and / or the at least one participant node and / or authentication and / or authorization mechanism(s) authentication and / or authorizing the at least one participant node for data exchange.

[0045] The data transactions or exchanges may be based on a decentral network protocol including decentral identifier(s) associated with at least one supply chain product and / or at least one participant node. Based on the decentral identifier(s) of the participant nodes the participant nodes for peer-to-peer communication may be identified. Based on the decentral identifier(s) of the supply chain products the product data associated with the supply chain product for peer- to-peer data exchange may be identified.

[0046] The data transactions or exchanges via the decentral network may be based on a decentral network protocol including authentication and / or authorization mechanism(s), e.g. based on an identification associated with the production environment (such as a decentral participant identifier) or security credential(s). Based on the authentication and / or authorization mechanism(s) a peer-to-peer network between participant nodes of the decentral network may be established. The one or more authentication mechanism(s) may be associated with or linked to the decentral identifier(s) related to participant nodes of the decentral network. The one or more authentication mechanism(s) associated with the decentral identifier(s) may be providedto participant node(s) for authenticating the peer-to-peer communication between respective participant nodes, via which e.g. an operator of a production environment may participate (e.g. for data exchanges / transactions) in the decentral network. The one or more authentication mechanism(s) associated with the decentral identifier(s) may be accessible by the data providing component and / or the data consuming component of respective participant node. The one or more authorization mechanism(s) may include at least one authorization rule for providing and / or consuming and / or accessing product data. The one or more authorization mechanism(s) may be associated with or linked to the decentral identifier(s) associated with the participant node and / or the supply chain product. The one or more authorization mechanism(s) may be associated with or linked to the decentral identifier(s) related to product data to be accessed and / or exchanged. The decentral configuration allows for more efficient use of computing resources and strengthens control by each data owner of the decentral network.

[0047] A product data set may relate to the digital twin of a product, e.g. a supply chain product, such as an output product. The digital twin may be a digital representation of the physical entity, e.g. may be or comprise product data associated with the product, such as property data associated with the product. The digital twin may be used to represent the physical entity of the product or output product in a digital representation of a real-world system. The product data set (e.g. digital asset) associated with at least one supply chain product may relate to data points related to one or more properties of the supply chain product. The data points related to one or more properties of the product may relate to properties measured or derived from data measured before, during and / or after production of the product. A product data set may be provided in the decentral network. It may be generated according to a selected or provided at least one data model by or via a peer-to-peer communication according to the selected or provided at least one decentral network protocol for exchanging product data associated with the supply chain product. For generating an output product data set at least the following steps may be carried out: providing data associated with the output product including at least one output product identifier; gathering - based on the provided data associated with the output product - output product data from one or more databases; generating the output product data set by transforming the gathered output product data using a rule-based engine including one or more rule(s) associated with at least one of the product(s); providing the generated output product data set for access by a decentral data consuming node under control of or controlled by a decentral data providing node associated with data owner of the generated output product data set.

[0048] A product data set or output product data set associated with an output product may be at least a part of a digital twin of the physical entity of the output product, e.g. a product data set for an output product. The output product may be used as input material to produce one or more product(s), such as a battery or a car. The product data set(s) or output product data set(s) may be stored in a database (DB), e.g. an assets DB associated with a data transfer service. An assets DB may be associated with a decentral provider node. The decentral provider node may be associated with a decentral network participant. The decentral network participant may be the producer of the output product. The decentral network participant may be the data owner of the output product data set(s) stored in the database. A product data set may be generated by a decentral network participant, such as the operator of the production environment, using (highly complex) semantic model(s) to ensure uniform data set(s) containing all required data point(s) from the respective product data. Alternatively, output product data may be gathered from the data owner and a rule-based engine may be used e.g. to aggregate the gathered output product into a given data format, such as a tabular representation, without the use of complex semantic models, hence facilitating simple generation and reliable sharing of output product data set(s) within the product ecosystem. The output product data set(s) may be associated with authorization rule(s), hence avoiding unauthorized access to such data and ensuring safe sharing of such data. The tabular representation stored within a dedicated storage can be readily consumed by a decentral data consuming node from the data providing node associated with such tabular representation under the control of the decentral data providing node. Hence, the output product data set can be directly consumed via a decentral data providing node from the dedicated storage without having to use any intermediary registries, such as decentral registries storing access elements pointing to 'data set(s) generated using highly defined semantic models. By using a decentral network, the generated output product data set(s) may be shared in a secure and controlled manner by avoiding access to such data set(s) by unauthorized decentral network participant(s), hence avoiding unwanted transparency on the supply chains by third parties.

[0049] A rule-based engine may be used to transform at least part of gathered output product data associated with output product(s). The rule-based engine may be a software or software component that applies one or more rules to at least part of the gathered output product data. The rule(s) may include or correspond to executable logic. The executable logic may be generated from a rule template including unstructured data associated with instructions related to transformation operation(s). The rule-based engine used to transform at least part of the gathered output product data may include one or more rule(s) associated with product(s) produced from the output product(s) (e.g. by using the output products as input materials). Theone or more rule(s) may be associated with or derived from a semantic model of the product. Hence, the one or more rule(s) may ensure that data point(s) for such output products required according to the semantic model may be included in the generated output product data set. The one or more rule(s) may be defined by the mandatory output product data points present within the semantic model. The one or more rule(s) may be generated based on the mandatory output product data points present within the semantic model. This may ensure that output product data required by the semantic model is included in the generated output product data set. The one or more rule(s) may hence be generated or defined by the semantic model associated with the product produced from the output product(s) as input materials. The one or more rule(s) may hence not be derived or generated based on a semantic model associated with the produced output products. This may allow to avoid generation of complex semantic models for produced output products but may instead allow to use existing semantic models of product(s) for generation of rule(s) to transform gathered output product data to output product data set(s).

[0050] The rule-based engine may operate on individual data point level, combination of data points or the whole gathered output product data. The operation of the rule-based engine may be defined by the one or more rule(s). The rule(s) may include or correspond to executable logic. The executable logic may be generated from a rule template including unstructured data associated with instructions related to validation operation(s). Rule(s) associated with individual data point(s) may include one or more rule(s) defining condition(s) for individual data points present with the gathered output product data. Use of such rule(s) allows to ensure that data point(s) required by the semantic model associated with the product are contained in the generated output product data sets. Rule(s) associated with multiple data points may include one or more rules defining required combination of data points. Use of such rule(s) allows to ensure that combination(s) of data point(s) required by the semantic model associated with the product are contained within the generated output product data sets. A product data set may be generated based on a network protocol used by the at least one decentral network (which may be derived from data for identifying the at least one decentral network) and / or at least one data model. The decentral network protocol may include one or more data model(s) defining product data, such as property data points, associated with the supply chain product. The one or more data model(s) may relate to one or more supply chain products and / or supply chain product class(es). The data model may be or may relate to specific supply chain product(s) or class(es). The data model may define or specify at least types of product data, such as property data points, that relate to properties of the supply chain product. The data model may define or specify at least a context or descriptions of the product data, such as property data points, that relate to properties of the supply chain product. The data model may define or specify key-value pairsfor product data, such as property data points, that relate to properties of the supply chain product. The data model may define or specify the digital twin of the supply chain product by product data, such as property data points, that relate to properties of the supply chain product. The data model may be defined or specified per decentral network and / or per supply chain product(s) or class(es). Generating the product data set may comprise: generating at least one product data set according to a selected at least one data model by generating at least one selection instruction for selecting at least one data point from a product data base based on the selected at least one data model and / or at least in part based on at least one user and / or system interaction, wherein the product data base provides product data associated with the at least one supply chain product; generating the product data set by providing one or more decentral identifier(s) uniquely associated with the supply chain product based on the selected decentral network protocol and by providing a locator to the generated product data set; and providing the product data set for access by one or more data consuming node(s) according to the selected at least one decentral network protocol.

[0051] The product data set may include at least one decentral identifier and at least one locator to or digital representation locating the product data associated with the supply chain product. The digital asset may include at least one decentral identifier and at least one locator to the data points related to one or more properties of the supply chain product. The digital asset may be provided to the decentral network to be discoverable and / or accessible by the participant nodes. The digital asset may be provided to the decentral network to provide access to the product data associated with the supply chain product e.g. via the locator. The digital asset may be provided to the decentral network to provide access to the data points related to one or more properties of the supply chain product. The digital asset may be provided to the decentral network to provide the locator for accessing to the product data associated with the supply chain product. The digital asset may be provided to the decentral network to provide the locator for accessing to the data points related to one or more properties of the supply chain product. The digital asset may be provided to the decentral network in relation to, linked to or including authentication and / or authorization information. The digital asset may be provided to the decentral network to provide the locator for accessing to the product data associated with the supply chain product after authentication and / or authorization of the by the participant node(s) requesting access. The digital asset may be provided to the decentral network to provide the locator for access to the data points related to one or more properties of the supply chain product by the participant node(s) after authentication and / or authorization of the by the participant node(s) requesting access.

[0052] The product data set may include the decentral identifier(s) associated with the supply chain product and the decentral identifier(s) may be linked to one or more product data points associated with the supply chain product. The decentral identifier(s) may be linked to the one or more product or property data points. The one or more product data points or property data points may be linked to the decentral identifier included in the digital asset. The one or more product data points, or one or more property data points may be stored in a data base of the supply chain product producer for access by supply chain product user(s), e.g. authenticated and / or authorized for such access. The one or more product data points, or one or more property data points may be stored in a data base of the supply chain product producer for transfer to the supply chain product user e.g. when accessed or on providing the supply chain product.

[0053] The decentral identifier(s) may comprise any unique identifier uniquely associated with the participant node of the supply chain product producer, the supply chain product producer, the physical entity of the supply chain product, product data sets associated with the supply chain product such as properties of the supply chain product and / or any subset or combination thereof. The decentral identifier(s) may include a Universally Unique IDentifier (UUID) or a Digital IDentifier (DID). The decentral identifier(s) may be issued by a central or decentral identity issuer of the decentral network according to the decentral network protocol. The decentral identifier(s) may be discoverable and or accessible to participant nodes of the decentral network according to the decentral network protocol. The decentral identifier(s) may be linked to authentication and / or authorization information. Via the decentral identifier(s) and the unique association with the participant node of the supply chain product producer, the supply chain product producer, the physical entity of the supply chain product, product data sets associated with the supply chain product such as properties of the supply chain product and / or any subset or combination thereof, access to the supply chain product data may be controlled by the supply chain product producer. This contrasts with central authority schemes, where identifiers are provided by such central authority and access to data is controlled by such central authority. Decentral in this context refers to the usage of the identifier in implementation as controlled by the data owner, such as the supply chain product producer.

[0054] A decentral network protocol may specify a communication protocol for peer-to-peer communication according to a communication standard defined for the decentral network. For example, some decentral networks may define fully decentral protocols using trust systems based on distributed ledgers such as blockchain implementations. Further for example, other decentral networks may define partially decentral protocols using trust systems that are only partially based on distributed ledgers such as blockchain implementations. For example,verifiable credentials used to verify product data by an issuer node may be implemented based on distributed ledger for the verification information such as blockchain implementations. Further for example, management and orchestration of decentral identifiers associated with participant nodes and / or supply chain products may be implemented based on central management and orchestration systems. The decentral network protocol may differ between different decentral networks. Participant nodes may participate in multiple decentral networks with different decentral network protocols.

[0055] The decentral network protocol may include one or more data model(s) defining product data, such as property data points, associated with the supply chain product. The one or more data model(s) may relate to one or more supply chain products and / or supply chain product class(es). The data model may be or may relate to specific supply chain product(s) or class(es). The data model may define or specify at least types of product data, such as property data points, that relate to properties of the supply chain product. The data model may define or specify at least a context or descriptions of the product data, such as property data points, that relate to properties of the supply chain product. The data model may define or specify key-value pairs for product data, such as property data points, that relate to properties of the supply chain product. The data model may define or specify the digital twin of the supply chain product by product data, such as property data points, that relate to properties of the supply chain product. The data model may be defined or specified per decentral network and / or per supply chain product(s) or class(es).

[0056] The user instruction and / or data provision instruction may be provided in natural language, e.g. it may be a natural language representation of a task that is to be performed. The user instruction and / or data provision instruction may indirectly or semantically relate to the product (e.g output or supply chain product), the decentral network and / or the data model. The user instruction and / or data provision instruction may correspond to the one or more indication(s) as provided by one or more decentral network protocols and / or as stored by a local data base associated with single participant nodes such as the knowledge database or the product data base. The user and / or system interaction may provide context or descriptions in natural language relating to the supply chain product, the decentral network and / or the data model as provided by one or more decentral network protocols and / or as stored by a local data base associated with single participant nodes such as the knowledge database.

[0057] The one or more instruction(s) may be generated based on instruction templates as may be retrieved from a template store. The instruction template may be selected based on the user instruction- e.g. by way of similarity search based on embeddings as described above. Theinstruction template may include placeholders for including or merging further instructions into the template instruction. The one or more instruction(s) may be generated by including or merging the user instruction into the instruction template. The one or more instruction(s) may be generated by including or merging further information as specified by the instruction template to the instruction template. This may be an example of contextualizing the the at least one instruction template based on the user instruction.

[0058] For selection of the instruction templates, the user instruction may be embedded. Embedding may include a mapping of at least part of the user instruction in natural language to a numeric representation of at least part of the user instruction. The numeric representation may also be referred to as embedding. The instruction templates may relate to one or more instruction types. The instruction type may relate to identification of data, access to data, selection of machine-readable instructions and / or execution of machine-readable instruction. The instruction template(s) and / or the instruction type(s) may be provided in natural language and / or as numeric representations of the natural language version. The instruction template(s) and / or the instruction type(s) may be mapped in at least part from natural language or text to a numeric representation of at least part of the user instruction.

[0059] For selection of the instruction template the embedding of the user instruction and the embedding of the instruction template(s) and / or type(s) may be compared. The similarity between respective embeddings of the user instruction and the embedding of the instruction template(s) and / or type(s) may be determined. The similarity may be based on a similarity measure such as Euclidian or cosine similarity. The embedded instruction template with the lowest distance to the embedded user instruction may be selected.

[0060] Further for example, instructions(s) including a task instruction may be generated to select an instruction template, the user instruction, and / or the template instruction types. The generated instructions may be provided to a generative data-driven model configured to provide the selected instruction template.

[0061] These are only some possibilities for selecting an instruction template. Other possibilities may rely on string or semantic comparison. If no suitable instruction template can be found according to a pre-defined similarity measure, a signal may be sent for asking the operator to further specify the user instruction or to free chat agent or Q&A agent to further process the user instruction to have the operator specify the user instruction.

[0062] In an embodiment, the method further includes a step of generating group access data associated with a generated output product data set and optionally one or more authorization rule(s) defining access to and / or usage of the generated output product data set. The groupaccess data may identify access control groups including decentral participant identifier(s) associated with decentral network participants permitted to access at the generated output product data set. This access group data may allow to control access to such data sets, hence avoiding access of unauthorized data consumers to such data. This may ensure that the output product data sets are only shared with such data consumers that require such data, for example for the generation of product passports.

[0063] The method according to any aspect or any step of the method may for instance be performed, carried-out, executed and / or controlled by an / the computing apparatus, for instance a server, a server cloud, a computer-system, or part thereof. The method or steps of the method may be computer-implemented. For instance, the method or any step of the method may be performed and / or controlled by using at least one processor e.g. of an / the apparatus.

[0064] Alternatively, the method according to any aspect may be performed, carried-out, executed and / or controlled by more than one apparatus, for instance a server cloud comprising at least two servers or a system of apparatus, e.g. a system comprising at least one server providing at least one data base comprising instruction templates, a server providing a data base comprising integration facilitation templates, at least one server providing a generative data-driven model, and an apparatus comprising means for carrying-out the respective steps of the method according to the first aspect.

[0065] According to a further example aspect, an apparatus is disclosed, the apparatus comprising respective means for carrying out or performing the steps of the method according to the first aspect and / or any embodiment or example and combinations thereof of the method.

[0066] Additionally or alternatively the apparatus may comprise at least one processor and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus at least to carry out the steps of the method according to the first aspect and / or any embodiment or example and combinations thereof of the disclosed method. Additionally or alternatively the apparatus may comprise circuitry (e.g. hardware-only circuitry, digital circuitry and / or a combination of hardware circuits and software) designed or configured to implement the functions for carrying out the steps of the method according to the first aspect and / or any embodiment or example and combinations thereof of the method. Circuitry may be implemented in a chipset or a chip or an integrated circuit.

[0067] Various units, entities, nodes or other computing components may be described as “configured to” perform a task or tasks. Configured to shall recite structure meaning “having circuitry that” performs the task or tasks on operation. The units, circuits, entities, nodes or other computing components can be configured to perform the task even when theunit / circuit / component is not operating. The units, circuits, entities, nodes or other computing components that form the structure corresponding to “configured to” may include hardware circuits and / or memory storing program instructions executable to implement the operation. The units, circuits, entities, nodes or other computing components may be described as performing a task or tasks, for convenience in the description. Such descriptions shall be interpreted as including the phrase “configured to.”

[0068] In general, the methods, apparatuses, systems, computer elements, nodes or other computing components described herein may include memory, software components and hardware components. The memory can include volatile memory such as static or dynamic random-access memory and / or nonvolatile memory such as optical or magnetic disk storage, flash memory, programmable read-only memories, etc. The hardware components may include any combination of combinatoric logic circuitry, clocked storage devices such as flops, registers, latches, etc., finite state machines, memory such as static random-access memory or embedded dynamic random-access memory, custom designed circuitry, programmable logic arrays, etc.

[0069] Input material may refer to any good which is bought from suppliers and brought to the respective production plant or production environment. The input material may include starting material used in the production process of the production plant to produce the product. An input material can be used in any production step used to produce the product. This means, the product of the one production plant can be the input material of the other production plant. Likewise, the output product of one supply chain participant may be used as input material to produce a further output product by a downstream participant. Input material may include recycled material. The input material may comprise or be any input material entering a production. The input material may comprise or be any input material provided at any entry point of the production.

[0070] Output product may include any product produced from one or more input materials. The output product may be produced via one or more process steps. The process steps may involve chemical reactions and / or physical processes and / or assembly processes. The input material may be used in one or more of such production step(s). The output product may comprise or be any product produced by a production and provided at any exit point of the production. The output product may be used as input material to produce one or more product(s). The product(s) may be produced by one or more downstream participants which may use the output product(s) produced by one or more upstream participants as input material(s). The output product may be associated with an output product identifier. The output product identifier may be a digital or virtual output product identifier. The output product identifier may uniquelyidentify the output product within the entity producing the output product. The output product identifier may uniquely identify the output product within the decentral network. The output product identifier may be associated with an identifier element physically connected to the output product. The identifier element may encode the digital output product identifier. The output product identifier may include an output product name, an output product number, a LOT number, a batch number, a serial number, etc.

[0071] The output product / input material and the product may be part of a product ecosystem. The product ecosystem may include chemical products. The product ecosystem may include production chains to produce a product. The product may be a chemical product, an intermediate chemical product, a component, a component assembly or an end-product. The product ecosystem may include processing chains to process used products resulting from the use of produced products. Processing chains may include recycling chains to recycle at least part of the used product or a component thereof. Processing chains may include re-use chains to reuse the used product. The product ecosystem may include various participants, such as raw input material producers, chemical product producers, chemical product users, end-product producers, end-product users, EOL product collectors and recyclers. The product ecosystem may allow to use of recycled materials resulting from recycling of end-of-life end products to produce new products, such as chemical products. The product ecosystem may be associated with the production and / or re-use and / or recycling of physical products.

[0072] The participants (e.g. an operator of a production environment) of the product ecosystem may be connected via a decentral network. The decentral network may include one or more decentral network node(s) configured to perform data transactions. The decentral network node(s) may be associated with participants of the product ecosystem. The data transactions may be based on a transaction protocol including authentication and / or authorization mechanism(s). Based on the authentication and / or authorization mechanism(s) a peer-to-peer network between decentral network node(s) of the decentral network may be established. The one or more authentication mechanism(s) may be associated with or linked to decentral identifier(s). The one or more authentication mechanism(s) associated with decentral identifier(s) may be provided to decentral network node(s). The one or more authentication mechanism(s) associated with decentral identifier(s) may be accessible by decentral network node(s). The decentral configuration allows for more efficient use of computing resources and strengthens control by each data owner of the decentral network.

[0073] The decentral data providing network node may comprise computer-executable instructions for providing and / or processing data within a decentral network, such as the outputproduct data sets, by a decentral data consuming network node. The decentral data providing network node may be associated with or connected to one or more dedicated data storage(s) storing the output product data sets. The decentral data providing network node may be directly or indirectly connected to the data storage(s) storing the output product data sets. Hence, the decentral data providing network node may be associated with the output product data sets. The dedicated data storage(s) may be under control of the data owner of the output product data sets. The data owner may be an entity having access to the output product data sets and controlling access by data consuming services of the decentral network to the output product data sets. The data owner may be the output product producer. Via the output product identifier and its unique association with the data owner and output product data set access to the output product data set may be controlled by the data owner. The output product data sets may be accessible for the data owner. The data owner may hence directly or indirectly own the output product data sets. The output product data sets may be stored in a database of or associated with the data owner. The output product data sets may be stored in a database accessible by the data owner. The data owner may control access to the output product data sets via the data providing service of the data owner. The data owner may control access to the output product data sets. The output product data sets may be associated with the data owner. The data owner may be the owner of the output product data sets or the output product data set owner. The output product data sets may be stored in a data base of or under control by the data owner.

[0074] The decentral data consuming network node may comprise computer-executable instructions for accessing and / or processing data within a decentral network, such as output product data, provided by a decentral data providing network node. The decentral data consuming network node may be controlled or owned by or associated with a consumer of the output product data (e.g. the entity generating the product passport). The consumer may be any entity processing the output product data. The consumer may be any entity operating a production configured to process the output products associated with the output product data as input material(s) Processing may include using the output products as input materials to produce further products. Processing may include performing one or more recycling step(s) on the output products as input material. The consumer may be an upstream participant of the output product producer in the product ecosystem.

[0075] The disclosed apparatus according to any aspect may be a module or a component for a device, for example a chip. Alternatively, the disclosed apparatus according to any aspect may be a device, for instance a server, server cloud, a personal computer or a user device. The disclosed apparatus according to any aspect may comprise only the disclosed components, forinstance means, processor, memory, or may further comprise one or more additional components, such as a graphical user interface.

[0076] According to a further example aspect, a computer element is disclosed, the computer element comprising instructions, which when executed by a processor or a computing apparatus perform or carry out the steps according to the methods or as defined by the apparatuses disclosed herein.

[0077] According to a further example aspect, an apparatus is disclosed, configured to perform and / or control or comprising respective means for performing and / or controlling the method according to any example aspect. According to a further example aspect, an apparatus is disclosed comprising at least one processor; and at least one memory storing instructions that, when executed by the at least one processor, cause the first apparatus at least to perform the method according to any example aspect.

[0078] According to a further example aspect, a computer program or computer program product is disclosed, the computer program or computer program product when executed by a processor causing an apparatus, for instance a server, to perform and / or control the actions of the method according the any aspect.

[0079] According to a further example aspect, a (e.g. tangible and / or non-transitory) computer readable storage medium is disclosed, the computer readable storage medium comprising a computer program, the computer program when executed by a processor causing an apparatus, for instance a server, to perform and / or control the actions of the method according the any aspect.

[0080] Any disclosure herein relating to any example aspect is to be understood to be equally disclosed with respect to any subject-matter according to the respective example aspect, e.g. relating to an apparatus, a method, or a computer program. Thus, for instance, the disclosure of a method step shall also be considered as a disclosure of means for performing and / or causing to perform the respective method step. Likewise, the disclosure of means for performing and / or causing to perform a method step shall also be considered as a disclosure of the method step itself. The same holds for any passage describing at least one processor; and at least one memory including instructions; the at least one memory and the instructions configured to, with the at least one processor, cause an apparatus at least to perform a step.

[0081] In the following example features and example embodiments of all aspects will be described in more detail.

[0082] A product passport may refer to a data set having a defined semantic structure. The defined semantic structure may be obtained by applying a semantic model, such as an aspectmodel, to validated input material data and product data associated with the respective product. The product passport may include a product identifier, at least one decentral identifier, validated input material data and product data (e.g. passport data). The product passport may include one or more authentication mechanisms associated with the decentral identifier(s), the validated input material data and the product data. The product passport may relate to one or more authorization mechanisms associated with the decentral identifier(s), the validated input material data and the product data. The one or more authorization mechanisms may include authorization rules determining if access to the at least a part of the validated input material data and / or product data is granted. The product passport may be associated with one or more digital representations of the passport data. The digital representations may be regarded as access element(s) providing access to the product passport or parts thereof. The digital representation may include a decentral identifier and access data. The access data may include a locator or pointer, such as am url or uri, to a dedicated storage, such as a dedicated storage address, associated with the data owner of the product passport. The pointer or locator may point directly to the dedicated storage. The pointer or locator may point to a data providing network node associated with the dedicated storage. The access element may include one or more authentication mechanisms associated with the decentral identifier(s) and the access data. The access element may be associated with one or more authentication mechanisms associated with the decentral identifier(s) and the access data. The access element may be provided to a decentral registry storing access elements. The decentral registry may be associated with a data providing network node. This may allow to control access to such registry and access to access element(s) stored in such registry via the data providing network node. The decentral registry may be associated with a participant of the product ecosystem. The decentral registry may be associated with the data owner of the product passport. The decentral registry may be part of the decentral network but may not be associated with a particular participant of the product ecosystem, e.g. may be regarded as infrastructure node of the decentral network. A decentral identifier may be linked to a digital representation of the product passport, wherein the digital representation may include a representation for accessing the digital product passport. A digital product passport may be stored in a database associated with the product producer (e.g. the operator of the production environment) for access by the product consumer (e.g. the operator of the other production environment). Access to the database may be controlled (e.g. via the decentral identifier) by the data owner of the digital product passport (e.g. the product producer).

[0083] In an embodiment, the one or more databases (e.g. storing output product data) are distributed databases, wherein at least one of the databases stores instance(s) of the output product data. A distributed database may be a collection of data stored at different sites of acomputer network. Each site might expose a degree of autonomy, providing services for the execution of local applications, but also participating in the execution of a global application. For instance, a distributed data source may be a distributed database. A distributed database can be created by splitting and scattering the data of an existing database over different sites or by federating together multiple existing databases. Each data source may contain only a fragment of the data associated with the respective output product. This leads to a fragmentation of said data. Two common types of data fragmentation are horizontal fragmentation, wherein (e.g. possibly overlapping) subsets of data tuples are stored at different sites; and vertical fragmentation, wherein (e.g. possibly overlapping) subtuples of data tuples are stored at different sites. More generally, the data associated with the output product may be fragmented into a set of relations (e.g. tables of a relational database, distributed across multiple sites).

[0084] In an embodiment, the output product is a chemical intermediate product, a chemical product, a part, a component or a component assembly. The chemical intermediate product and / or the chemical product may be produced from virgin input material(s) and / or recycled input material(s). The component may be a battery component, such as an electrode, a battery cell, a battery module, a separator, a cell casing, a battery management system, a cooling system, mechanical subsystem or a battery pack housing.

[0085] In an embodiment, the gathered output product data includes output product identifier data, property data associated with the output product, output product name data, output product producer data, output product declaration data, output product safety data, emission data associated with the output product, recyclate content data associated with the output product, biobased content data associated with the output product, biodegradability data associated with the output product, production data associated with the output product, certificate of analysis data associated with the output product, certificate data associated with the output product, life cycle data associated with the output product, storage instruction data associated with the output product, assembly instructions associated with the output product, operating conditions associated with the output product or a combination thereof.

[0086] Output product identifier data may include a batch number, a serial number, a LOT number or a combination thereof.

[0087] Property data may include at least one measured chemical and / or physical property of the produced output product and / or at least one chemical and / or physical property determined from collected data associated with the production of the output product. The data may be collected before, during and / or after production of the output product. The collected data may be used to determine at least one physical and / or chemical property of the produced outputproduct. For instance, the at least one physical and / or chemical property may be determined from sensor data obtained from sensor(s). The data may be collected with a suitable sensor configured to measure the chemical and / or physical property. The chemical property may be a property of the output product that becomes evident during, or after, a chemical reaction. Hence, the chemical property may be any quality that can be established only by changing the chemical identity of the output product. Examples of chemical properties include heat of combustion, enthalpy of formation, toxicity, chemical stability in a given environment, flammability, oxidation state(s), ability to corrode, combustibility, acidity and basicity, chemical composition, recyclate content used for producing or manufacturing the product, bio-based content used for producing or manufacturing the product, renewable content used for producing or manufacturing the product and / or pH value. The physical property may be any property of the output product that is measurable. Hence, the value of a physical property describes a state of the output product. Examples of physical properties include absorption, brittleness, boiling point, capacitance, color, concentration, continuous discharge, density, ductility, physical dimensions, distribution, efficacy, elasticity, electric charge, electrical conductivity, electrical impedance, electric potential, flow rate, fluidity, hardness, capacity, inductance, intrinsic impedance, luminance, luminescence, luster, mass, melting point, opacity, permeability, permittivity, plasticity, pulse discharge, power, pressure, radiance, resistivity, reflectivity, refractive index, solubility, specific heat, strength, stiffness, temperature, tension, thermal conductivity, thermal resistance, weight, viscosity, volume and / or wave impedance. The measured at least one physical and / or chemical property may be obtained by sensors configured to measure such property. The sensor may be included in a measuring device. The sensor may correspond to the measuring device.

[0088] Recyclate content data and / or bio-based content data may comprise any data related to the recyclate content or the bio-based content used for providing or manufacturing the output product.

[0089] Emission data may comprise any data related to environmental footprint. The environmental footprint may refer to the output product and its associated environmental footprint. The environmental footprint may be output product specific. For instance, the environmental footprint may relate to the output product or additional output product-specific relations. Emission data may include data relating to the carbon footprint of the output product or a Product Carbon Footprint (PCF). Emission data may include data relating to greenhouse gas emissions e.g. released in production of the output product. Emission data may include data related to greenhouse gas emissions. Greenhouse gas emissions may include emissions such as carbon dioxide (CO2) emission, methane (CH4) emission, nitrous oxide (N2O) emission,hydrofluorocarbons (HFCs) emission, perfluorocarbons (PFCs) emission, sulphurhexafluoride (SFe) emission, nitrogen trifluoride (NF3) emission, combinations thereof and additional emissions.

[0090] Emission data may include data related to greenhouse gas emissions of an entities or companies own operations (e.g. production, power plants and waste incineration). Scope 2 may comprise emissions from energy production which is sourced externally. Scope 3 may comprise all other emissions along the value chain. Specifically, this may include the greenhouse gas emissions of raw materials obtained from suppliers. Product Carbon Footprint (PCF) may sum up greenhouse gas emissions and removals from the consecutive and interlinked process steps related to a particular product. Cradle-to-gate PCF may sum up greenhouse gas emissions based on selected process steps: e.g. from the extraction of resources up to the factory gate where the product leaves the company. Such PCFs may be called partial PCFs. In order to achieve such summation, each company providing any products may provide the scope 1 and scope 2 contributions to the PCF for each of its products..

[0091] Production data may comprise any data related to the production of the output product. Production data may include monitoring and / or control data associated with the production of the output product. Production data may be acquired prior to, during and / or after production of the output product.

[0092] An agent may be a self-operating computational unit, e.g. a software component or system that can independently execute tasks or operations. This execution may involve processing input data, applying predefined algorithms, and making decisions based on the results. It may also include the ability to adjust or modify its operations in response to changes in input data or outcomes of its operations. The self-operating computational unit may carry out these tasks without the need for continuous human supervision or intervention, although human input may be incorporated as part of its decision-making processes or to change its operational parameters. It should be understood that the specific functionalities, operations, and level of independence of the self-operating computational unit can vary based on the design and requirements of the specific system it is implemented in.

[0093] Data provision information may be tabulated information on which data is needed e.g. to generate security credentials. For instance, a list may be provided indicating different steps to be performed and associated data needed to complete said step, so that data provision instruction may comprise instructions to provide the respective associated data.

[0094] A data provision instruction may prompt a computer system of a production environment or an operator of the production environment to provide certain data, e.g. on the productionenvironment, provided IT infrastructure intended to be used for accessing the decentral network etc. The in order to for accessing the decentral network. An operation data set may be provided based on or in response to receiving a data provision instruction.

[0095] An operation data set may be a data set associated with a certain step of the access procedure. For instance, accessing the decentral network may require the operator to provide a certain technical infrastructure and to enter data on which infrastructure is being used, so that access to the decentral network may be possible. An operation data set may be provided based on or in response to receiving a data provision instruction.

[0096] Pre-defined routing information may relate to natural language and or text related to the routing of the user instruction for executing the method for accessing a decentral network, it may comprise information on a specific decentral network.

[0097] Security credentials may be tokens or certificates, which may ensure secure access to authenticate an operator or a production environment and authorize data transactions, such as providing and / or consuming product data sets in the decentral network.

[0098] A user instruction may be provided including natural language e.g. text such as a question. The user instruction(s) may be provided by an operator of the distributed chemical production environment. The user instruction(s) may be provided by a web-interface configured to retrieve instruction(s) in natural language or text by an operator of the distributed chemical production environment.

[0099] A task instruction may refer to an objective of a task to be executed. It may relate to a generation task instructing a generative data-driven model to generate at least a part of a data provision instruction for accessing the at least one decentral network and / or a classification task to classify e.g. a user instruction, on which classification e.g. a routing class may be based . The task instruction may e.g. be a prompt for the generative data-driven model. The task instruction may comprise a sequence of one or more text elements. A task instruction may comprise, e.g. in form of a respective text element, at least one of the following: role information, context information, information on the format of the tool sequence or data provision instruction to be generated, example information (e.g. for few-shot learning). The task instruction may comprise at least part of the user instruction. A task instruction may be embedded by the generative data- driven model into its embedding space, e.g. by tokenization and encoding, wherein the task instruction is divided into smaller units or tokens, e.g. words or subwords. Each token may be assigned a unique identifier and may represent a discrete unit of meaning. Tokenization may help organize the task instruction into manageable parts for further processing by the generative data-driven model. The task instruction or tokens of the task instruction may then be encodedinto numerical representation or numerical representations compatbile with the embedding space of the generative data-driven model. Encoding of tokens may map the tokens e.g. of the task instruction to vectors or numerical values that may be trained to capture the semantic and contextual information of the task instruction. The model then may, based on the encoded tokens of the task instruction, generate an encoded tool sequence, which may then be decoded into the tool sequence.

[0100] A generative data-driven model may be a model, e.g. implemented in a computer system, that, based on historical data it has been trained on, may generate new instances of said data e.g. by sampling from a probability distribution, wherein the probability distribution may have been learned during training, and generate an according output data set after receiving a task instruction. The generative data-driven model may have been trained on general purpose training data sets (and e.g. via that training may be configured to) to generate an output data set in response to obtaining (e.g. receiving) the task instruction. A generative data-driven model may be or comprise a transformer-based data driven model, preferably a decoder-only transformer-based model such as a generative pre-trained transformer, e.g. Large Language Model Meta Al (Llama), Llama 2, Llama 3, Mistral 7B, GPT 3.5, GPT 3.5 turbo, GPT 4, GPT 4o, GPT 5, Gemini 3. A generative data-driven model may be or comprise a mixture of experts architecture based model e.g. Mixtral 8x7B, Mixtral 8x22B. In a mixture of experts model several decoder blocks may be operated in parallel representing different experts or a feed-forward layer in a block may be split into separate parallel feed-forward layer, wherein each of the parallel feed-forward layers may be regarded as an expert and may learn to focus on different tasks during training. A gating network may be used to switch a particular input to the respective expert, e.g. by training the gateway network alongside the experts for instance using an expectation-maximization algorithm or a gradient descent algorithm. A generative data-driven model may be or comprise a selective state space sequence architecture based model e.g. Mamba. A generative data-driven model may be or comprise a combined architecture such as Mamba LLM or Mamba Mixture of Experts (which may comprise alternating Mamba and mixture of experts layers). A selective or structured state space sequence architecture (e.g. SSMs, S4, or S6 models) may allow for using more context in generation and allow for generating larger output output data sets. A generative data-driven model may be trained on general-purpose training data sets or based on (e.g. distilled from) a generative data-driven model trained on general-purpose training data sets.

[0101] For instance, when receiving a task instruction the generative data-driven model may process the task instruction. For instance, the (received) task instruction may be tokenized, e.g.by dividing the task instruction into tokens (i.e. smaller units), such as words or subwords. The task instruction or tokens of the task instruction may be embedded e.g. converted into a vector in the model's latent space, e.g. by an embedding layer of the generative data-driven model.

[0102] A generative data-driven model may be or comprise an artificial neural network (ANN), which may comprise several layers.

[0103] A generative data-driven model may comprise an embedding layer for embedding a received token into the model's latent space, which may be a numerical representation or vector of the token in the latent space.

[0104] The generative data-driven model may comprise a positional encoding layer, which may encode the position of a received token relative to the task instruction. For instance, using trigonometric functions such as sine and cosine a unique positional encoding vector for each position of a token in the task instruction may be generated. The positional encoding vector may then be added element-wise to the embedding of the token obtained by an embedding layer, so that the position of the token is encoded together with the embedding of the token in the embedded token passed to e.g. an encoder or decoder block.

[0105] A generative data-driven model may comprise one or more encoder layers forming an encoder block and / or one or more decoder layers forming a decoder block.

[0106] A decoder and / or encoder block may comprise a self-attention layer. A self-attention layer may be configured to (re-)encode an embedded token (e.g. a vector) by taking into account the context provided by all other tokens. A self-attention layer may be configured to weigh the relevance of different parts (i.e. tokens) of its input with regard to the (e.g. overall) input. The input of a self-attention layer may have the form of a matrix (e.g. input matrix) or tensor, in which each row or column may correspond to an embedded token, which may be a vector, The output of a self-attention layer may then also be a matrix (e.g. output matrix) or tensor, wherein each row or column may additionally contain information on the relevance of the token relative to the other tokens. For instance, the first self-attention layer may be configured to weigh the relevance of different tokens of a task instruction based on their relevance to the (e.g. overall) task instruction, where the overall task instruction may be represented as the matrix of all embedded tokens of the task instruction. In this case the input matrix for the first self-attention layer e.g. after an input layer, may comprise the embedded tokens of the task instruction, i.e. vector representations of each token of the task instruction. To determine the self-attention, an individual embedded token (e.g. a token vector or row / column vector of the input matrix) may be transformed into a set of vectors (also named a head) namely a query vector, a key vector, and a value vector by e.g. Linear projection, such as multiplying a token vector by a respectiveweight matrix. Then a scaled dot-product attention may be applied, e.g. by forming the dotproduct of each query vector with each key vector to calculate an attention score that may represent the relevance of a given token relative to the (e.g. overall) input of the self-attention layer, e.g. In particular for the first self-attention layer the relevance of the token to the task instruction. The attention scores may then be normalized using a softmax function, which converts each attention score into a respective softmax score, which may represent probabilities that sum up to 1 , so that e.g. the weight of higher attention scores is increased and the weight of lower attention scores is decreased. Then, each value vector may be multiplied by the respective softmax score to calculate a weighted value vector. The weighted value vectors - corresponding to the tokens of the task instruction - may then be summed up to determine a self-attention vector. The output matrix of the self-attention layer may then comprise the selfattention vectors. The self-attention layer may also utilize several sets of trained vectors (or heads) each comprising a query vector, a key vector, and a value vector. In this case, the output matrices resulting from calculating the self-attention of each head may be concatenated and multiplied by an additional trained weight matrix, which may allow to transform the matrix to the dimensions of the input matrix, the resulting matrix may correspond to the output matrix of the self-attention layer. This multi-head approach to self-attention may allow the self-attention layer to capture different types of information from the input matrix in each of the heads. For example, one head might focus on syntactic information, another on semantic information.

[0107] A decoder and / or encoder block may comprise at least one feed-forward layer, which may perform at least one linear transformation, preferably two linear transformation, wherein a Rectified Linear Unit activation function is applied between the two linear transformations.

[0108] A decoder block may comprise a cross-attention layer. For instance, similar to the selfattention layer an attention score between different data sets may be calculated. For example, a query vector may be determined based on tokens from the task instruction like in self-attention, whereas the key and value vectors may be determined based on context information, e.g. provided separately, which may e.g. be part of a data provision instruction.

[0109] A generative data-driven model may comprise an output layer, which may e.g. be configured to determine a probability distribution for the next token of an output sequence, e.g. which may be or be comprised by the data provision instruction. For instance, the output matrix of the last decoder block may be linearly transformed to e.g. a vocabulary size (which may be larger than the size of the corresponding dimension of the output matrix) and a softmax function may be applied to create a probability distribution, e.g. over the vocabulary. From this distribution a sampling module may sample the next token of the output sequence, e.g. byselecting the token having the highest probability or by selecting the k tokens with the highest probability and selecting one from the k tokens at random. The sampling module may use top k-sampling or top p-sampling. The output sequence generated may then be attached to the prompt and again processed by the generative data-driven model to determine the next token and so forth until a end token is generated or a maximum length threshold is reached, wherein the end token may be determined during training of the generative data-driven model.

[0110] The generative data-driven model may be a (pre-)trained or parametrized general purpose model parametrized or trained based on general data sets including input-output-data pairs not specific to input data related to accessing the decentral network.

[0111] A production environment may be configured to produce a product e.g. based on input material. It may comprise at least one production plant and / or production line comprising . The production environment may be a distributed production environment. The production environment may be a chemical production environment, wherein the output product is a chemical product. A chemical production environment may comprise a chemical apparatus. A chemical apparatus may be an apparatus, device, or equipment used in a chemical production process. A chemical apparatus may e.g. be a heater, cooler, heat exchanger, pump, compressor, pressure valve, a chemical reactor, a column such as a fractionating column, a furnace, a reaction chamber, a cracking unit, a storage tank, an extruder, a pelletizer, a precipitator, a blender, a mixer, a cutter, a curing tube, a vaporizer, a filter, a stack, an actuator, a mill, a transformer, a conveying system, a circuit breaker, a machinery e.g., a heavy duty rotating equipment such as a turbine, a generator, a pulverizer, a transport element such as a conveyor system, or a motor.

[0112] Information related to the at least one decentral network or set of decentral networks, such as routing information, e.g. network access routing information, data provision information, may be based on web data. For instance a knowledge database e.g. comprising data sets for (e.g. each) decentral networks in a set of decentral networks may be populated and / or updated based on web data. For instance, data sets descriptive of respective decentral networks may be obtained from web data and provided in the knowledge data base or may be generated based on the web data, web data that may be incorporated in data sets descriptive of respective decentral networks may e.g. comprise descriptions of production environments associated with (e.g. each of) the decentral network(s), products or output products associated with each of the decentral network(s) of the set, which data may be available on the internet. Hence, also instruction templates relating to to identification of at least one decentral network, setting up at least one edge data connector, preparing and testing end-to-end integration, deploying accessinfrastructure to the production environment, may be based on web data, e.g. by being generated based on said web data e.g. via a rule set or a generative data-driven model.

[0113] In an embodiment the knowledge database is populated and / or updated based on web data provided for or related to the one or more decentral networks. Web data may relate to decentral network specifications, decentral network identifiers, decentral network context data, data models per decentral network, data models per product, data model context data or any combinations or subsets thereof. Context data may relate to descriptions in natural language of respective data. Web data may be retrieved by or via a web interface configured to access web repositories and / or scrape webpages. Web data may be retrieved by or via a web interface configured to access web repositories and / or scrape webpages based on an update event received by the web repositories and / or scrape webpages. Web data may be mapped to data for populating and / or updating the knowledge database. Web data may be mapped to the data structure of the data is stored in by the knowledge database. Web data may include or relate to public data such as publicly accessible over the internet.

[0114] For instance, an operating system may be configured to fetch web data including data related to decentral networks such as data related to accessing at least one decentral network to provide and / or obtain a product data set associated with the output product. The operating system may include a web interface configured to fetch web data based on one or more Uniform Resource Locators (URLs) or application programming interfaces associated with a path to a web repository e.g. / repos / {owner} / {repo}. The operating system may include a data component communicatively coupled or connected to the web interface. The data component may include a web scraping component configured to receive and / or fetch web data including data as described above through the web interface based on web protocols such as html or xhtml data, metadata or semantic markups and annotations defined by web protocols such as Cascading Style Sheets (CSS) standard as defined by W3C, Selectors Level 4, W3C Working Draft, 11 November 2022. The data component may include a web repository component configured to receive and / or fetch web data through the web interface from internet repositories accessible through an application programming interface such as REST APIs. The web repositories or the URLs may provide access to web data including e.g. data related to decentral networks, decentral network protocols for respective decentral networks, supply chain product specific data models and / or context information for the respective decentral networks, protocols and / or data models based on web data. The web repositories or the URLs may provide the web data including e.g. data related to decentral networks, decentral network protocols for respective decentral networks, product specific data models and / or context information for the respectivedecentral networks, protocols and / or data models. The web data may be fetched in response to update event(s) provided by web components configured to notify subscribers when web content is updated. The data component may be configured to update the knowledge database storing data e.g. related to decentral networks, decentral network protocols for respective decentral networks, supply chain product specific data models and / or context information for the respective decentral networks, protocols and / or data models based on web data. The data component may be configured to receive update event(s) via the web interface e.g. by connection to web repositories accessible through application programming interface such as REST APIs. The data component may be connected, e.g. via an API, to pre-defined paths for accessing pre-defined web repositories storing web data including data as described above such as data related to one or more decentral network(s), decentral network protocol(s) for respective decentral network(s), supply chain product specific data model(s) for respective decentral network(s), and / or context information for the respective decentral networks, protocols and / or data models, such as semantic models stored on GitHub. For example, the data component may be configured to access such pre-defined web repositories. Further for example, the data component may be configured to receive updates from such pre-defined web repositories. The data component may receive notifications when one of the pre-defined web repositories is updated and may collect updated data in response to such notification(s). The data component may receive updated data when one of the pre-defined web repositories is updated. The web data may be fetched based on one or more lists including pre-defined paths to web content relating to decentral network protocols for exchanging product data associated with the supply chain product and / or respective one or more data model(s) related to product data associated with the supply chain product to be exchanged by the decentral network protocol. The data component may be configured to provide paths or representations such as urls locating the web content to be scraped e.g. to the web interface. For example, the data component may be configured to access and / or provide to the web interface one or more pre-defined lists of urls related to one or more decentral network(s), decentral network protocol(s) for respective decentral network(s), supply chain product specific data model(s) for respective decentral network(s), and / or context information for the respective decentral networks, protocols and / or data models and to be scraped. The web data may be fetched based on one or more pre-defined search terms for searching web content related to decentral network protocols for exchanging product data associated with the supply chain product and / or respective one or more data model(s) related to product data associated with the supply chain product to be exchanged by the decentral network protocol.

[0115] For example, the data component may be configured to access and / or provide to the web interface one or more pre-defined search terms for searching content related to one or more decentral network(s), accessing one or more decentral networks to provide and / or obtain product data sets associated with an output product, decentral network protocol(s) for respective decentral network(s), product specific data model(s) for respective decentral network(s), and / or context information for the respective decentral networks, protocols and / or data models and locating respective urls to be scraped. The data component may be configured to receive web repository and / or web data. The data component may be configured to receive data related as described above. The data may include the specification of the decentral network, respective decentral network protocols defined for the specified decentral network, product specific data models defined for the specified decentral network and / or context information for the specified decentral network, protocols and / or data models. An operating system may be configured to generate data for the knowledge database by mapping the fetched web data to a data structure as stored by the knowledge database. The knowledge database may store key-value pairs of decentral network identifiers and respective edge connector set up instructions. The web repositories, or the URLs may provide web data that may be mapped to the data as stored by the knowledge database. The data component may be configured to receive web repository and / or web data and extract data in a structured format such as table formats or semi-structured format such as JSON or JSON Lines. This may include cleansing of HTML data, validating scraped data e.g. by checking that the data items contain certain fields or checking for duplicates and dropping them. The data component may be configured to receive data in a structured format such as table formats or semi-structured format such as JSON or JSON Lines. Processing and storing of the data in the knowledge database may be logged and stored in a documentation storage. Data in the knowledge database may be updated frequently in preset or dynamic time intervals. The data component may be configured to store data in a structured format such as table formats or semi-structured format such as JSON or JSON Lines. Descriptions of the decentral networks or network protocols for exchanging product data (e.g providing and / or obtaining product data sets associated with an output product) may be extracted from the fetched web data. The data component may be configured to receive context information for the specified decentral network e.g. in natural language e.g. from html content of webpages or web repositories. The knowledge database may store descriptions in natural language for data points stored as key-value pairs. The data component may be configured to generate context information for the specified decentral network e.g. by using a generative data- driven such as a Large Language Model (LLM). The generative data-driven model may be general purpose in the sense that it was trained or is pre-trained on unstructured naturallanguage data and is configured to process natural language. The data-driven model may be general purpose in the sense that is not trained for a specific task. The data-driven model may include a general-purpose model trained at least on unstructured data or natural language data. Descriptions of the decentral networks or network protocols for exchanging product data (e.g providing and / or obtaining product data sets associated with an output product) may be generated using a Large Language Model. One or more instruction(s) may be generated and provided to the Large Language Model. The instructions may include at least one task instruction specifying the task to generate descriptions of the decentral networks. The instructions may include fetched web data as context, wherein the instructions may indicate the key-value pair(s) of data structure to be described. One or more instruction(s) may be provided to the generative data driven model e.g. together with parameters for the generative data-driven model such as maximal token specification or authentication / authorization parameters. The generative data- driven model may be selected from one or more data-driven models based on a functionality, a sensitivity and / or confidentiality level of processing by the data-driven model. The generative data-driven model may generate and / or provide context descriptions for the specified decentral network. The context descriptions may be stored in relation to or for the specified decentral network. The context descriptions for the specified decentral network, protocols and / or other data related to the decentral network as described above may be provided in natural language or text and stored as metadata in association with the specified decentral network in the knowledge database. The data component may be communicatively coupled or connected to the knowledge database.

[0116] Further possible implementations or alternative solutions of the invention also encompass combinations - that are not explicitly mentioned herein - of features described above or below in regard to the embodiments. The person skilled in the art may also add individual or isolated aspects and features to the most basic form of this disclosure.

[0117] Other features will become apparent from the following detailed description considered in conjunction with the accompanying drawings. It is to be understood, however, that the drawings are designed solely for purposes of illustration and not as a definition of the limits, for which reference should be made to the appended claims. It should be further understood that the drawings are not drawn to scale and that they are merely intended to conceptually illustrate the structures and procedures described herein.BRIEF DESCRIPTION OF THE SEVERAL VIEWS OF THE DRAWINGS

[0118] In the following, the present disclosure is further described with reference to the enclosed figures. The same reference numbers in the drawings and this disclosure are intended to refer to the same or like elements, components, and / or parts.

[0119] FIG. 1 shows an example of a production environment providing an output product to two further production environments, wherein the production environments are connected via decentral networks.

[0120] FIG. 2 illustrates an example of a decentral network of a product ecosystem including a material loop and being associated with a decentral peer-to-peer network for exchange of data associated with raw materials, chemical product(s), discrete product(s), end product(s) and recycled material(s).

[0121] FIG. 3 illustrates an example of a decentral network of a battery ecosystem including a material loop and being associated with a decentral peer-to-peer network for exchange of data associated with raw materials, chemical product(s), batteries, end product(s) and recycled material(s).

[0122] FIG. 4 illustrates another example method for accessing decentral network to provide and / or consume a product data set associated with an output product.

[0123] FIG. 5 shows an example of the method according to the first aspect.

[0124] FIG. 6A illustrates another example method for accessing decentral network to provide and / or consume a product data set associated with an output product.

[0125] FIG. 6B illustrates another example method for accessing decentral network to provide and / or consume a product data set associated with an output product.

[0126] FIG. 7 illustrates an example of fine-tuning a general-purpose generative data-driven model.

[0127] FIG. 8 illustrates an embodiment of input embedding.

[0128] FIG. 9 illustrates an embodiment of a transformer encoder architecture.

[0129] FIG. 10 illustrates an embodiment of a transformer decoder architecture.

[0130] FIG. 11 illustrates an embodiment of a transformer encoder-decoder architecture.

[0131] FIG. 12 illustrates an embodiment of training and / or deploying the transformer encoder.

[0132] FIG. 13 illustrates an embodiment of a structured space-state sequence architecture.

[0133] FIG. 14 shows a schematic block diagram of an example apparatus.DETAILED DESCRIPTION

[0134] The following embodiments are mere examples for implementing the method, the system or application device disclosed herein and shall not be considered limiting. The following description serves to deepen the understanding and shall be understood to complement and be read together with the description as provided in the above summary and embodiment sections of this specification. Some aspects may have a different terminology than e.g. provided in the description above. The skilled person will nevertheless understand that those terms refer to the same subject-matter, e.g. by being more specific.

[0135] FIG. 1 shows a plant 148 (as an example of a production environment) producing an output product 164 used as an input material for the production of products 168, 170 by other participants (here other production environments 166, 172) of the respective product ecosystem. Producer 148 needs to access the correct decentral networks 162, 158 via which to provide (e.g. output) product data sets associated with product 164 to producer 166 via a node (e.g. a decentral participant node) of decentral network 162 and to producer 172 via a node (e.g. a decentral participant node) of decentral network 158. Production environment 148 may be associated with a network node of the decentral network 158 and with another network node (e.g. a decentral participant node) of decentral network 162 to provide product data in respective product data sets. Production environment 148 may be a data owner of the product data. The product data sets may need to adhere to different respectively required data structures for the different decentral networks, which in turn may further depend on the respective end product produced, e.g. a car or washing machine. Using the respective product data sets a digital product passport may be generated e.g. for car 168 or washing machine 170 both using output product 164 as input material. Further examples of such decentral networks with different participants will be discussed with reference to FIG. 2 and FIG. 3.

[0136] FIG. 2 illustrates an example of a participant network of a product ecosystem associated with a decentral network (or peer-to-peer network) for exchange of data associated with output products such as raw materials, chemical product(s), discrete product(s), end product(s) and recycled material(s). The decentral participant network 230 may include one or more decentral network participants, such as decentral participants 202 to 214 (e.g. a chemical production environment 206). The decentral network participants may be part of a product ecosystem including chemical products. The product ecosystem may include production chains to produce an end-product. The product ecosystem may include recycling chains to recycle at least part of an end-of-life product resulting from the use of the end product. The product ecosystem may include a raw material producer 204, a chemical product producer 202, a chemical product user 206, an end-product producer 208, an end-product user 210 an EOL product collector 212 anda recycler 214. The decentral participant network 230 may include a chemical supply chain. The product ecosystem may allow to use of recycled materials resulting from recycling of end-of-life products to produce new products, such as chemical products. The product ecosystem may be associated with the production and / or recycling of physical products. The product may be a chemical product, an intermediate chemical product, a component, a component assembly, an end product, an end-of-life product or a recycled material.

[0137] At least a part of the participant(s) of the decentral participant network 230 may be associated with the production of the product and / or the recycling of end-of-life products resulting from the use of the product by product users, such as end-product users 210. The decentral network participant 202 to 214 may refer to a manufacturer of physical products, such as raw material producer 204, chemical product producer 202, chemical product user 206, endproduct producer 208, a user of physical goods, such as end-product user 210, and / or a participant of a recycling chain associated with the physical product, such as EOL product collector 212 and recycler 214. The decentral network participant may be associated with a decentral participant identifier. The decentral participant identifier may uniquely identify the decentral network participant within the decentral participant network 230.

[0138] At least a further part of the participant(s) of the decentral participant network 230 may be associated with the generation of product passport(s). Such decentral participants may not be associated with the production of the product and / or the recycling of the end-of-life products. Such decentral participants may gather input material data and may generate product passports using at least a part of the gathered input material data. The generated product passports may be provided by such participants for access via the decentral network 230. For instance, recycler 214 may access such product passports to determine the composition of the end-of-life product, allowing adaption of the recycling process to the determined composition.

[0139] The participant(s) of the decentral participant network 230 may be connected via material flows. The material flow may be a loop material flow 236. The loop material flow 236 may be a closed loop material flow. A closed loop material flow may refer to a material loop where recycled material is used to produce the same end products the recycled material is obtained from via recycling. The loop material flow 236 may be an open loop material flow. An open loop material flow may refer to a material loop where recycled material is used to produce different end products than the one the recycled material is obtained from. The material flow may be a linear material flow (e.g. not including recycling). The material flow 236, 238 may correspond to the flow of product from one participant of the decentral participant network 230 to the downstream participant of the decentral participant network 230. The material flow 236,238 may refer to a continuous or a discontinuous flow of product. The flow of product may include any means of transportation suitable to transport the product from a participant to the downstream participant. The means of transportation may include pipes, containers, barrels, packages. The material flow 238 may be associated with raw materials used to produce a chemical product, such as virgin raw materials. The raw materials may be provided to chemical product producer 202 for producing chemical product(s) and / or intermediate chemical product(s) (not shown). The loop material flow 136 may be associated with chemical product(s) and discrete product(s). The chemical product(s) may be provided from chemical product producer 102 to chemical product user 106 for producing discrete product(s). In contrast to chemical production, the discrete products being produced are distinct units sold as individual products. The loop material flow 136 may be associated with recycled material. The recycled material may be provided from recycler 114 to chemical product producer 102 for the production of chemical product(s) using the recycled material.

[0140] At least part of the participants of the decentral participant network 230 may be associated with decentral participant network nodes 216 to 228. The decentral participant nodes 216 to 228 may be under control of the respective decentral participant associated with the respective decentral participant node. The decentral participant nodes 216 to 228 may form decentral network 234. The decentral network 234 may be a peer-to-peer communication network. The decentral network 234 may be configured to perform data transactions 232. The data transactions 232 may be based on a transaction protocol including authentication and / or authorization mechanism(s). Based on the authentication and / or authorization mechanism(s) a peer-to-peer communication between decentral network nodes 216 to 228 associated with decentral network participants 202 to 214 may be established. The one or more authentication mechanism(s) may be associated with or linked to an identifier. The one or more authentication mechanism(s) associated with the identifier may be accessible by the decentral participant nodes. The decentral configuration allows for more efficient use of computing resources and strengthens control by the data owners of the decentral network.

[0141] Data transactions / exchange (such as obtaining product data sets) between decentral network participant nodes may be based on an identifier associated with respective data to be accessed. The identifier may be uniquely associated with the physical entity of the respective product and associated product data. The identifier may be a decentral identifier uniquely identifying the product within the decentral network. The identifier may be a local identifier used by the respective product producer to uniquely identify the respective product. The identifier may be associated with further identifier(s), such as identifier(s) of production input(s) (e.g. inputmaterials) used to produce the output product. This may allow to track the product input(s) used to produce a product, such as an end-product. The identifier may, for example, be included in an output product data set associated with the product.

[0142] The data flow 232 (e.g. transactions) between decentral network participant nodes may be directly or indirectly associated with the material flow 236, 238 between the decentral network participants. For instance, data flow 232 may be directly associated with material flow 236, 238 if data associated with an input material provided from the raw material producer 204 to the chemical product producer 202 is accessed by decentral participant node 218 associated with said chemical product producer 202. For instance, data flow 232 may be indirectly associated with material flow 236, 238 if data associated with a chemical product produced by chemical product producer 202 is accessed by decentral participant node 228 associated with recycler 214.

[0143] The decentral participant nodes 216 to 228 may be decentral computing nodes. The decentral computing node may be any device or system that includes at least one physical and tangible processor, and a physical and tangible memory capable of having thereon computerexecutable instructions that are executed by a processor. The memory may take any form and depends on the nature and form of the computing node.

[0144] At least part of the decentral participant nodes 216 to 228 may be decentral data providing network nodes. At least part of the participant nodes 216 to 228 may be decentral data consuming network nodes. A participant of the decentral participant network 230 may be associated with a decentral data providing network node and / or a decentral data consuming network node depending on whether data is provided to downstream participants and / or consumed from upstream participants. For instance, end-product producer 208 may be associated with a decentral data providing network node configured to provide product data to a downstream participant (e.g. recycler 214). In addition to or alternatively, end-product producer 208 may be associated with a decentral data consuming network node configured to access data associated with a discrete product produced by an upstream participant (e.g. chemical product user 206).

[0145] The decentral network 234 may include further decentral network nodes. The further decentral network nodes may be decentral infrastructure service nodes (not shown in FIG. 2). The decentral infrastructure service nodes may not be associated with a participant of the product ecosystem. The decentral infrastructure service nodes may provide services for decentral participant nodes 216 to 228, such as verifying the identity of the decentral network participant nodes 216 to 228 prior to performing a data exchange. The decentral networkparticipant nodes 216 to 228 may be associated with or include certificate(s), such as X.509 certificate(s). The certificate(s) may be associated with decentral infrastructure service node(s) including e.g. a certificate issuing service and / or a dynamic provisioning service providing dynamic attribute tokens (e.g. OAuth Access Tokens). This way the decentral network participant nodes 216 to 224 possess a unique identifier embedded in a X.509 certificate that identifies the respective decentral network participant node 216 to 228. The information required to verify the certificate may be provided via an authentication registry associated with the certificate issuing service and / or a dynamic provisioning service. For instance, in the IDSA Reference Architecture Model, Version 3.0 of April 2019, a decentral data providing network node associated with a data owner, a Certification Authority (CA), a Dynamic Attribute Provisioning Service (DAPS) and a decentral data consuming network node associated with a data consumer are used to verify the identity prior to performing a data exchange.

[0146] FIG. 3 illustrates an example of a participant network of a battery ecosystem including a material loop and being associated with a decentral (peer-to-peer) network for exchange of product data associated with raw materials, chemical product(s), batteries, end product(s) and recycled material(s). The decentral participant network 316 may include one or more decentral network participants, such as decentral network participants 148, 166, 302 to 312. The battery ecosystem may include production chains to produce an (end) product, such as a machine containing a battery. The battery ecosystem may include recycling chains to recycle at least part of an end-of-life battery resulting from the use of the end product. The battery ecosystem may include a miner 302, a refiner 304, a precursor cathode active material (PCAM) and cathode active material (CAM) producer 306, a battery producer 308, an end-product producer 302, an EOL product collector 304, a black mass producer 310 and a metal extractor 312. The battery ecosystem may allow to use recycled materials, such as recycled metals and metal salts, resulting from recycling of end-of-life batteries or components thereof to produce new products, such as PCAM and CAM. The product ecosystem may be associated with the production and / or recycling of batteries.

[0147] The participant(s) of the decentral participant network 316 may be associated with the production of a battery containing end product and / or recycling of end-of-life batteries or components thereof. The decentral network participant may be a manufacturer of physical products, such as miner 314, refiner 316, PCAM & CAM producer 318, battery producer 320, end-product producer 302 and / or a participant of a recycling chain associated with the end-of- life batteries or components thereof, such as EOL product collector 304, black mass producer 322 and metal extractor 324. For instance, the metal extractor 324 and the PCAM & CAMproducer 318 may be a single entity performing recycling operations to obtain recycled material, such as recycled metals and / or metal salts, and producing new chemical product(s), such as PCAM and / or CAM using the recycled metal and / or metal salts. The decentral network participant may be associated with a decentral participant identifier. The decentral participant identifier may uniquely identify the decentral network participant within the decentral participant network 316.

[0148] At least a further part of the participant(s) of the decentral network 308 may be associated with the generation of battery passport(s). Such decentral participants may not be associated with the production of the battery containing end product and / or the recycling of the end-of-life batteries or components. Such decentral participants may gather input material data, for example as described in the context of FIG. 1 and may generate battery passports using at least a part of the gathered input material data. The generated battery passports may be provided by such participants for access via the decentral network 308. For instance, a recycler may access such battery passports to determine the chemical composition of the end-of-life batteries or components thereof, allowing adaption of the recycling process to the determined composition.

[0149] The participant(s) of the decentral participant network 316 may be connected via material flows as described in the context of FIG. 2. The raw materials, such as metals, may be provided to refiner 316 for refinement. The refined metals may be provided to PCAM & CAM producer 318 for the production of cathode active material. The CAM may be used, for example by battery producer 320, to produce battery cells. The battery cells may be used to produce batteries or battery packs. The batteries or battery packs may be provided to end-product producer 302 to produce battery containing end products, such as electric vehicles. Scrape from battery production may be provided to black mass producer 310.

[0150] At least a part of the participants of the decentral participant network 316 may be associated with decentral participant network nodes 328 to 342 as described in the context of FIG. 2. The decentral participant nodes 328 to 342 may form decentral network 308. The decentral network 308 may be a peer-to-peer communication network as described in the context of FIG. 2. The decentral configuration allows for more efficient use of computing resources and strengthens control by the data owners of the decentral network.

[0151] Data transactions between decentral network participant nodes may be based on an identifier associated with respective data to be accessed. The identifier may be uniquely associated with the physical entity of the respective product and associated product data. The identifier may be a decentral identifier uniquely identifying the product within the decentralnetwork. The identifier may be a local identifier used by the respective product producer to uniquely identify the respective product. The identifier may be associated with further identifier(s), such as identifier(s) of production input(s) (e.g. input materials) used to produce the output product. This may allow to track the product input(s) used to produce a product, such as an end-product. The identifier may, for example, be included in input material data (or its respective product data set) associated with input materials.

[0152] The data flow (e.g. transactions) between decentral network participant nodes may be directly or indirectly associated with the material flow 310 between the decentral network participants. The decentral participant nodes 328 to 342 may be decentral computing nodes

[0153] At least part of the decentral participant nodes 328 to 342 may be decentral data providing network nodes. At least part of the participant nodes 328 to 342 may be decentral data consuming network nodes. A participant of the decentral network 308 may be associated with a decentral data providing network node and / or a decentral data consuming network node depending on whether data is provided to downstream participants and / or consumed from upstream participants (see also FIG. 1). The decentral network 308 may include further decentral network nodes as described in the context of FIG. 2.

[0154] FIG. 4 illustrates an example method for accessing decentral network to provide and / or consume a product data set associated with an output product. From an operator of the production environment, a user instruction including natural language is obtained (e.g. via a web interface). A routing or orchestrator agent receives the user instruction including at least data indicative of the production environment, the output product, and at least one other production environment, the at least one other production environment configured to produce a product based on the output product and / or the at least one other production environment providing input material for producing the output product. The routing or orchestrator agent determines from a number of agents specific for different decentral networks which agent is suitable to process the user instruction. In this example, it determines an agent specific for decentral network 1 as part of a tool sequence to be carried out, that may e.g. comprise further tools such as a conversion tool to convert certain data from the user instruction into a format better processed by the agent specific for decentral network 1 . To determine a tool sequence the routing or orchestrator agent may obtain a set of tool data sets, wherein each tool data set of the set of tool data sets is associated with a tool of the at least one tool; wherein the determining the tool sequence is further based on the set of tool data sets.

[0155] The one or more tool(s) comprising the at least one agent(s) are carried out according to the tool sequence. In this example, the agent specific for decentral network 1 determines thatthe user instruction requires specific questions better handled by a further agent specific for decentral network 1 , in this case an agent specific for setting up an endge data connector. To determine the further agent, the agent specific for decentral network 1 may obtain a set of tool data sets specific for decentral network 1 comprising at least an indication of the agent specific for setting up the edge data connector and an indication of an agent specific for deploying access infrastructure to the production environment. The agent agent specific for setting up the edge data connector may then be carried out.

[0156] This staged set-up with routing at different stages may enable more robust and reliable results. However, a set of tool data sets obtained by the routing agent may already comprise indications of agents specific for different decentral networks, e.g. agent specific for setting up the edge data connector in a first decentral network and agent specific for setting up the edge data connector in a second decentral network. In this example the routing agent may already determine them as part of the tool sequence.

[0157] Based on received output data from the one or more one or more tool(s) or agents the data provision instruction for accessing the at least one decentral network is determined, for instance by the routing or orchestrator agent or a tool configured to determine the data provision instruction based on the output data, e.g. via the generative data-driven model as described above in an analogous manner.

[0158] In the example of FIG. 4, the routing agent determines that the user instruction relates to decentral network 1 and routes it to an agent specific for decentral network 1. The agent specific for decentral network 1 may generate a data provision instruction or in turn may route the user instruction further to an agent configured for generating a data provision instruction, when it is determined that an agent exists that is more specifically related to the user instruction, e.g. related to a certain task such as setting up an edge data connector. Routing in this case may be implemented similar as described above e.g. in relation to the routing agent, as information for routing e.g. a list of available agents alongside a description of the agent may be used.

[0159] The agent specific for setting up the edge data connector may be provided e.g. to the operator or a computer system associated with the production environment.

[0160] FIG. 5 shows an example of the method according to the first aspect. An operator sends a user instruction comprising data indicative of a production environment (e.g. the operator's production environment), the output product, and another production environment configured to produce a product based on the output product. The user instruction may be pre-processed, e.g. prior to passing it to the orchestrator agent or by the orchestrator agent. Pre-processing maycomprise identifying a term associated with a decentral network, production environment, product or the like. For instance, a term may be searched, e.g. using a keyword search, in a data structure that allows associating terms with terms of similar meaning or further information related to the term, which may enable retrieval of e.g. routing information to enrich the user instruction.

[0161] An orchestrator agent may based on the user instruction or pre-processed user instruction determine a tool sequence, wherein a tool sequence comprises an indication of one or more tool(s), wherein the one or more tools comprises at least one agent configured to determine at least a part of the at least one data provision instruction, by providing a task instruction, based on the user instruction, to at least one generative data-driven model, the at least one generative data-driven model being configured to generate the tool sequence in response to obtaining the task instruction.

[0162] The tool sequence may be determined by providing a task instruction for generating the tool sequence to a / the generative data-driven model, wherein the task instruction is generated based on the user instruction and the set of tool data sets, the generative data-driven model having been trained on general purpose training data sets and being configured to generate a tool generated data set related to the tool sequence, in response to receiving the task instruction, wherein the tool generated data set may be provided as the tool sequence.

[0163] The orchestrator agent may retrieve / receive a set of tool data sets from a tool database, wherein the tool data sets may comprise descriptions of respective tools (e.g. functions or agents for specific tasks) including any input and output data required to execute the tools, e.g. in json format. A set of tool data sets may e.g. be a list of tools and their associated descriptions.

[0164] Based on the set of tool data sets, the user instruction or pre-processed user instruction and the tool data set(s) a tool sequence is determined or caused to be determined (e.g. by the orchestrator agent). A tool sequence may comprise an ordered list of tools specifying which tools to execute in what order for processing the user instruction. For instance, a tool could be a function or an agent requiring specific input and providing specific output. For instance, determining the tool sequence may comprise providing a task instruction for generating the tool sequence to a generative data-driven model, the generative data-driven model having been trained on general purpose training data sets and being configured to generate (e.g. and provide) the tool sequence, in response to receiving the task instruction (such a generative data- driven model may e.g. be GPT4, GPT4o, GPT5, GPT5-mini, Gemini 3). The task instruction may be based on the set of tool data sets and the user instruction.

[0165] The tool sequence may be determined by providing a task instruction for generating the tool sequence to a / the generative data-driven model, wherein the task instruction is generated based on the user instruction and the set of tool data sets.

[0166] The orchestrator agent retrieves / receives a set of tool data sets from a tool database, wherein the tool data sets may comprise descriptions of respective tools including any input and output data required to execute the tools, and data indicative of the function of the respective tool e.g. in json format. A set of tool data sets may e.g. be a list of tools and their associated descriptions.

[0167] Based on the set of tool data sets, the user instruction or pre-processed user instruction and the tool data set(s) a tool sequence is determined or caused to be determined (e.g. by the orchestrator agent). A tool sequence may comprise an ordered list of tools specifying which tools to execute in what order for processing the user instruction.

[0168] Subsequently, the tools are carried out, caused to carry out or executed according to the tool sequence e.g. by a routing module or component of the orchestrator agent. The output data provided by the tools may then be received e.g. by the orchestrator agent.

[0169] As an example of carrying out the tools according to the tool sequence, a first tool according to the tool sequence is carried out. Output data from the first tool indicated in the tool sequence is received. Upon receiving output data from the first tool, the orchestrator agent determines an operation instruction to a second tool indicated in the tool sequence, e.g. by providing a task instruction for generating the operation instruction to the generative data-driven model, the task instruction comprising at least a part of the output data from the first tool and optionally data on the required input for the second tool, e.g. a specific format, or other information on the second tool which may be available from the tool data set. Subsequently, the second tool is carried out based on the operation instruction. This process may be repeated for some or all tools according to the tool sequence, taking for instance into account at least part of the output data from previous tools for carrying out subsequent tools according to the tool sequence.

[0170] When output data from all tools of the tool list is received, which include at least a part of the at least one data provision instruction, the at least one data provision instruction is determined, e.g. by passing the output data to a generative data-driven model to generate the at least one data provision instruction based on the output data and the user instruction.

[0171] In case the orchestrator agent determines that data is missing, the user may be instructed to provide additional data.

[0172] The data provision instruction prompts e.g. a computer system of the production environment or an operator of the production environment to provide certain data, e.g. on the production environment, provided IT infrastructure intended to be used for accessing the decentral network etc. in form of an operation data set. Based on the operation data set, at least one of an identification associated with the production environment or security credential(s) for accessing the at least one decentral network for obtaining and / or providing the product data set is determined and provided to the operator.

[0173] FIG. 7 shows an example of a training and fine-tuning process to obtain a fine-tuned generative data-driven model 706, which may be used as a generative data-driven model in the disclosed method. A general-purpose generative data-driven model 704 may have been (predrained using a large number of training data sets, which may be unlabeled data sets, in an unsupervised manner. Training may involve tokenizing input texts and masking a number of tokens of the input text. The weights of the generative data-driven model may then be adjusted based on the accuracy of generating the masked tokens, wherein the accuracy may be measured by a metric such as in cross-entropy loss or maximum likelihood estimation. The pretrained model 704 may then be fine-tuned 720 using for example a number of labeled training data sets 722 specific for the tasks a certain agent is meant to perform, e.g. an access or routing agent. The training data may be based on web data obtained from websites providing information for accessing at least one decentral network to provide and / or obtain a product data set associated with an output product. The traning data may comprise historic user instructions and respective (correct) data provision instruction. Preferably, low-rank adaptation or parameter-efficient fine-tuning (PEFT) is used for fine-tuning 720, which may allow for efficient fine-tuning and may reduce the risk of the pre-trained general purpose generative data-driven model 704 losing the pre-trained weights (i.e. catastrophic forgetting). The fine-tuning 720 may involve creating a number of training user instructions, e.g. by consulting experts on accessing the respective decentral network to provide and / or obtain a product data set. The specific finetuned model 706 may be used as the generative data-driven model accessed or used by the correspondingly tasked agent.

[0174] FIG. 8 illustrates an example embodiment of input embedding in particular for generative data-driven models or machine-learning architectures using an embedding layer 802 such as a transformer encoder, transformer decoder or transformer encoder decoder architecture.

[0175] An input embedding may be obtained by training for example a continuous bag of words model (CBOW) or a skip-gram model. The embedding layer may be suitable for generatingembedded input data based on input data. Generating embedded input data may refer to embedding input data. Embedding input data may result in a representation associated with the input data. Thus, the embedded input 814 may be the representation associated with the input data. The input data may comprise one or more elements. The one or more elements may be represented by the input vector 806. In particular, the embedded input 814 and / or the input vector 806 may be machine- readable and / or processable by a processor. For this purpose, the embedded input 814 and / or the input vector 806 may be a tensor, in particular a first-rank tensor. Specifically, the input vector 806 may be a one-hot vector or a summation of a plurality of one- hot vectors. A one-hot vector may be a vector with one entry unequal to zero. Examples for one- hot vectors may be 808, 810 and 812. The entries unequal to zero in the one-hot vector and / or in the input vector 806 may indicate the element. For example, a lookup table may define the relation between the position of the entries unequal to zero and the element indicated by the one-hot vector. The lookup table may specify a plurality of different elements. The number of different elements may be equal to the number of entries in the one-hot vector. The number of different elements may be referred to as vocabulary size. In an example, the elements may be represented by tokens and a sequence of elements may refer to at least a part of a sentence. The at least a part of the sentence may be represented by a plurality of tokens. A token may represent at least a part of the element and / or word. For example, where one element would be associated with only one word, words such as “embeddings", “embedding” or “embed” would constitute different elements. A first token may represent the stem “embed” and the endings, typically appearing in a plurality of word, may be represented by a second token, a third token and a fourth token. The second token, the third token and the fourth token may be used for representing other words such as “look”, “looking” or the like, preferably together with a fifth token representing the stem “look”. Ultimately, this tokenization of elements associated with a plurality of stems and a plurality of endings results in less tokens to be used for representing a plurality of elements and thus, uses less computational resources.

[0176] A lookup table specifying a subset of the vocabulary size e.g. of the English language may comprise 10,000 words or more. The embedded input 814 may be a lower-dimensional representation than the input vector 806. For example, typical embedded inputs 814 may comprise some hundreds of different entries. Followingly, the embedded inputs 814 constitute a densified representation of one or more elements using less computational resources. More than that, the embedded input 814 may represent a relation between two or more elements. For example, the words “Italy” and “Germany” may be similar or may be more closely related since they both define European countries, whereas the the word “embodiment” may be very different from the two respective words. The smaller the dot product between two embedded inputs 814may be the more similar the two elements associated with the embedded inputs 814 may be. Hence, the embedded inputs 814 may represent one or more elements accurately and lead to accurate results based on processing the embedded inputs 814.

[0177] For transforming the input vector 806 into the embedded input 814, the embedding layer may comprise a number of neurons equal to the number of entries in the embedded input 814. Based on the embedded inputs 814, the output layer may generate the output vector 816. The output vector may be a vector and / or may indicate one or more elements. The output vector 816 may indicate one or more elements different from the input vector 806 and / or the one-hot vectors associated with the input vector 806. For this purpose, the output layer may comprise a number of neurons equal to the number of entries of the input vector 806 and / or the output vector 816. The output layer may apply a softmax function to the embedded inputs 814. By doing so, the output vector may comprise the probabilities associated with the elements associated with the entries of the output vector 816 unequal to zero. Hence, from the output vector 816 one or more elements may be obtained with a corresponding probability. Where the input vector 806 may specify one or more sequence(s) of elements, the output vector 816 may specify one or more elements corresponding to the sequence(s) of elements specified by the input vector 806. In the example of FIG. 8, the element associated with vector 818 may correspond to the input vector with a probability of 71 %. Additional or alternative elements may correspond to the input vector as indicated by the output vector with lower probability. By defining a threshold to which the probability may be compared, the selection of the corresponding elements may be tailored to the needs of the user. The elements generated by the model comprising the embedding layer 802 and the output layer 804 may refer to the most probable elements indicated by the output vector 816. Hence, the model depicted in FIG. 8 may generate the element associated with the vector 818 with a confidence score of 71 %.

[0178] The model of FIG. 8 may be continuous bag of words (CBOW) model. The CBOW model may be trained based on a training data set comprising a plurality of input vectors and corresponding output vectors. As the training data set may not be labeled, the training of the CBOW model may be referred to as self-supervised. Before training of the CBOW model, the CBOW model may be initialized with random values assigned to the weights of the neurons. During the training of the CBOW model, the input vectors may be passed through the initialized embedding layer and the output layer and a loss may be determined by comparing the output vector obtained by passing the input vector 806 through the model to the output vector corresponding to the input vector 806 as specified by the training data set. Based on the determined loss, backpropagation may be applied to determine the gradients associated withthe neurons of the embedding layer 802 and the output layer 804 to lower the loss. According to the determined gradients, the weights of the neurons may be updated by using a gradient descent algorithm. If a predetermined loss may be achieved by the CBOW model, the training may be terminated and a trained CBOW model may be obtained. From the trained CBOW model, the embedding layer 802 may be suitable for embedding input data comprising one or more elements. This embedding layer 802 may be used in other machine-learning architectures requiring an embedding layer 802 such as a transformer encoder, transformer decoder or transformer encoder decoder architecture as described within the context of FIG. 9, FIG. 10 and FIG. 11. For training these architectures, a trained embedding layer 802 may be required. Hence, a model such as a CBOW model may be trained prior to training the transformer encoder, transformer decoder or transformer encoder decoder architecture.

[0179] Further, applying input embedding may include determining a numerical representation of the input data by determining the number of elements and / or parts of the input data. Hence, the numerical representation of the two or more elements, in particular of a predefined size, may be indicative of a number of occurrences of the elements and / or parts of the input data. In an example, the numerical representation of the two or more elements indicative of a number of occurrences of the elements of the input data may be a vector with a plurality of entries where one entry may be indicative of the occurrence of one element of the input data.

[0180] FIG. 9 illustrates an embodiment of a transformer encoder architecture e.g. of an encoder-only transformer model.

[0181] The transformer encoder comprises an encoder input 924, one or more encoder blocks 920, 914 and an encoder output 922. In particular, the transformer encoder may be referred to as X-former. The transformer encoder architecture may correspond to the encoder architecture associated with the transformer encoder-decoder architecture with an additional encoder output instead of connecting the encoder block directly to the decoder of the transformer encoderdecoder architecture. An example of a transformer encoder architecture is the bi-directional encoder representations from transformers (BERT).

[0182] The input data may be received at the encoder input 924. The input data may comprise at least one of text data, numerical data, tabular data, image data or the like. Where the input data may comprise one of text data, numerical data, tabular data, image data or the like, input embedding of a type corresponding to the type of input data may be applied. The type of the input data may be text data, numerical data, tabular data, image data or the like. In an embodiment, the input data may be associated with two or more types of input data. The input embedding may be associated with the two or more types of input embedding, in particularaccording to the input data. Hence, the input embedding may be configured to map text data, numerical data, tabular data, image data or the like to a numerical representation of the input data. In particular, at least one first type of input embedding may be applied to at least a part of the input data associated with one first type of input data. Further, at least one second type of input embedding may be applied to at least a part of the input data associated with one second type of input data. The model associated with the input embedding comprising the at least one first and at least one second type of input data may be referred to as multimodal model. The type of the input data may correspond to a modality. An example of input embedding associated with text data can be found in the context of FIG. 8.

[0183] Receiving and / or providing the input data may comprise identifying two or more elements of the input data. This may be referred to as tokenization. For this purpose, a vocabulary may be available. The vocabulary may specify a plurality of elements, in particular elements typically repeating in data of the type of the input data. For example, where the input data may be text data, the vocabulary may comprise several endings and / or word stems. In an embodiment, the elements of the input data may be specified by a selection indicative of the plurality of elements provided.

[0184] The encoder input 924 may apply an input embedding 902, in particular to the two or more elements of the input data. Applying the input embedding 902 may refer to passing the input data, in particular the two or more elements of the input data preferably separately, through one or more embedding layer e.g. as described within the context of FIG. 8. Applying the input embedding may comprise mapping the input data, in particular the two or more elements of the input data to a numerical representation of the input data. The numerical representation may be indicative and / or may be related to the input data. Mapping the input data to the numerical representation of the input data may comprise identifying two or more elements of the input data. For example, where the input data may be text data, the text may be divided into one or more token(s). The one or more element(s) may be mapped to a numerical representation of the one or more part(s). In particular, the number of element(s) may be equal to the number of numerical representation of the element(s). The numerical representation may be a tensor, in particular a vector and / or a matrix.

[0185] Further, the numerical representation of the two or more elements may be mapped to a numerical representation of a predefined size related to the numerical representation of the two or more elements. This may be referred to as padding. Data-driven model(s) may require data input of a predefined size. Hence, padding may allow for processing of input data of irregular size by the generative data-driven model. Padding may include concatenating anumerical representation independent of the input data with the numerical representation of the two or more elements to generate the numerical representation of predefined size related to the numerical representation of the two or more elements. The numerical representation independent of the input data may be indicative of a zero.

[0186] Further, the encoder input 924 may apply positional encoding 904. Applying positional encoding 904 may refer to adding a positional factor to the embedded input obtained via input embedding. Applying positional encoding 904 may comprise mapping the numerical representation of the predefined size related to the numerical representation of the two or more elements to a numerical representation of the two or more elements and a relation between the two or more elements. Preferably, the input data may specify a sequence of elements. The positional factor pposmay be indicative of the position of the elements within the sequence. For example, the positional factor pposmay be obtained based on the following equation:10000 d

[0189] where pos may refer to the position of the element within the sequence, / may refer to the dimension associated with the input embedding and d may refer to the dimension of the model, e.g. transformer decoder, transformer encoder or transformer encoder-decoder. This may be referred to as absolute positional embeddings. Alternatively, the positional encoding may be based on rotary positional embeddings (RoPE). Positional encoding is beneficial since it enables the processing of sequential data without requiring further dimensions indicating the position of each element. Followingly, the positional encoding 904 reduces the computational resources needed for embedding the input data. By passing the input data through the encoder input, the input data may be transformed into a second-rank tensor representing the sequence of elements. This second-rank tensor may be referred to as embedded input data. The embedded input data may be processed by the encoder block. The embedded input data may be provided to the layer normalization 908 by a residual connection. Multi-head self-attention 906 may be applied to the embedded input data. Multi-head self-attention 906 may comprise the two components multi-head and self-attention. Self-attention may be understood as being a filter applied to the embedded input data. By applying the filter to the embedded input data, the elements associated with the embedded input data contributing to the to be generated output data may be identified for generating the output data. Hence, the filter may represent the degree of contributing to the to be generated output data by the elements associated with the embedded input data. Applying the filter may be referred to as weighting the elements associated with theembedded input data. This is advantageous specifically regarding long sequences of elements. The filter may be learned and improved during the training by learning to identify the contribution of elements associated with the embedded input data. For example, in the partial sentence “I went to the bakery to buy a” the last word may be generated by the generative data-driven model such as the transformer encoder. The self-attention may focus the transformer encoder to attend to the word “bakery” and “buy” mostly to generate the word “bread”. Self-attention may refer to attention generated based on the input data. Hence, the filter may be determined based on the input data, preferably the embedded input data. The embedded input data may serve as query Q, key K and value V with respect to the self-attention operation. The self-attention may refer to attention based on the received input data. Hence, the filter may be calculated based on the following formula by inserting the respective tensors based on the embedded input data:

[0191] where dkcorresponds to the dimension of the key.

[0192] For improving the efficiency of the transformer encoder further, the multiple heads are used to apply the filter resulting in the multi-head self-attention 906. Multi-head self-attention 906 may comprise applying the filter to two or more elements of the embedded input data. Hence, the tensor may be split into two or more elements and the filter may be applied to the two or more elements separately by two or more heads according to the following equation: head i = Attention(QWiQ, KWtK, VWiV)

[0193] with parameter matrices WtQe ]Rdxd<?, wtKee Rdxdv where i may refer to the number of heads, dK, dKand dQmay refer to the dimensions of the value, key and query.

[0194] The result of the two or more head may be concatenated according to the following equation: MultiHead Q, K, V) = Concat(head l, . . . , headh)W°

[0195] heads.

[0196] The embedded input data may be transformed via the multi-head self-attention 906 into a context tensor. The context tensor may represent the sequence of elements and the relation between two or more elements of the input data. The context tensor may be a second rank tensor and / or may comprise one or more first rank tensor(s). After the multi-head self-attention 906 layer normalization 908 may be applied based on the context tensor and / or the embedded input data from the residual connection. Applying layer normalization 908 may refer to normalizing the context tensor. Normalizing the context tensor may lower the values of the entries of the context tensor. This reduces the computational cost associated with processingthe context tensor. Further, it improves the training by contributing the loss to converge and preventing instabilities.

[0197] Layer normalization 908 may be followed by passing the context tensor to a feedforward layer 910 again followed by layer normalization 912 based on the residual connection to the context tensor and / or the output of the feed-forward layer 910. The feed-forward layer 910 may be a feed-forward neural network. The feed-forward neural network may comprise of a plurality of fully connected neurons. Passing the context tensor through the feed-forward neural network may result in transforming the context tensor linearly. Additionally or alternatively, the neural network may comprise one or more activation functions such as a rectified linear unit (ReLLI). Hence, the neural network may be configured for performing one or more non-linear operations to the context tensor and / or transforming the context tensor non-linearly. After the context tensor has been transformed and / or normalized by the feed-forward layer 910 and the layer normalization 912, the context tensor may be provided to one or more further encoder blocks 914. Having passed the context tensor through the feed-forward layer 910 may adapt the context tensor for the processing by a further attention layer of the one or more further encoder blocks 914 for applying a self-attention filter, preferably multi-head self-attention 906. The context vector after being transformed by the layer normalization 912 and the feed-forward layer 910 may be referred to as hidden state.

[0198] The encoder output 922 comprises of a linear layer 916 and a softmax layer 918. The linear layer 916 may transform the context vector into a logits vector. The linear layer may be fully-connected. The logits vector obtained by passing the context tensor through the linear layer 916 may be passed through the softmax layer 918. Passing the logits vector through the softmax layer 918 may refer to applying the softmax function to the logits vector. Applying the softmax function to the logits vector may result in a probability distribution of one or more elements corresponding to the sequence of elements in the input data. From the probability distribution based on predefined selection criteria, one or more elements may be chosen. The one or more chosen elements may be referred to as the one or more elements generated by the transformer encoder. The one or more generated elements may be provided to the encoder input for generating further one or more elements corresponding to the sequence of the input data and the one or more elements generated by the transformer encoder as described within the context of FIG. 11.

[0199] Hence, processing the numerical representation of the two or more elements and the relation between the two or more elements by the generative data-driven model may comprise at least one of• generating two or more numerical representations of the two or more elements and the relation between the two or more elements from the numerical representation of the two or more elements and the relation between the two or more elements,• modifying the two or more numerical representation of the two or more elements and the relation between the two or more elements by applying a filter to the two or more numerical representations of the two or more elements and the relation between the two or more elements, wherein the filter may be configured to modify the contribution of the two or more elements to the numerical representations of the two or more elements and the relation between the two or more elements,• concatenating the two or more numerical representations of the two or more elements and the relation between the two or more elements• mapping the concatenated numerical representation of the two or more elements and the relation between the two or more elements to a numerical representation of the output data

[0200] or a combination thereof.

[0201] In particular the encoder block may be configured to• split the numerical representation of the two or more elements and the relation between the two or more elements into two or more numerical representations of the two or more elements and the relation between the two or more elements,• modify the two or more numerical representation of the two or more elements and the relation between the two or more elements by applying a filter to the two or more numerical representations of the two or more elements and the relation between the two or more elements, wherein the filter may be configured to modify the contribution of the two or more elements to the numerical representations of the two or more elements and the relation between the two or more elements,• concatenate the two or more numerical representations of the two or more elements and the relation between the two or more elements

[0202] or a combination thereof. Applying self-attention may comprise modifying the two or more numerical representation of the two or more elements and the relation between the two or more elements by applying a filter to the two or more numerical representations of the two or more elements and the relation between the two or more elements , wherein the filter may be configured to modify the contribution of the two or more elements to the numerical representations of the two or more elements and the relation between the two or more elements. The filter may be obtained during training of the generative data-driven model. The filter may be obtained based on, in particular related to the input data. Multi-head self-attention may comprisegenerating two or more numerical representations of the two or more elements and the relation between the two or more elements from the numerical representation of the two or more elements and the relation between the two or more elements, modifying the two or more numerical representation of the two or more elements and the relation between the two or more elements by applying a filter to the two or more numerical representations of the two or more elements and the relation between the two or more elements , wherein the filter may be configured to modify the contribution of the two or more elements to the numerical representations of the two or more elements and the relation between the two or more elements and / or concatenating the two or more numerical representations of the two or more elements and the relation between the two or more elements.

[0203] The encoder output may be configured to map the concatenated numerical representation of the two or more elements and the relation between the two or more elements to a numerical representation of the output data. The numerical representation of the output data may be mapped to output data, eg by providing a vocabulary indicative of a relation between numerical representations and data of a type according to the input data. Additionally or alternatively, a decoding model may be used to map the concatenated numerical representation of the two or more elements and the relation between the two or more elements to a numerical representation of the output data. The decoding model may be trained to relate a numerical representation of data of a type according to the input data.

[0204] FIG. 7 illustrates an embodiment of a transformer decoder architecture e.g. of an decoder-only transformer or transformer-based model.

[0205] The transformer decoder comprises a decoder input 1024, one or more decoder blocks 1020 , 1014 and a decoder output 1022. The transformer decoder may be referred to as X- former. The transformer decoder architecture may correspond to the decoder architecture associated with the transformer encoder-decoder architecture independent of receiving one or more hidden states from the encoder of the transformer encoder-decoder. An example of transformer decoder architectures is the generative pretrained transformer (GPT).

[0206] The decoder input 1024 may apply input embedding 1002 and positional encoding 1004 analogous to analogous to the input embedding 1002 and the positional encoding 1004 as described within the context of FIG. 9.

[0207] The decoder block 1020 may comprise the layer normalizations 1008, the masked multihead self-attention 1006, the feed-forward layers 1010 and / or the layer normalization 1012. The embedded input data resulting from passing the input data through the decoder input 1024 may be provided to the layer normalization 1008 via a residual connection. Further, masked multi-head self-attention 1006 may be applied to the embedded input data. Masked multi-head selfattention 1006 corresponds to the multi-head self-attention 906 as described within the context of FIG. 9 with additionally masking a part of the embedded input data associated with elements later in the sequence than the element to be generated. Additionally or alternatively, the part of the input data associated with elements later in the sequence than the element to be generated may not be received and / or transformed into the embedded input data. Thus, the transformer decoder may be suitable for generating a subsequent element to a sequence, whereas the transformer encoder may be suitable for generating a missing element in within one sequence and / or between two or more sequences. Therefore, the transformer encoder may be configured for classification tasks. The transformer decoder may be configured for text generation. Masked multi-head self-attention may comprise applying a filter obtained based on elements of the sequence of the input data appearing previously to the to be generated part of the sequence. Similar to the transformer encoder as described within the context of FIG. 9, a context tensor may be generated by applying the masked multi-head self-attention 1006 and the layer normalization 1008. The context tensor may be provided to the layer normalization 1012 via a residual connection. Further, the feed-forward layer 1010 and the layer normalization 1012 may be analogous to the feed-forward layer 910 and the layer normalization 912 as described within the context of FIG. 9. The context tensor may be provided to one or more further decoder blocks 1014.

[0208] The decoder output 1022 may comprise of a linear layer 1016 and a softmax layer 1018. The linear layer 1016 and the softmax layer 1018 may be analogous to the linear layer 916 and the softmax layer 918 as described within the context of FIG. 9.

[0209] FIG. 11 illustrates an embodiment of a transformer encoder-decoder architecture e.g. of an encoder-decoder transformer(-based) model. The transformer encoder-decoder may comprise the encoder input 1140, the one or more encoder blocks 1138, 1128, the decoder input 1146, the decoder block 1142 and the decoder output 1144. The encoder input 1140 may correspond to the encoder input 924 of FIG. 9. The one or more encoder block 1138, 1128 may correspond to the one or more encoder blocks 920, 914 of FIG. 6. The decoder input 1146 may correspond to the decoder input 1024 of FIG. 10.

[0210] The architecture described with respect to FIG. 11 may allow that the transformer encoder-decoder may receive and process input data at the encoder input 1140 and the one or more encoder blocks 1138, 1128 and the decoder block 1142 and the decoder output 1144. Based on the input data, the transformer encoder-decoder may generate output data part by part or sequentially. The sequentially generated output data may be provided to and / or may beprocessed by the decoder input 1146, the one or more decoder blocks 1142, 1106 and the decoder output 1144. Preferably, a sequence may be provided to the encoder input 1140 and after having generated at least a part of the output data, the decoder input 1146 may be provided with at least the part of the elements of the output data already generated. By doing so, the next elements of the output data may be generated with a higher accuracy by taking the input data and the generated output data into account since more data is received by the transformer encoder-decoder may be received over time.

[0211] Because of the transformer encoder-decoder architecture, the transformer encoderdecoder may be configured for transforming a sequence into another representation of the sequence. An example for transforming one sequence into another representation may be translation of one sentence into another language. A plurality of transformer encoder-decoders may be used such as BART, T5 or the like.

[0212] In an embodiment, the layer normalization 1136, 1112 may be applied prior to the masked multi-head self-attention 1134, multi-head self-attention 1114 and / or the feed-forward layer 1102 in the transformer decoder, the transformer encoder and / or the transformer encoderdecoder. By doing so, the computational resources for applying the multi-head self-attention 1114 and / or the feed-forward layer 1102 to the embedded input data and / or the context tensor may be decreased as the entries of the respective tensors may be lower after normalization.

[0213] In an embodiment, the decoder output 1144 may comprise of a classification neural network, further feedforward layers, convolutional layers, fully connected layers or the like. For example, the transformer encoder-decoder may be configured for choosing between a plurality of options. For this purpose, the transformer encoder-decoder may be provided with three different input data sets and may classify the context vectors obtained from the one or more decoder blocks 1142 via one or more linear layers. Followingly, the architecture may be extended depending on the use case to be solved.

[0214] FIG. 12 illustrates an embodiment of training and / or deploying the transformer encoder, the transformer decoder and / or the transformer encoder-decoder.

[0215] The encoder / decoder / encoder-decoder architecture 1202 may correspond to the transformer decoder, the transformer encoder and / or the transformer encoder-decoder as described within the context of FIG. 9 - FIG. 11.

[0216] The output data generated by the encoder / decoder / encoder-decoder architecture 1202 may comprise of one or more elements, in particular a sequence of elements. The previously generated elements of the output data may be provided as input for generating the next element in the sequence of the output data.

[0217] The input data may comprise of N elements, in particular input tokens. An input token may be a token dedicated to be inputted into a data-driven model such as the transformer decoder, the transformer encoder or the transformer encoder-decoder. The output data to be generated may comprise of M elements. The encoder / decoder / encoder-decoder architecture 1202 may generate one element of the output data based on receiving the input data and optionally previously generated elements of the output data at a timestep. Hence, for generating M elements M time steps are required. A time step comprises of providing input 1210, 1212, 1214 to the encoder / decoder / encoder-decoder architecture 1202 and receiving output data 1204, 1208, 1206 from the encoder / decoder / encoder-decoder architecture 1202. In a first timestep, the input 1210 may comprise of N input tokens. The N input tokens may be associated eg with N words, stems or endings. Preferably, the N input tokens may specify a question. One or more input tokens may specify the beginning of the sequence of tokens and / or the end of the sequence of tokens. The input 1210 may be processed by the encoder / decoder / encoder- decoder architecture 1202. Based on the input 1210 at least a part of the output data 1204 may be generated. The at least a part of the output data may comprise a first output token. In the next timestep, the generated first output token may be provided together with the input 1212. Specifically, where the input 1212 may be received by a transformer encoder-decoder the input tokens may be received at the encoder input 1140 and the first output token may be received at the decoder input 1146. Where the input 1212 may be received by the transformer encoder, the input 1212 may be received by the encoder input 1140 and analogously regarding the transformer decoder and the decoder input 1146. Based on the input 1212, the output data 1208 comprising the first output token and a second output token may be generated. Generating the output data 1208 based on the input 1212 may refer to generating the second token based on the first token and the N input tokens, wherein the first token may have been generated based on the N input tokens. This process may be repeated until the last token in the sequence of the output data 1206 may be generated. Preferably, the last token may be an end token. The end token may terminate the generation of a further output token.

[0218] Similarly, to the data processing during deployment of the encoder / decoder / encoder- decoder architecture 1202, the encoder / decoder / encoder-decoder architecture 1202 may be trained. The training data set may comprise a plurality of sequences comprising a plurality of elements. The sequences may be associated with the input data and / or the output data. Additionally or alternatively, the sequences may be independent of the input data and / or the output data. For example, where the input data and the output data may refer to chemical compositions represented via text, the training data set may comprise sequential text data independent of chemical compositions. In this example, the training data set may comprisesequences of words originating from a conversation. In an embodiment, the training data set may comprise at least partially input data sets and / or output data sets.

[0219] The training may be initialized by initializing the encoder / decoder / encoder-decoder architecture 1202. In an embodiment, the parameters associated with the encoder / decoder / encoder-decoder architecture 1202 may be initialized randomly. Additionally or alternatively, the input embedding of the encoder / decoder / encoder-decoder architecture 1202 may be obtained by training a CBOW model or a skip gram model as described within the context of FIG. 8. The trained embedding layer may be used during training. The parameters associated with the embedding layer may be kept constant and / or may be updated after a predefined number of training epochs. By doing so, the number of parameters to be updated is lower enabling a faster and less computational resources-consuming training. Further, the accuracy associated with the embedding layer may be constant and / or may be increased by avoiding error compensation in relation to the just initialized encoder / decoder / encoder-decoder architecture 1202.

[0220] During the training of the encoder / decoder / encoder-decoder architecture 1202, at least a part of the sequences of the training data set may be provided to the encoder / decoder / encoder-decoder architecture 1202 one by another and one or more elements may be generated based on the sequences of the training data set one by another. The elements generated based on the sequences may follow the elements of the parts of sequences the encoder / decoder / encoder-decoder architecture 1202 may have been provided with. The generated one or more elements may be compared to the one or more elements following the at least a part of the sequences provided to the encoder / decoder / encoder-decoder architecture 1202 as specified by the training data set. Hence, during the training the encoder / decoder / encoder-decoder architecture 1202 may generate a guess on the next element and the guess on the next element in a sequence may be compared to the ground truth specifying the actual next element according to the training data set. Based on the guess on the next element and the ground truth a loss may be determined. The loss may define the similarity between the guess on the next element and the ground truth. The loss may be determined by forming a vector dot product between the token associated with the one or more elements and the token associated with the ground truth. A loss unequal to zero may result in updating the parameters associated with encoder / decoder / encoder-decoder architecture 1202. Preferably the parameters associated with the encoder / decoder / encoder-decoder architecture 1202 may be independent of the embedding layer. For example, the parameters associated with theencoder / decoder / encoder-decoder architecture 1202 may be weights of the neurons of the encoder / decoder / encoder-decoder architecture 1202.

[0221] Based on the determined loss, backpropagation may be applied to determine the gradients associated with the parameters of the parameters associated with encoder / decoder / encoder-decoder architecture 1202 to lower the loss. According to the determined gradients, the parameters associated with the encoder / decoder / encoder-decoder architecture 1202, preferably the weights of the neurons associated with the encoder / decoder / encoder-decoder architecture 1202, may be updated by using a gradient descent algorithm.

[0222] The training data set may be unlabeled. The sequences of elements within the training data set may inherently comprise the ground truth for determining the loss with respect to the one or more elements generated during the training of the encoder / decoder / encoder-decoder architecture 1202. Hence, the encoder / decoder / encoder-decoder architecture 1202 may be trained self-supervised. This is advantageous since time and resources for creating a labeled training data set may be saved. Furthermore, this enables the usage of large training data sets associated with a size of several tera bytes. Consequently, the data-driven model may be accurate in generating elements of a sequence. In addition, the large training data set enables few shot predictions or even zero shot predictions. Hence, the generative data-driven model(s) trained as described above are versatile contributing to saving resources needed for training and / or hosting a plurality of purpose-driven models such as convolutional neural networks. The training described above may be referred to as pretraining. Pretraining may refer to training a generative data-driven model based on data with a plurality of contexts

[0223] The generative data-driven model may be configured for performing few shot or even zero shot predictions with respect to a plurality of use cases after pretraining. The performance of the data-driven model may be increased further by additional training referred to as finetuning. Finetuning may refer to training a pretrained data-driven model for a concrete task, e.g. by providing task instructions to the pretrained data-driven model and adapting the parameters of the pretrained data-driven model to decrease the distance of the generated output data by the pretrained data-driven model in response to receiving the task instructions from predefined output data corresponding to the provided task instructions.

[0224] Models based on the architecture according to FIG. 9 to FIG. 11 and / or pretrained generative data-driven model(s) and / or finetuned data-driven model(s) may be referred to as large language models. Examples of large language models include Llama models, Mistral models, GPT models, BERT models or the like. Such models have been tested. Testing data-driven model(s), in particular pretrained and / or finetuned data-driven model(s), may include comparing output data generated by the one or more data-driven model(s) in response to receiving the input data with target data, e.g. obtained from domain experts. These domain experts may be a current bar for performing tasks the data-driven model(s) may be parametrized and / or trained for. The target data may specify output data desired to be generated in response to receiving the input data.

[0225] FIG. 13 illustrates an embodiment of a Mamba architecture that may be used as a generative data-driven model. A Mamba architecture may enhance inference speed in relation to a transformer based model.

[0226] The Mamba architecture with its layered structure may be similar to the transformer decoder architecture discussed in relation to FIG. 10. However, instead of decoder blocks mamba blocks 1332, 1304 are stacked. Mamba block 1332 may be based on a selective space state sequence model (S6).

[0227] An input token may be linearly projected via linear layer 1312, 1320 into an expanded latent space (which may allow to capture more information during processing in the selective state space layer 1310), followed by a convolution via a convolutional layer 1314 and a nonlinear function (e.g. a sigmoid linear unit (SiLu) or swish activation function). The convolution before the selective state space layer 1310 may prevent independent token calculations. The selective state space layer 1310 performs a selective state space operation. Further, a learnable skip connection may be provided via linear layer 1320, this may use a linear transformation to map the input to the output, similar to a residual connection in a transformer model this may help to mitigate vanishing gradient effects.

[0228] A selective state space layer 1310 may be a linear recurrent network that selectively process data based on the input token, which may allow to focus on relevant data and discard irrelevant data. For instance in each step a separate weight vector may be determined based on the respective input token. The determined weight vector may then be used in a selective scan.

[0229] A selective state space layer 1310 may be used in a convolutional mode e.g. for parallelizable training and a recurrent mode for near-constant time generation of output data. A state space operation may be based on solving the state and output equations, wherein a state equation may describe how a state changes based on how the input influences the state and an output equation may describe how the state is translated to the output. Further how the input influences the output may be represented by a learnable linear transformation, e.g. a matrix D, used in a learnable skip connection.

[0230] The state equation for a hidden state may be (in discretized form):

[0231] hk= Ahk-1+ Bxk

[0232] The output may be expressed by (in discretized form):

[0233] yk= Chk

[0234] This discretized space state model may be unfolded into a recurrent form similar to a recurrent network, exemplifying that a selective state space model may be or comprise a linear recurrent model. However, here matrices A, B, and C may also be used as a kernel of a convolution of the state space model. Kernel K for this may e.g. be:

[0235] K = (CA2B, CAB, CB)

[0236] which may allow to determine an output:

[0237] yk+1= CA2Bxk-±+ CABxk+ CBxk+1

[0238] So, in this representation of the state space model training may be performed in a parallel manner like in convolutional neural networks.

[0239] Matrix A may be a matrix that represents recent tokens well and decays older tokens and may be initialized using HiPPO:

[0241] where every entry below the diagonal is set to 0. This may allow to create a long-term memory for the selective state space model.

[0242] For a Mamba block 1332, the matrices B and C as well as the step size A used for discretization of the matrices may be dependent on the input token and may be trained during training, so that for each input token different matrices B and C are determined, which may enhance the content-awareness and may act similar to a multi-head self-attention in a transformer model. However, unlike in space state models with fixed matrices A, B, and C, here the convolutional representation may not be easily determined. Hence, to operate the selective state space layer 1310 in convolutional mode a selective scan may be applied utilizing associative properties of the hidden states calculation, allowing parallel determination of the sequence in parts and iteratively combining them, so that parallel training may be used. Further reading and writing operations may be decreased by using kernel fusion of the described step size, the selective scan, and the multiplication with C.

[0243] Linear layer 1302 may project the generated output back into the same dimension as the input.

[0244] Mamba blocks may be used together with transformer decoder blocks or mixture of expert blocks (e.g. decoder blocks wherein the feed-forward layer is exchanged for a gating network and a number of parallel feed-forward layers, wherein the gating network switches between the feed-forward layers depending on the input), which may allow leveraging advantages of the different architectures.

[0245] An example of the architecture of a mamba block may be found in “Mamba: Linear- Time Sequence Modeling with Selective State Spaces” by Albert Gu and Tri Dao arXiv:2312.00752v2 [cs.LG] 31 May 2024, , which is incorporated herein by reference.

[0246] FIG. 14 shows a schematic block diagram of an example apparatus 1414 or computation apparatus according to an example aspect, which may for instance represent the apparatus according to the fourth example aspect. Apparatus 1414 may for instance be configured to perform and / or control or comprise respective means (e.g. at least one of memory 1404, processor 1402, communication interface 1406, user interface 1408) for performing and / or controlling the method according to any example aspect. Apparatus 1414 may as well constitute an apparatus comprising at least one processor (1402) and at least one memory (1404) storing instructions that, when executed by the at least one processor, cause an apparatus, e.g. apparatus 1414 at least to perform and / or control the method according to all example aspects. Processor 1402 may for instance execute program code stored in memory 1404, which may for instance represent a readable storage medium comprising program code that, when executed by processor 1402, causes the processor 1402 to perform the method according an example aspect. Processor 1402 may for instance further control memory 1404 and / or further memories, the communication interface(s) 1004, the optional user interface 1408, e.g. a graphical user interface 1408. Processor 1402 (and also any other processor mentioned in this specification) may be a processor of any suitable type. Memory 1404 may be included in processor 1402 or memory 1404 may be fixedly connected to processor 1402, or be at least partially removable from processor 1402, for instance in the form of a memory card or stick. Memory 1404 may for instance be non-volatile memory. It may for instance be a FLASH memory (or a part thereof), any of a ROM, PROM, EPROM and EEPROM memory (or a part thereof) or a hard disc (or a part thereof). Memory 1404 may also comprise an operating system for processor 1402. Memory 1404 may also comprise a firmware for apparatus 1414. Memory 1404 may also for instance be a Random Access Memory (RAM) or Dynamic RAM (DRAM). It may for instance be used by processor 1402 when executing an operating system and / or computer program. Communication interface (s) 1406 may enable apparatus 1414 to communicate with other entities, e.g. another apparatus, such as a decentral network node or a server providing a generative data-drivenmodel or a user device or computer e.g. used by an operator and providing a user interface. The communication interface(s) 1004 may for instance comprise a wireless interface and / or wirebound interface for instance to communicate with entities via an Intranet. User interface 1408 is optional and may comprise a display for displaying information to a user and / or an input device (e.g. a keyboard, keypad, touchpad, mouse, etc.) for receiving e.g. a query from an operator. Some or all of the components of the apparatus 1414 may for instance be connected via a bus. Some or all of the components of the apparatus 1414 may for instance be combined into one or more modules.

[0247] The publication Prior Art Disclosure; Issue 684; paragraphs

[1000] to

[8005] ; ISSN: 2198-4786; published: February 12, 2024 will be regarded as Reference RF1 , which is incorporated herein by reference in its entirety. Preferably, the product is a chemical product as described in Reference RF1 ; paragraphs

[1000] to

[8005] , Preferably, the method / process described herein is further a method / process for the production of a product.

[0248] The converting step to obtain the product preferably comprises one or more step(s) as described below and can be performed by conventional methods well known to a person skilled in the art. The converting step preferably comprises one or more step(s) selected from:• recycling, preferably depolymerizing, gasifying, pyrolyzing, and / or steam cracking; and / or• purifying, preferably crystallizing, (e.g. solvent) extracting, distilling, evaporating, hydrotreating, absorbing, adsorbing and / or subjecting to ion exchanger; and / or• assembling, preferably foaming, synthesizing, chemical conversion, chemically transforming, polymerizing and / or compounding; and / or• forming, preferably foaming, extruding and / or molding; and / or• finishing, preferably coating and / or smoothing.

[0249] In addition, the one or more step(s) are described in detail in Reference RF1 ; paragraphs

[1000] to

[8005] ,

[0250] The present disclosure has been described in conjunction with preferred embodiments and examples as well. However, other variations can be understood and effected by those persons skilled in the art and practicing the claimed subject-matter, from the studies of the drawings, this disclosure and the claims. Notably, in particular, any steps presented can be performed in any order, i.e. the present disclosure is not limited to a specific order of these steps. Moreover, it is also not required that the different steps are performed at a certain place or at one node of a distributed system, i.e. each of the steps may be performed at different nodes using different equipment / data processing.

[0251] The sequence of all method steps presented above is not mandatory, also alternative sequences may be possible. Nevertheless, the specific sequence of method steps shown as examples in the figures shall be considered as one possible sequence of method steps, e.g. for the respective embodiment described by the respective figure or an embodiment comprising at least some of the steps described by the respective figure.

[0252] In the present specification, any presented connection in the described embodiments is to be understood in a way that the involved components are operationally coupled. Thus, the connections can be direct or indirect with any number or combination of intervening elements, and there may be merely a functional relationship between the components.

[0253] As used herein ..determining" may also include ..initiating or causing to determine", “generating" may also include ..initiating and / or causing to generate", “providing” may also include “initiating or causing to determine, generate, select, send and / or transmit”, and "obtaining" may also include “initiating or causing to determine, generate, select, retrieve and / or receive”. “Initiating or causing to perform an action” may include any processing signal that triggers a computing node or device to perform the respective action.

[0254] The term “comprising” or “including” is to be understood in an open sense, i.e. in a way that an object that “comprises an element A” may also comprise further elements in addition to element A. Further, the term “comprising” or “including” may be limited to “consisting of”, i.e. consisting of only the specified elements.

[0255] The indefinite article “a” or “an” is not to be understood as “one”, i.e. use of the expression “an element” does not preclude that also further elements are present. A single element or other unit may fulfill the functions of several entities or items recited in the claims. The mere fact that certain measures are recited in the mutual different dependent claims does not indicate that a combination of these measures cannot be used in an advantageous implementation or further elements may be included.

[0256] The expressions “A and / or B” and “at least one of: A or B” are considered interchangeable and meant to comprise any one of the following three scenarios: (i) A, (ii) B, (iii) A and B. More generally, the expression “at least one of the following: ” and “at least one of <a list of two or more elements:*” and similar wording, where the list of two or more elements are joined by “and” or “or”, mean at least any one of the elements, or at least any two or more of the elements, or at least all the elements, so the wording includes any combinations of the elements.

[0257] Providing in the scope of this disclosure may include any interface configured to provide data. This may include an application programming interface, a human-machine interface suchas a display and / or a software module interface. Providing may include communication of data or submission of data to the interface, in particular display to a user or use of the data by the receiving entity.

[0258] Obtaining in the scope of this disclosure may include any interface configured to obtain or receive data. This may include an application programming interface, a human-machine interface such as a display and / or a software module interface. Obtaining may include communication of data or submission of data from the interface, in particular use of the data by the receiving entity. Any obtaining of data, data structures, data sets, or the like may comprise receiving the data, data structures, data sets, or the like from a server providing (e.g. hosting) a data base comprising the data, data structures, data sets, or the like.

[0259] Various units, circuits, entities, nodes or other computing components may be described as “configured to” perform a task or tasks. Configured to shall recite structure meaning “having circuitry that” performs the task or tasks on operation. The units, circuits, entities, nodes or other computing components can be configured to perform the task even when the unit / circuit / component is not operating. The units, circuits, entities, nodes or other computing components that form the structure corresponding to “configured to” may include hardware circuits and / or memory storing program instructions executable to implement the operation. The units, circuits, entities, nodes or other computing components may be described as performing a task or tasks, for convenience in the description. Such descriptions shall be interpreted as including the phrase “configured to.” Any recitation of “configured to” is expressly intended not to invoke 35 U.S.C. § 112(f) interpretation.

[0260] In general, the methods, apparatuses, systems, computer elements, nodes or other computing components described herein may include memory, software components and hardware components. The memory can include volatile memory such as static or dynamic random-access memory and / or nonvolatile memory such as optical or magnetic disk storage, flash memory, programmable read-only memories, etc. The hardware components may include any combination of combinatorial logic circuitry, clocked storage devices such as flops, registers, latches, etc., finite state machines, memory such as static random-access memory or embedded dynamic random-access memory, custom designed circuitry, programmable logic arrays, etc.

[0261] In the present specification, any presented connection in the described embodiments is to be understood in a way that the involved components are operationally coupled. Thus, the connections can be direct or indirect with any number or combination of intervening elements, and there may be merely a functional relationship between the components.

[0262] Moreover, any of the methods, method steps, processes and actions described or illustrated herein may be implemented using executable instructions in a general-purpose or special-purpose processor and stored on a computer-readable storage medium (e.g., disk, memory, or the like) to be executed by such a processor. References to a ‘computer-readable storage medium’ should be understood to encompass specialized circuits such as signal processing devices, and other devices.

[0263] A processor may be a processor of any suitable type, and is preferably a processor configured for parallel processing of at least a hundred or at least a thousand threads in parallel, e.g. a graphical processing unit (GPU). For instance, the processor comprises at least a hundred or a at least a thousand parallel processing cores. In particular, the processor may comprise at least one (preferably at least a thousand) compute unified device architecture (CUDA) core(s), which may allow for using a graphical processing unit as the processor, which may increase computational efficiency. For instance, the processor may comprise at least one (e.g. at least a hundred) streaming multiprocessor cores, which may allow for increasing the data throughput. As a further example, the processor may comprise one or more (e.g. at least a hundred) tensor core(s) and / or (e.g. at least a hundred) tensor processing units (TPUs) . A tensor core may be specifically adapted to perform matrix operations and may allow to accelerate large matrix operations. A tensor core may be configured to perform mixed-precision matrix multiply and accumulate calculations in a single operation. For instance, a tensor core may perform mixed- precision floating-point matrix arithmetic, specifically utilizing FP16 (half-precision) inputs to produce either full-precision (FP32) or half-precision (FP16) outputs. In the case of FP16 output, a tensor core may provide a performance boost by storing the intermediate accumulation results in FP32 format, thereby maintaining the precision necessary for accurate results. A tensor processing unit may be an application-specific integrated circuit (ASIC). It may comprise a matrix multiplication unit (MXU), which may be specifically adapted or configured for dense linear algebra operations. TPUs may be configured to handle large-scale matrix operations efficiently, which may provide high computational throughput for Al tasks. A TPU may be equipped with on- chip high-bandwidth memory (HBM), which may enhance the capability for the use of larger models and batch sizes. TPUs may be connected in groups called Pods, which may scale up workloads with minimal code changes. An MXU may be specifically configured for performing matrix multiplications. A TPU may comprise a tensor core.

[0264] For example, a processor may comprise several thousand tensor cores, each capable of performing 64 floating point FMA (Fused Multiply-Add) operations per clock cycle or (e.g. at least several hundred) tensor processing units (TPUs) being specifically configured foraccelerating machine learning (ML) workloads, particularly for cloud-based applications. Additionally, Field-Programmable Gate Arrays (FPGAs) and Application-Specific Integrated Circuits (ASICs) may provide flexibility and performance benefits for specific Al tasks.. With these capabilities, such a GPU may allow for hundreds of TFLOPs (Tera Floating-Point Operations per Second) of performance in mixed-precision computations. Furthermore, a tensor core may support a variety of numerical formats, including IEEE standard half-precision, singleprecision, and double-precision floating-point formats, as well as a range of integer formats.

[0265] A processor may be a central processing units (CPU) configured with an advanced architecture, such as Intel’s Xeon Scalable processors or AMD’s EPYC series. A CPU may be configured for sequential processing and general-purpose computing. These CPUs may incorporate vector instruction sets, such as AVX-812, to accelerate mathematical computations that may e.g. enhance Al model training and inference. Furthermore, CPUs may integrate Al accelerators i.e. a CPU may be specifically configured for deep learning workloads.

[0266] The processor may be coupled to memory having a memory bandwidth of at least a hundred gigabytes per second, which may allow efficient handling of extensive data sets and may allow faster reading, processing, and writing compared to a general-purpose processor such as a computational processing unit.

[0267] The memory may be a high-capacity memory configured to manage the data-intensive nature of Al applications, providing necessary bandwidth and storage capacity for complex datasets. The memory may for instance be DDR4, DDR5, High Bandwidth Memory (HBM) and / or GDDR6X memory, which may improve data transfer rates and reduce latency. Such memory may enhance e.g. modeling and real-time sensor data for monitoring and control. Further, the memory may be operated with memory optimization techniques, such as caching and prefetching, which may enhance the execution speed of Al algorithms. Non-volatile Memory (NVM) technologies, including NAND Flash and 3D XPoint, may provide persistent storage solutions with high-speed access, which may enhance rapid data storage and retrieval for Al applications.

[0268] Any disclosure and embodiments described herein relate to the methods, the systems, devices, the computer program element lined out above and vice versa. Advantageously, the benefits provided by any of the embodiments and examples equally apply to all other embodiments and examples and vice versa.

[0269] All terms and definitions used herein are understood broadly and have their general meaning if not indicated otherwise.

[0270] It will be understood that all presented embodiments are only examples, and that any feature presented for a particular example embodiment may be used with any aspect on its own or in combination with any feature presented for the same or another particular example embodiment and / or in combination with any other feature not mentioned. In particular, the example embodiments presented in this specification shall also be understood to be disclosed in all possible combinations with each other, as far as it is technically reasonable and the example embodiments are not alternatives with respect to each other. It will further be understood that any feature presented for an example embodiment in a particular category (method / apparatus / computer program / system) may also be used in a corresponding manner in an example embodiment of any other category. It should also be understood that presence of a feature in the presented example embodiments shall not necessarily mean that this feature forms an essential feature and cannot be omitted or substituted.

Claims

75CLAIMS1. A method for accessing at least one decentral network to provide and / or obtain a product data set associated with an output product, wherein the output product is produced or producible from one or more input materials by a production environment, the method comprising the steps:- obtaining, from an operator of the production environment, a user instruction including natural language and at least data indicative of the production environment, the output product, and at least one other production environment, the at least one other production environment configured to produce a product based on the output product and / or the at least one other production environment providing input material for producing the output product;- determining, based on the user instruction, a tool sequence, wherein a tool sequence comprises an indication of one or more tool(s), wherein the one or more tools comprises at least one agent(s) configured to generate at least one data provision instruction for accessing the at least one decentral network, wherein determining the tool sequence comprises providing a task instruction, based on the user instruction, to at least one generative data-driven model, the at least one generative data-driven model configured to generate the tool sequence in response to obtaining the task instruction;- carrying out or causing to carry out the one or more tool(s) comprising the at least one agent(s) according to the tool sequence;- receiving output data from the one or more one or more tool(s), the output data comprising at least a part of the at least one data provision instruction for accessing the at least one decentral network;- providing the at least one data provision instruction;- receiving, based on the at least one data provision instruction, an operation data set related to the at least one data provision instruction;- determining, based on receiving the operation data set, at least one of an identification associated with the production environment or security credential(s) for accessing the at least one decentral network for obtaining and / or providing the product data set; and- providing, to the operator, at least one of the identification associated with the production environment or the security credential(s).

2. The method of claim 1 , wherein determining the tool sequence comprises:76 retrieving a set of tool data sets, wherein each tool data set of the set of tool data sets is associated with a tool of the one or more tool(s); wherein the determining the tool sequence is further based on the set of tool data sets.

3. The method of claim 1 or 2, wherein determining the tool sequence further comprises at least the following steps: providing a task instruction for generating the tool sequence to a generative data-driven model, wherein the task instruction is generated based on the user instruction and the set of tool data sets, the generative data-driven model configured to generate a tool generated data set related to the tool sequence, in response to receiving the task instruction; providing the tool generated data set as the tool sequence or in case the tool generated data set is not of the same format of type of a tool sequence: parsing or causing to parse the tool generated data set into the tool sequence and providing the tool sequence.

4. The method of any one of claims 1 to 3, wherein determining the tool sequence comprises: determining whether input data required for carrying out the at least one tool is provided by the user instruction; upon determining that the input data required for carrying out the at least one tool is provided by the user instruction: including the at least one tool in the tool sequence to be carried out based on the input data.

5. The method of claim 1 , wherein the agent configured to generate at least one data provision instruction carries out at least the following steps:- obtaining the user instruction and optionally routing information;- obtaining one or more instruction templates relating to accessing at least one decentral network;- selecting at least one instruction template of the one or more instruction template(s) based on the user instruction;- generating one or more instruction(s) including at least one task instruction related to accessing at least one decentral network, based on the at least one instruction template, the user instruction and optionally the routing information;- generating at least one data provision instruction for accessing the at least one decentral network by providing the one or more generated instruction(s) to a generative data-driven model, wherein the generative data-driven model has been trained on unstructured natural language data and is configured to process natural language.

776. The method of claim 5, wherein depending on the user instruction multiple instruction templates may be selected and processed sequentially, wherein the instruction templates relate to identification of at least one decentral network, setting up at least one edge data connector, preparing and testing end-to-end integration, deploying access infrastructure to the production environment, wherein optionally generating instructions includes providing at least part of the data provision instruction generated by the generative data-driven model from one or more preceding instruction(s) to a subsequent instruction template.

7. The method of any one of claims 1 to 6, the method further comprising: obtaining routing information comprises data for identifying the at least one decentral network among a set of decentral networks associated with different products and / or other production environments, based on the data indicative of the production environment, the output product, and at least one other production environment; wherein determining the a tool sequence is further based on the routing information.

8. The method of claim 7, wherein the routing information is obtained from a knowledge database, in particular a knowledge database comprising data sets for decentral networks in a set of decentral networks comprising the at least one decentral network, wherein the data sets are descriptive of a respective decentral network.

9. The method of any one of claims 1 to 8, wherein the at least one agent(s) configured to generate at least one data provision instruction for accessing the decentral network provides the one or more instruction(s) to a generative data-driven model, the generative data-driven model having been trained to generate the at least one data provision instruction in response to obtaining the one or more instruction(s).

10. The method of claim 9, wherein the at least one data provision instruction is generated based on similarity between the user instruction and pre-defined data provision information, wherein the at least one data provision instruction is generated based on a similarity level for the similarity between the user instruction and the pre-defined data provision information.

11. The method of claim 10, wherein, if no pre-defined data provision information is determined to fulfill the similarity level, the at least one data provision instruction is generated based on interaction with a generative data-driven model, wherein the generative data-driven model has been trained on unstructured natural language data and is configured to process natural language.7812. The method of claim 10 or 11 , wherein, if the pre-defined data provision information is determined to fulfill the similarity level, the at least one data provision instruction is generated based on a similarity score for the respective pre-defined data provision information.

13. A method for generating a digital product passport associated with a product, the method comprising: accessing a decentral network according to a method of any one of claims 1 or 5-12; providing product data associated with the product, providing a decentral identifier associated with the product data; generating the digital product passport including the decentral identifier and at least part of the product data; providing the digital product passport for access by a decentral network node associated with a product consumer under control or controlled by a decentral network node associated with a data owner of the product passport.

14. An apparatus comprising respective means for carrying out or performing the steps of any one of claims 1 or 5-13 or comprising at least one processor and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus at least to carry out the steps of the method according to any one of claims 1 or 5-13.

Images