Monitoring and / or control of chemical plants

Abstract

A method for monitoring and / or controlling the production of at least one chemical product, To provide chemical product data related to the production and / or processing of chemical products, To provide multiple chemical product data templates that show structures associated with at least some of the chemical product data, Selecting at least one chemical product data template by determining that at least one chemical product data template corresponds to at least a portion of the chemical product data, To generate contextualized chemical product data by merging at least a portion of chemical product data and at least one selected chemical product data template, Providing contextualized chemical product data to a data-driven model for generating environmental characteristics data relating to the processing and / or production of chemical products, wherein the data-driven model is configured to provide environmental characteristics data in response to the contextualized chemical product data provided by it. A method comprising providing environmental characteristics data for monitoring and / or controlling the processing and / or production of at least one chemical product.

Description

[Technical Field]

[0001] Technical field The claimed invention relates to climate change mitigation through advanced manufacturing in an industrial environment of chemical production, the use of environmental characteristics data associated with the processing and / or production of one or more chemical products, an apparatus for monitoring and / or controlling the production of chemical products, and a method for monitoring and / or controlling the production of at least one chemical product. [Background technology]

[0002] Technical background Chemical production is complex and consumes numerous resources, including energy and raw materials. Because chemical production typically involves only a few tons, even small relative changes can lead to significant improvements. Therefore, enhancing the sustainability of industries that produce and use chemical products is highly desirable. [Overview of the Initiative] [Means for solving the problem]

[0003] overview Any disclosures, embodiments, and examples described herein relate to the methods, systems, apparatus, chemical products, and computer elements described above and below. Advantageously, any benefits derived from any embodiment or example are also applicable to all other embodiments and examples.

[0004] In another aspect, the Disclosure relates to a computer implementation for generating environmental characteristics data relating to the processing and / or production of a chemical product, the method comprising: providing chemical product data relating to the production and / or processing of a chemical product; providing a plurality of chemical product data templates indicating structures associated with at least a portion of the chemical product data; selecting at least one chemical product data template based on at least a portion of the chemical product data; generating contextualized chemical product data by combining at least a portion of the chemical product data and at least one chemical product data template; and providing the contextualized chemical product data to a data-driven model for generating environmental characteristics data, the data-driven model being parameterized and / or trained to provide environmental characteristics data in response to what is provided by the contextualized chemical product data.

[0005] In another embodiment, the present invention relates to a computer implementation method for generating chemical product production and / or processing data based on environmental characteristics data relating to environmental characteristics associated with the processing and / or production of chemical products, the method comprising: providing environmental characteristics data relating to the production and / or processing of chemical products; providing a plurality of environmental characteristics data templates showing structures associated with at least a portion of the environmental characteristics data; selecting at least one chemical product data template based on at least a portion of the environmental characteristics data; generating contextualized environmental characteristics data by combining at least a portion of the environmental characteristics data and at least one chemical product data template; and providing the contextualized chemical product data to a data-driven model for generating chemical product production and / or processing data based on the environmental characteristics data, the data-driven model being parameterized and / or trained to provide chemical product production and / or processing data in response to what is provided by the contextualized environmental characteristics data; and providing the chemical product production and / or processing data.

[0006] In another embodiment, the present invention relates to an apparatus for monitoring and / or controlling the production of chemical products, wherein the apparatus is Processor and A memory that stores instructions that, when executed by the processor, configure the device to perform any one step of the methods described herein, It is equipped with.

[0007] Another application relates to the use of environmental characteristics data relating to the processing and / or production of one or more chemical products, obtained by any one of the methods described herein, for controlling and / or monitoring the production and / or processing of one or more chemical products.

[0008] In another aspect, the present invention relates to a method for monitoring and / or controlling the production of at least one chemical product, particularly a computer-implemented method, wherein the method To provide chemical product data related to the production and / or processing of chemical products, To provide multiple chemical product data templates that show structures associated with at least some of the chemical product data, Selecting at least one chemical product data template by determining that at least one chemical product data template corresponds to at least a portion of the chemical product data, To generate contextualized chemical product data by merging at least a portion of chemical product data and at least one selected chemical product data template, Providing contextualized chemical product data to a data-driven model for generating environmental characteristics data relating to the processing and / or production of chemical products, wherein the data-driven model is configured to provide environmental characteristics data in response to the contextualized chemical product data provided by it. This includes providing environmental characteristics data to monitor and / or control the processing and / or production of at least one chemical product.

[0009] In another aspect, the present invention relates to the use of chemical product production data relating to the production of chemical products obtained by any one of the methods presented herein.

[0010] In another aspect, the present invention relates to a device and / or system including a processor and a memory for storing instructions that, when executed by the processor, constitute a device to perform any one step of the methods presented herein.

[0011] In another embodiment, the present invention relates to a non-temporary computer-readable storage medium that, when executed by a computer, includes instructions causing the computer to perform any one of the methods presented herein.

[0012] In another embodiment, the present invention relates to a computer implementation method for generating environmental characteristics data relating to the processing and / or production of a chemical product, the method comprising: receiving chemical product data relating to the production and / or processing of a chemical product; receiving a chemical product data template relating to a structure associated with at least a portion of the chemical product data; generating contextualized chemical product data by combining at least a portion of the chemical product data and the chemical product data template; and generating environmental characteristics data by providing the contextualized chemical product data to a data-driven model, the data-driven model being parameterized and / or trained to provide environmental characteristics data in response to being provided by the contextualized chemical product data based on a training dataset including historical data and environmental characteristics data relating to the chemical product data; and providing environmental characteristics data.

[0013] In another embodiment, the present invention relates to a computer implementation method for generating second chemical product data indicating the production and / or processing of a chemical product, wherein the first chemical product data comprises a first portion of chemical product data, and the second chemical product data comprises a second portion of chemical product data, the method comprising: receiving the first chemical product data indicating the production and / or processing of chemical product and environmental characteristic data; receiving a chemical product data template indicating a structure associated with the first chemical product data and at least a portion of the environmental characteristic data; generating contextualized chemical product data by combining the first chemical product data, environmental characteristic data, and at least a portion of the chemical product data template; and generating second chemical product data by providing the contextualized chemical product data to a data-driven model, the data-driven model being parameterized and / or trained to provide the second chemical product data in response to being provided by the contextualized chemical product data based on a training dataset including historical data and environmental characteristic data related to the chemical product; and providing the second chemical product data.

[0014] Embodiment Hereafter, terms used herein and / or the technical field of this disclosure will be outlined by definitions and / or examples. Where examples are given, it should be understood that this disclosure is not limited to those examples.

[0015] Chemical products are usually at the beginning of multiple diverse supply chains. Depending on the consumers of chemical products, the performance requirements, quantity, and quality of chemical products vary greatly. Therefore, chemical products adjusted for different application purposes are desired. Furthermore, the production of chemical products consumes large amounts of energy, water, and raw materials, emits carbon dioxide, and generates toxic waste. Additionally, strengthening the collaboration between the production and processing of chemical products is important for saving resources such as energy, water, and raw materials while reducing the amount of carbon dioxide emissions and toxic waste. This can be achieved by adjusting the production of chemical products according to the necessity of chemical product treatment. Moreover, providing accurate information regarding available chemical products helps to match the requirements for chemical products with the available chemical products. In order to reduce carbon dioxide emissions, the use of water, the amount of toxic waste, product safety, energy consumption, the production and / or processing of chemical products need to be judged according to the environmental characteristics of chemical products and the products obtained based on the processing of chemical products. By doing so, it is possible to compare multiple similar processes having similar error-free operation rates and conversion rates with respect to resource efficiency and the potential harm to the environment of the production and / or treatment of chemical products. Thereby, while ensuring an efficient process for producing and / or treating chemical products, the positive effect of the process on the environment is enhanced. Therefore, the present invention strongly contributes to maintaining a livable environment from a long-term perspective.

[0016] For this purpose, a data-driven model can be developed to generate environmental characteristic data as well as production and / or treatment data of chemical products. This is achieved by providing data based on chemical product data and environmental characteristic data to a data-driven model for generating environmental characteristic data as well as production and / or treatment data of chemical products. Furthermore, the data-driven model can be easily integrated into existing workflows and can generate output data such as environmental characteristic data and production and / or treatment data of chemical products quickly and reliably.

[0017] By generating contextualized chemical product data, it is possible to obtain further improved regulation of the production and / or processing of chemical products and to reduce the environmental impact associated with the production and / or processing of chemical products. This enhances user requirements and significantly improves human-machine interaction. As a result, the environmental characteristic data and the production and / or processing data of chemical products thus obtained are further improved in terms of accuracy and precision, thereby significantly reducing the amount of resources such as energy, water, and raw materials while reducing carbon dioxide emissions and the amount of toxic waste. Due to the large-scale production associated with the production and processing of chemical products, even a slight improvement will ultimately have a significant impact on various supply chains.

[0018] These objectives and other objectives that will become apparent upon reading the following description are solved by the subject matter of the independent claims. The dependent claims refer to preferred embodiments of the invention.

[0019] In one embodiment, the chemical product may refer to a product obtained by a chemical production process. The chemical production process may refer to a process including one or more chemical reactions. The chemical product may be characterized by at least one functional group. The functional group may be at least one of an alkyl group, an alkenyl group, an alkynyl group, a phenyl group, a carbonyl group, a ketone group, an aldehyde group, a hydroxyl group, a haloformyl group, an ester group, a carboxylate group, a halo group, a carboxyl group, a peroxy group, a carboalkoxy group, a hydroperoxyl group, an ether group, an acetal group, a hemiacetal group, a hemiketal group, a ketal group, a carboxylic acid anhydride group, a carboxamide group, an amidine group, an amine group, a ketamine group, an aldimine group, an imide group, a cyanate group, an azo group, a nitrite group, a nitrate group, a nitro group, a nitrile group, a sulfide group, a thiol group, a sulfinyl group, a sulfonyl group, a sulfo group, a thiocyanate group, a thionoester group, a thioester group, a phosphino group, a phosphono group, a phosphate group, or any combination thereof.

[0020] In one embodiment, chemical product data may relate to and / or indicate the production and / or processing of chemical products. Chemical product data may include and / or chemical product processing data. Furthermore, chemical product data may include environmental characteristics data.

[0021] In one embodiment, chemical product data may represent the production and / or processing of a chemical product. Chemical product data may include chemical product producer data, chemical product production orders, potential chemical product processing data and / or potential chemical product data, one or more material properties associated with a chemical product, chemical product processor data, chemical product processing orders, potential chemical product data, potential chemical product processing data, one or more material properties associated with a product based on a chemical product, one or more input products for producing a product based on a chemical product, one or more chemical compounds on which the chemical product is based, or a combination thereof. A product based on a chemical product may refer to a product obtained from processing a chemical product.

[0022] Chemical product producer data may indicate producers of chemical products and / or production facilities associated with the production of chemical products. For example, chemical product producer data may indicate one or more producers of a chemical product, quantitative production data, qualitative production data, or a combination thereof. Qualitative production data may indicate the chemical structure associated with the chemical product being produced. For example, qualitative production data may indicate the name of the chemical product, the quality of the chemical product, one or more chemical compounds associated with the chemical product, one or more material properties associated with the chemical product, one or more chemical compounds on which the chemical product is based, or a combination thereof. Quantitative production data may indicate the quantity of the chemical product.

[0023] Potential chemical product data may indicate projected production of a chemical product based on historical production and / or production associated with substitute products of the chemical product. For example, potential chemical product data may indicate historical production associated with a chemical product, substitutes of the chemical product, products produced based on the chemical product, application areas for the chemical product, or a combination thereof. In one embodiment, chemical product data may be received in audio and / or text format.

[0024] Chemical product processor data may indicate chemical product processors and / or processing equipment associated with the processing of chemical products. For example, chemical product processor data may indicate one or more chemical product processors for a chemical product, quantitative processing data, qualitative processing data, application fields for the chemical product, products produced based on the chemical product, application fields for products produced based on the chemical product, or a combination thereof. Qualitative processing data may include the name of the chemical product, the quality of the chemical product, one or more desired material properties associated with the chemical product, one or more chemical compounds contained in the chemical product, or a combination thereof. Quantitative processing data may indicate one or more components associated with the chemical product and their associated quantities. Chemical product processing instructions may indicate processing instructions associated with the processing of a chemical product. For example, chemical product processing instructions may relate to processing conditions such as temperature and pressure.

[0025] Potential chemical treatment data may indicate expected treatment for a chemical product based on its historical treatment and / or treatment associated with substitute products. For example, potential chemical treatment data may indicate historical treatment associated with a chemical product, substitute products for the chemical product, or a combination thereof.

[0026] In one embodiment, chemical product data may be received in audio and / or text format. In one embodiment, chemical product data may include one or more elements, preferably one or more chemical products and environmental characteristic parameter values. One or more elements, preferably chemical products and environmental characteristic parameter values, may be suitable for combination with a chemical product data template. Preferably, one or more elements, preferably one or more chemical products and environmental characteristic parameter values, may be inserted into a chemical product data template where the chemical product data template specifies the chemical products and environmental characteristic parameters. The chemical products and environmental characteristic parameters may be placeholders indicating where at least a portion of the chemical product data, preferably the chemical products and environmental characteristic parameter values, should be inserted.

[0027] In one embodiment, a chemical product data template represents a structure associated with at least a portion of the chemical product data. The chemical product data template may represent a sequence of one or more elements associated with at least a portion of the chemical product data. Preferably, the chemical product data template may represent a sequence of one or more elements associated with at least a portion of the chemical product data and a sequence of one or more elements included in the chemical product data template. The chemical product data template may represent one or more chemical product parameters and one or more elements associated with the relationship between one or more chemical product parameters, particularly one or more chemical product parameter values. A chemical product parameter may specify a location where a chemical product parameter value may be inserted. One or more chemical product parameters and one or more elements associated with the relationship between one or more chemical product parameters, particularly one or more chemical product parameter values, may constitute a sequence of two or more elements. Therefore, an element associated with the relationship between one or more chemical product parameters may specify a portion of the sequence of elements associated with the contextualized chemical product data. Additionally or alternatively, an element associated with the relationship between one or more chemical product parameters may specify one or more parts of the sequence into which a chemical product parameter value may be inserted.

[0028] In one embodiment, the chemical product data template may include at least a portion of the chemical product data. The chemical product data template may include chemical product producer data, chemical product production orders, potential chemical product processing data and / or potential chemical product data, one or more material properties associated with a chemical product, chemical product processor data, chemical product processing orders, potential chemical product data, potential chemical product processing data, one or more material properties associated with a product based on a chemical product, one or more input products for producing a product based on a chemical product, or at least one combination thereof.

[0029] In one embodiment, chemical product processing data may indicate processing related to a chemical product. Chemical product processing data may include chemical product processor data, chemical product processing instructions, potential chemical product data, and / or potential chemical product processing data.

[0030] In one embodiment, chemical product production and / or processing data may include chemical product data. Chemical product production and / or processing data may indicate the production and / or processing of a chemical product. Chemical product production and / or processing data may include chemical product producer data, chemical product production orders, potential chemical product processing data and / or potential chemical product data, one or more material properties associated with a chemical product, chemical product processor data, chemical product processing orders, potential chemical product data, potential chemical product processing data, one or more material properties associated with a product based on a chemical product, one or more input products for producing a product based on a chemical product, one or more chemical compounds on which a chemical product is based, or a combination thereof. A product based on a chemical product may refer to a product obtained by processing a chemical product.

[0031] In one embodiment, chemical product data may indicate production related to a chemical product. Chemical product data may include chemical product producer data, chemical product production orders, potential chemical product processing data, and / or potential chemical product data.

[0032] In one embodiment, a data-driven model may include one or more machine learning architectures and model parameters. The one or more machine learning architectures may be at least one of one of the following: one or more convolutional layers, one or more fully connected layers, one or more pooling layers, one or more transformer encoders, one or more classification layers, one or more transformer decoders, one or more feedforward layers, one or more linear layers, one or more transformer encoder decoders, or a combination thereof.

[0033] A data-driven model may be trained and / or parameterized based on a training dataset. The training dataset may include chemical product data and historical data related to chemical product production and / or processing data. Furthermore, the training dataset may include at least one dataset containing chemical product data and corresponding chemical product production and / or processing data. The training data may include one or more element sequences. Preferably, at least one dataset within the training dataset may include a sequence of elements specifying chemical product data and corresponding chemical product production and / or processing data. A data-driven model may be trained and / or parameterized based on the training dataset to provide chemical product production and / or processing data in response to what is provided by contextualized chemical product data. A data-driven model may also be a generative model. A generative model may be trained and / or parameterized based on a training dataset, preferably based on what is provided by contextualized chemical product data, to generate chemical product production and / or processing data. Preferably, the training dataset may include chemical product production and / or processing data and numerical and / or sentence-based historical data relating to chemical product data. Numerical historical data may refer to historical data containing one or more numerical values. Sentence-based historical data may refer to historical data that includes at least a portion of a sentence. Data-driven models can be trained and / or parameterized to sequentially generate chemical production and / or processing data, particularly based on the provision of contextualized chemical data.

[0034] A data-driven model may be trained and / or parameterized to sequentially generate chemical production and / or processing data, particularly based on the provision of contextualized chemical data. The data-driven model may be trained and / or parameterized to generate chemical production and / or processing data including element sequences by generating elements of an element sequence, preferably the first element of the element sequence, based on the contextualized chemical data, and generating further elements, preferably elements following the first element in the sequence, based on the contextualized chemical data and previously generated elements in the sequence. By taking previously generated elements into account, the data-driven model can generate chemical production and / or processing data that is more accurately linked and better linked to other elements in the sequence. Ultimately, this enables customized production and helps to use resources as efficiently as possible.

[0035] In one embodiment, the data-driven model may be a pre-trained data-driven model. The pre-trained data-driven model may be trained on general data to provide output data based on what is provided by the input data. In one embodiment, the data-driven model may be a fine-tuned data-driven model. The fine-tuned data-driven model may be trained on general data, preferably in a first training process, to provide output data based on what is provided by the input data. The fine-tuned data-driven model may be further trained in a second training on historical chemical data to provide chemical production and / or processing data based on what is provided by contextualized chemical data. The general data may include historical input data and / or historical output dates. Input data and / or output data may refer to data that can be processed by the data-driven model, preferably by a parameterized and / or initialized data-driven model. Input data and / or output data may include and / or represent one or more element sequences, and an element sequence may include two or more elements. In particular, an element sequence may represent a sequence of two or more elements of a sequence.

[0036] Training a data-driven model means the process of building a data-driven model, in particular the process of determining and / or updating the parameters of the data-driven model, and / or the training process may be the process of building a data-driven model, in particular the process of determining and / or updating the parameters of the data-driven model. During the training process, the data-driven model may be adjusted to achieve the best fit with the training dataset, for example, by associating at least one ON input value with at least one desired output value in the best fit. For example, if the neural network is a feedforward neural network such as a CNN, a backpropagation algorithm may be applied to train the neural network. In the case of an RNN, a gradient descent algorithm may be used for training purposes. The gradient descent algorithm can use gradients to update parameters. Gradient can indicate the degree of change in the parameters of the data-driven model. Gradient can be obtained by backpropagation. Therefore, the gradient descent algorithm may be based on backpropagation. The training process may be terminated when the deviation of the output generated by the data-driven model compared to the target output specified by the training dataset is within a given range. Determining and / or updating the parameters of the data-driven model may be terminated when the training process can be terminated. The output generated by the data-driven model may be chemical production and / or processing data, historical data related to chemical production and / or processing data, or a combination thereof. The target output specified by the training dataset may be chemical production and / or processing data, historical data related to chemical production and / or processing data, or a combination thereof. During the training process and / or training of the data-driven model, the training dataset may contain one or more element sequences, one or more element sequences may be sequentially provided to the data-driven model, and / or the data-driven model may sequentially generate outputs.Therefore, a data-driven model may generate a first element of one or more element sequences based on contextualized chemical product data, and may generate further elements, such as a second element of one or more sequences, based on one or more previously generated elements and contextualized chemical product data. A second training process following the first training process may be beneficial for tailoring the data-driven model to use cases of chemical product production and processing. This helps to further improve the accuracy of the data-driven model.

[0037] In one embodiment, a transformer encoder may comprise an encoder input, one or more encoder blocks, and / or an encoder output. The encoder input may generate embedded input chemical processing data based on the reception of contextualized chemical processing data. One or more encoder blocks may generate a chemical data context tensor based on the reception of embedded input chemical processing data. The encoder output may generate chemical production and / or processing data based on the reception of the chemical data context tensor. A transformer decoder may comprise a decoder input, one or more decoder blocks, and / or a decoder output. The decoder input may generate embedded input chemical processing data based on the reception of contextualized chemical processing data. One or more decoder blocks may generate a chemical data context tensor based on the reception of embedded input chemical processing data. The decoder output may generate chemical production and / or processing data based on the reception of the chemical data context tensor. A transformer encoder decoder may comprise a transformer encoder comprising an encoder input and one or more encoder blocks, and a transformer decoder comprising a decoder input, one or more decoder blocks, and a decoder output.

[0038] In one embodiment, generating chemical production and / or processing data by providing contextualized chemical product data to a data-driven model may include, for example, receiving contextualized chemical product data at encoder inputs and / or decoder inputs; applying input embeddings to the contextualized chemical product data to obtain, for example, embedded input chemical product processing data; applying position coding to the contextualized chemical product data and / or embedded input chemical product processing data; applying self-attention to the contextualized chemical product data and / or embedded input chemical product processing data to obtain, for example, a chemical product data context tensor; and applying a softmax function to the contextualized chemical product data.

[0039] In one embodiment, the method may further include embedding contextualized chemical product data by a data-driven model, depending on whether it is provided by contextualized chemical product data. Embedding contextualized chemical product data may mean applying input embedding and / or position coding to the contextualized chemical product data. Embedding contextualized chemical product data may include converting the contextualized chemical product data into machine-processable contextualized chemical product data such as tensors. The data-driven model may process the machine-processable contextualized chemical product data into chemical product data. In this way, the contextualized chemical product data can be received in a user language, and accurate, robust, and efficient processing of user input can be handled by the data-driven model. This makes it possible to generate chemical product data using the data-driven model in a barrier-free manner. Furthermore, by providing access to all users, all users are provided with the opportunity to deploy the method and system presented herein. This increases the number of problems that can be handled by the method and system, resulting in a reduction in the amount of errors in the processing and production of chemical products, while leading to more coordinated processing and production of chemical products. Therefore, resources for processing and producing chemical products are used more efficiently and with fewer errors.

[0040] Additionally or alternatively, embedding contexted chemical product data may involve passing the contexted chemical product data through an embedding layer. The embedding layer may be suitable for converting the contexted chemical product data into a machine-processable format. The machine-processable format may refer to a numerical, and in particular, tensor-based, representation of the contexted chemical product data. Embedding contexted chemical product data may yield embedded input chemical processing data. This embedded input chemical processing data may include tensors representing the contexted chemical product data. The tensor-based representation of the contexted chemical product data may be called the embedded input chemical processing data. This embedded input chemical processing data may also be called the machine-processable format of the contexted chemical product data.

[0041] In one embodiment, the method may further include processing embedded input chemical processing data into a chemical data context tensor using a data-driven model. Processing embedded input chemical processing data into a chemical data context tensor may mean converting embedded input chemical processing data into a chemical data context tensor. Preferably, embedded input chemical processing data can be converted into a chemical data context tensor by forming one or more tensor products and / or one or more representations associated with the embedded input chemical processing data. One or more representations associated with the embedded input chemical processing data can be obtained by applying one or more mathematical operations to the embedded input chemical processing data. One or more mathematical operations may include, for example, adding, subtracting, dividing, integrating, forming a derivative, multiplying, normalizing, or a combination thereof.

[0042] In one embodiment, embedded input chemical processing data may be sequence-specific. Sequence-specific embedded input chemical processing data may refer to a representation of embedded input chemical processing data that takes into account a set of data points associated with the embedded input chemical processing data. Sequence-specific embedded input chemical processing data may be obtained by applying input embedding and / or location coding to contextualized chemical data. For example, applying location coding may refer to adding and / or multiplying one or more portions of embedded input chemical processing data by location coefficients that indicate the location of one or more portions of embedded input chemical processing data within the embedded input chemical processing data. Sequence-specific embedded input chemical processing data may be processed, in particular, transformed in the same way as embedded input chemical processing data.

[0043] In one embodiment, the received chemical product production and / or processing data may be sentence-based. Therefore, a chemical product data template, which shows the structure associated with at least a portion of the chemical product data and / or contextualized chemical product data, may also be sentence-based. Thus, the data-driven model may be a natural language processing model. Sentence-based contextualized chemical product data can be embedded, resulting in embedded input chemical product processing data. Sentence-based contextualized chemical product data can be embedded by input embeddings, preferably word embeddings. Word embedding may refer to converting sentence-based contextualized chemical product data into a machine-processable format.

[0044] In one embodiment, the determined chemical product production and / or processing data may include two or more parts, specifically a first part of the chemical product production and / or processing data and a second part of the chemical product production and / or processing data. If the chemical product production and / or processing data may be sentence-based, the two or more parts may refer to words, numbers, and / or word portions. The first part of the chemical product production and / or processing data may refer to a first part in a sequence of chemical product production and / or processing data, and the second part of the chemical product production and / or processing data may refer to a second part in a sequence of chemical product production and / or processing data. The first part may be determined in a first time step, and the second part may be determined in a second time step.

[0045] The first part of the chemical product production and / or processing data may refer to one element associated with the chemical product production and / or processing data, preferably a first element associated with the chemical product production and / or processing data. The second part of the chemical product production and / or processing data may refer to one element associated with the chemical product production and / or processing data, preferably a second element associated with the chemical product production and / or processing data.

[0046] In a first time step, first chemical product production and / or processing data may be determined based on embedded input chemical product processing data. In a second time step, second chemical product production and / or processing data may be determined based on embedded input chemical product processing data and embedded first chemical product production and / or processing data. Embedded first chemical product production and / or processing data may refer to first chemical product production and / or processing data embedded in a manner similar to contextualized chemical product data. Subsequently, chemical product production and / or processing data may be generated based on contextualized chemical product data and at least partially based on at least a portion of the generated chemical product production and / or processing data, preferably by further providing at least a portion of the generated chemical product production and / or processing data to a data-driven model, in particular by providing at least a first portion of the generated chemical product production and / or processing data to a data-driven model.

[0047] In one embodiment, an element may refer to a part of a sequence. Preferably, an element may be a word, a number, a part of a word, or a combination thereof.

[0048] In one embodiment, environmental characteristics may relate to the processing and / or production of a chemical product. Environmental characteristics may include at least one of the following: product emission data, product recyclable content, product bio-based content, product renewable content, product declaration data, product safety data, proportion of the product to be recycled, proportion of the product to be biodegradable, proportion of the product to be non-biodegradable, or a combination thereof. In this regard, the product may be a chemical product and / or a product produced based on a chemical product.

[0049] Products produced based on chemical products may refer to products in which chemical products were used to produce the product.

[0050] Emission data may include any data related to the environmental footprint. The environmental footprint may refer to an entity and its associated environmental footprint. The environmental footprint may be entity-specific. For example, the environmental footprint may relate to a product, especially a chemical product, a company, a process such as a manufacturing process, raw materials or basic substances, chemical products or materials, components, component assemblies, finished products, combinations thereof, or additional entity-specific relationships. Emission data may include data related to the carbon footprint of a product, especially a chemical product. Emission data may include data related to greenhouse gas emissions released in the production of a product, especially a chemical product. Emission data may include data related to greenhouse gas emissions. Greenhouse gas emissions may include emissions such as carbon dioxide (CO2) emissions, methane (CH4) emissions, nitrous oxide (N2O) emissions, hydrofluorocarbon (HFC) emissions, perfluorocarbon (PFC) emissions, sulfur hexafluoride (SF6) emissions, nitrogen trifluoride (NF3) emissions, combinations thereof, and additional emissions. Emissions data may include data related to greenhouse gas emissions from the entity's or firm's own operations (production, power supply to plants, and waste incineration). Scope 2 includes emissions from externally supplied energy production. Scope 3 includes all other emissions along the value chain. Specifically, this includes greenhouse gas emissions from raw materials obtained from suppliers. Product carbon footprint (PCF) sums up greenhouse gas emissions and removals from a series of interconnected process steps associated with a particular product. Cradle-to-gate PCF may sum up greenhouse gas emissions based on selected process steps, from resource extraction to the factory gate where the product leaves the firm. Such a PCF is called a partial PCF. To achieve such a sum, each firm providing any product must provide Scope 1 and Scope 2 contributions to the PCF for each of its products as accurately as possible, and it must be possible to obtain reliable and consistent data for the PCF of purchased energy (Scope 2) and its raw materials (Scope 3).

[0051] In one embodiment, environmental characteristic data may refer to data related to environmental characteristics, and preferably, environmental characteristic data may indicate environmental characteristics. Furthermore, environmental characteristic data may indicate environmental characteristics related to chemical products and / or products obtained from the processing of chemical products. Environmental characteristic data may indicate environmental characteristics related to the processing and / or production of chemical products.

[0052] In one embodiment, the environmental characteristics of a chemical product may include at least one of the following: emissions data for the chemical product, the recycled content of the chemical product, the bio-based content of the chemical product, the renewable content of the chemical product, declaration data for the chemical product, safety data for the chemical product, the proportion of the chemical product that is recycled, the proportion of the chemical product that is biodegradable, the proportion of the chemical product that is non-degradable, or a combination thereof. Emission data may include any data related to the environmental footprint. The environmental footprint may refer to an entity and the environmental footprint associated with it. The environmental footprint may be entity-specific. For example, the environmental footprint may relate to a chemical product, a company, a process such as a manufacturing process, raw materials or basic substances, chemical products or materials, components, component assemblies, final products, combinations thereof, or additional entity-specific relationships. Emission data may include data related to the carbon footprint of the chemical product. Emission data may include, for example, data related to greenhouse gas emissions released in the production of the chemical product. Emission data may include data related to greenhouse gas emissions. Greenhouse gas emissions may include emissions such as carbon dioxide (CO2), methane (CH4), nitrous oxide (N2O), hydrofluorocarbon (HFC), perfluorocarbon (PFC), sulfur hexafluoride (SF6), nitrogen trifluoride (NF3), combinations thereof, and additional emissions. Emission data may include data related to greenhouse gas emissions from the operations of the entity or company itself (production, power supply to plants, and waste incineration). Scope 2 includes emissions from externally supplied energy production. Scope 3 includes all other emissions along the value chain. Specifically, this includes greenhouse gas emissions from raw materials obtained from suppliers. The product carbon footprint (PCF) sums up greenhouse gas emissions and removals from consecutive, interconnected process steps associated with a particular product. Cradle-to-gate plant-factory emissions (PCFs) can be the sum of greenhouse gas emissions based on selected process steps, from resource extraction to the factory gate where the product leaves the company. Such PCFs are called partial PCFs.To achieve such a total, each company providing any product must provide, as accurately as possible, the Scope 1 and Scope 2 contributions to the PCF for each of its products, and must be able to obtain reliable and consistent data on the PCF of purchased energy (Scope 2) and its raw materials (Scope 3).

[0053] In one embodiment, the environmental characteristic data template represents a structure associated with at least a portion of the environmental characteristic data. The environmental characteristic data template may represent a sequence of one or more elements associated with the environmental characteristic data and / or at least a portion of the environmental characteristic data template. The environmental characteristic data template may represent one or more environmental characteristic parameters and one or more elements associated with the relationship between one or more environmental characteristic parameters, particularly between environmental characteristic parameter values. The environmental characteristic parameters may specify where the environmental characteristic parameter values ​​may be inserted. One or more environmental characteristic parameters and one or more elements associated with the relationship between one or more environmental characteristic parameters, particularly between the values ​​of one or more environmental characteristic parameters, may constitute a sequence of two or more elements. Thus, the element associated with the relationship between one or more environmental characteristic parameters may specify a portion of a sequence of elements associated with contextualized environmental characteristic data. Additionally or alternatively, the element associated with the relationship between one or more environmental characteristic parameters may specify one or more portions of a sequence in which the environmental characteristic parameter values ​​may be inserted adjacent, preferably before, after, and / or in between.

[0054] In one embodiment, the environmental characteristics data template may include at least a portion of the environmental characteristics data.

[0055] In one embodiment, material properties may refer to physical properties, chemical properties, or a combination thereof. Chemical properties may be properties that can only be established by modifying one or more chemical structures associated with at least one chemical product. Examples of chemical properties may be acidity, oxidation state, or reactivity. Physical properties may be one of the following: mechanical properties, electrical properties, optical properties, thermal properties, etc. For example, physical properties may include one or more of the following: density, scratch resistance, conductivity, color, absorptiveness, heat capacity, etc.

[0056] In one embodiment, memory may be physical system memory, which may be volatile, non-volatile, or a combination thereof. Memory may include non-volatile mass storage devices such as physical storage media. Memory may be computer-readable storage media, such as any physical, tangible storage media that can be used to store desired program code means in the form of computer-executable instructions or data structures and can be accessed by a computing system. Furthermore, memory may be a computer-readable medium (also called a transmission medium) that carries computer-executable instructions. Furthermore, upon reaching various computing system elements, program code means in the form of computer-executable instructions or data structures may be automatically transferred from the transmission medium to the storage medium (or vice versa). For example, computer-executable instructions or data structures received via a network or data link may be buffered in RAM within a network interface module (e.g., a "NIC") and then finally transferred to computing system RAM and / or a less volatile storage medium in the computing system. Thus, it should be understood that the storage medium may be included in computing elements that also utilize (or rather exclusively utilize) the transmission medium.

[0057] In one embodiment, the processing data template represents a structure associated with at least a portion of the chemical product processing data. The processing data template may represent a sequence of one or more elements associated with at least a portion of the chemical product processing data. Preferably, the processing data template may represent a sequence of one or more elements associated with at least a portion of the chemical product processing data and one or more elements included in the processing data template.

[0058] A processing data template may represent one or more processing parameters and one or more elements associated with the relationships between the one or more processing parameters, in particular one or more processing parameter values. The processing parameters may specify where the processing parameter values ​​may be inserted. One or more processing parameters and one or more elements associated with the relationships between the one or more processing parameters, in particular one or more processing parameter values, may constitute a sequence of two or more elements.

[0059] Therefore, elements associated with relationships between one or more processing parameters may specify a portion of a sequence of elements associated with contextualized chemical product data. Additionally or alternatively, elements associated with relationships between one or more processing parameters may specify one or more portions of a sequence in which processing parameter values ​​can be inserted adjacent, preferably before, after, and / or in between.

[0060] In one embodiment, the processing data template may include at least a portion of the chemical product processing data. The processing data template may include one or more of the following: chemical product processor data, chemical product processing instructions, potential chemical product processing data, or a combination thereof.

[0061] In one embodiment, a processor may refer to any logic circuit and / or generally a device configured to perform calculations or logical operations, configured to perform basic operations of a computer or system. In particular, a processor or computer processor may be configured to process basic instructions that drive a computer or system. This may be a semiconductor-based processor, a quantum processor, or any other type of processor configured to process instructions. For example, a processor may be or include a central processing unit ("CPU"). A processor may be a graphics processing unit ("GPU"), a tensor processing unit ("TPU"), a composite instruction set computing microprocessor ("CISC"), a reduced instruction set computing ("RISC") microprocessor, a very long instruction word ("VLIW") microprocessor, or a processor implementing a combination of other instruction sets. The processing means may also be one or more dedicated processing units such as application-specific integrated circuits ("ASIC"), field-programmable gate arrays ("FPGA"), composite programmable logic circuits ("CPLD"), digital signal processors ("DSP"), network processors, etc. The methods, systems, and apparatus described herein may be implemented as software within a DSP, microcontroller, or any other subprocessor, or as hardware circuitry within an ASIC, CPLD, or FPGA. The term “processor” may also refer to one or more processing units, such as a distributed system of processing units deployed across multiple computer systems (e.g., cloud computing), and should be understood that it is not limited to a single device unless otherwise specified. A processor may also be an interface to a remote computer system, such as a cloud service. A processor may include, or may be, a secure enclave processor (SEP). An SEP may be a secure circuit configured to process a spectrum. A “secure circuit” is a circuit that prevents isolated internal resources from being directly accessed by external circuits.The processor may be an image signal processor (ISP) and may include circuitry suitable for processing images, particularly images containing personal and / or confidential information.

[0062] In one embodiment, chemical product data may include one or more elements associated with the chemical product data, and production and / or multiple chemical product data templates may include one or more elements. Chemical product data may include one or more chemical data elements, and / or at least one of the multiple chemical product data templates may include one or more template elements. Contextualized chemical product data may include one or more chemical data elements and one or more template elements.

[0063] In one embodiment, an element may refer to at least part of a word, at least part of a number, part of a table, or a combination thereof. In addition or alternatively, a sequence of two or more elements may refer to at least part of a sentence, at least part of a number sequence, at least part of a table, or a combination thereof. Part of a table may refer to a row number, a column number, or a table entry. Furthermore, contextualized chemical product data may include two or more elements that are equivalent to one or more elements associated with the chemical product data and / or one or more elements of at least one chemical product data template. Furthermore, contextualized chemical product data may include a sequence of two or more elements that are equivalent to one or more elements associated with the chemical product data and / or one or more elements of at least one chemical product data template. Additionally or alternatively, chemical product data may include one or more elements associated with the chemical product data. Furthermore, at least one of a plurality of chemical product data templates may include one or more elements.

[0064] In one embodiment, the method may further include receiving second chemical product data based on chemical product processing data or receiving second chemical product processing data based on chemical product data; optionally receiving a second chemical product data template or receiving a second processing data template; generating second contextualized chemical product data based on second chemical product data and optionally a chemical product data template which is a second chemical product data template or generating second contextualized chemical product data based on second chemical product processing data and optionally a processing data template which is a second processing data template; generating second chemical product processing data based on second contextualized chemical product data or generating second chemical product data based on second contextualized chemical product data; and providing second chemical product processing data or second chemical product data. Additionally or alternatively, the system may receive or receive second contextualized chemical product data based on at least a portion of chemical product processing data, generate or generate second chemical product data based on second contextualized chemical product data, or provide second chemical product processing data or second chemical product data. The second chemical product data, second chemical product processing data, second contextualized chemical product data, second contextualized chemical product data, or a combination thereof, may be received via a user interface. This allows users to obtain solutions tailored to their specific situations. Furthermore, additional input from the user may be obtained, for example, to further improve the generated data or to react to changes in conditions involving the production and / or processing of chemical products while building upon previously generated data. This makes it possible to cover resource-efficient use cases where conditions are constantly changing.

[0065] In one embodiment, the method may further include providing instructions for selecting a data-driven model from a plurality of data-driven models, and selecting a data-driven model from the plurality of data-driven models based on the instructions. Preferably, the instructions for selecting a data-driven model may be provided by the user, preferably via a user interface. Two or more data-driven models may be available. The two or more data-driven models may be distinguished by their intended application. For example, a first data-driven model among the two or more data-driven models may be more accurate than a second data-driven model among the two or more data-driven models. On the other hand, the second data-driven model may generate chemical product processing data and / or chemical product data faster than the first data-driven model. In situations where the accuracy of the generated data is valued more highly than the speed during data generation, the corresponding first data-driven model may be selected. In situations where the speed of data generation is of higher value than the accuracy of the generated data, the second data-driven model may be selected. Additionally or alternatively, one or more of the two or more data-driven models may be more resource-efficient than the other data-driven models among the two or more data-driven models. Thus, the model best suited to each application can be selected. This enables customized chemical processing data and chemical product data while conserving resources as much as possible. Instructions may be provided by the user. Additionally or alternatively, instructions may be based on one or more elements associated with chemical product data, and / or representations may be provided by providing chemical product data. Thus, instructions may include one or more elements associated with chemical product data. Data-driven models may be selected based on chemical product data, preferably based on one or more elements associated with chemical product data.

[0066] In one embodiment, a plurality of chemical product data templates may include a plurality of types of chemical product data templates. Selecting at least one chemical product data template based on at least a portion of the chemical product data and / or associated elements may mean selecting at least one chemical product data template based on the type of chemical product data template. Furthermore, the type of chemical product data template may be selected based on the determination that the type of chemical product data template corresponds to chemical product data, in particular to one or more elements associated with the chemical product data. Chemical product data may include one or more elements associated with the chemical product data. The type of chemical product data template may be associated with one or more elements associated with the chemical product data. Therefore, selecting at least one chemical product data template may mean selecting at least one chemical product data template, and the type of at least one chemical product data template may correspond to one or more elements associated with the chemical product data. The type of chemical product data template specifies that the chemical product data template includes at least one of the following: chemical product producer data, chemical product production orders, chemical product processor data, chemical product processing orders, one or more material properties related to the chemical product, and one or more material properties related to products based on the chemical product. By doing so, the interaction with the data-driven model can be tailored to the desired use case, resulting in more accurate chemical production and / or processing data. This enables greater resource efficiency in the production and processing of chemical products.

[0067] In one embodiment, the method may further include selecting a data-driven model from a plurality of data-driven models based on chemical product data. Preferably, the data-driven model may be selected from a plurality of data-driven models based on at least one chemical product data template, in particular the type of at least one chemical product data template. By doing so, the interaction with the data-driven model can be tailored to the desired use case, resulting in more accurate chemical product production and / or processing data. This enables greater resource efficiency in the production and processing of chemical products.

[0068] In one embodiment, a plurality of chemical product data templates may be provided via a database, and selecting at least one chemical product data template based on at least a portion of the chemical product data and / or associated elements may include generating embedded chemical product data, generating one or more embedded chemical product data templates, and selecting a chemical product data template based on the determination that the tensor product between the tensor associated with the embedded chemical product data and the tensor associated with at least one of the one or more chemical product data templates falls within a predefined numerical range.

[0069] Generating embedded chemical product data may refer to providing chemical product data to an encoder and receiving embedded chemical product data from the encoder. Generating one or more embedded chemical product data templates may refer to providing chemical product data templates to an encoder and receiving one or more embedded chemical product data templates from the encoder. The encoder may be configured to convert the chemical product data and chemical product data templates into a machine-readable format, such as a tensor. Thus, the encoder can reduce the dimensionality of the chemical product data and chemical product data templates. An example of such an encoder may be the embedding layer of a CBOW model, as described in the context of Figure 8. Multiple encoders are available for embedding data, such as word2vec, GloVe, and FastText. The tensor product between the embedded chemical product data and the embedded chemical product data templates may refer to determining the similarity between the embedded chemical product data and one or more embedded chemical product data templates. Selecting at least one chemical product data template based on the tensor product may refer to selecting at least one chemical product data template that is most similar to the chemical product data. Searching processed chemical product data templates based on embedded chemical product data enables cross-category searches and allows for comparison of content levels. For example, a compound is typically associated with multiple names, such as UIPAC nomenclature and historically developed self-evident names. Therefore, when ethylene and ethene are mentioned, they refer to the same compound, but with different names. Searching through embedded data makes it possible to link these two distinct words to the same compound. Thus, using an embedded database significantly improves search quality.

[0070] In one embodiment, chemical product data may include one or more elements associated with the chemical product data, and chemical product processing data may include one or more elements associated with the chemical product processing data. One or more elements may be part of a sequence of elements, in particular a sequence specified by contextualized chemical product data and / or contextualized chemical product data.

[0071] In one embodiment, the contexted chemical product data may include a sequence of two or more elements associated with the contexted chemical product data, and / or a sequence of two or more elements associated with the contexted chemical product data.

[0072] In one embodiment, an element may refer to a word and / or a part of a word. In one embodiment, a sequence of two or more elements may be a sentence and / or a part of a sentence.

[0073] By doing so, contextualized chemical product data, chemical product data, chemical product processing data, and contextualized chemical product data can be received in the user language, and accurate, robust, and efficient processing of user input can be handled by a data-driven model. This makes it possible to generate chemical product data and chemical product processing data using the data-driven model in a barrier-free manner. Furthermore, by providing access to all users, all users are given the opportunity to deploy the methods and systems presented herein. This increases the number of problems that can be handled by the methods and systems, resulting in more coordinated processing and production of chemical products, while reducing the amount of errors in the processing and production of chemical products. Thus, resources for the processing and production of chemical products are used more efficiently and with fewer errors.

[0074] In one embodiment, the system may further include one or more databases configured to provide processing data templates and / or chemical product data templates. Storing the processing data templates and chemical product data templates in the databases makes them readily available when needed, while also enabling robust retrieval of the processing data templates and chemical product data templates.

[0075] In one embodiment, the system may further include a user interface configured to receive chemical product processing data and / or chemical product data, in particular from the user, and / or to provide chemical product processing data and / or chemical product data, in particular to the processor. The user interface may enable the user to interact with a data-driven model. Thus, the adjusted chemical product data and chemical product processing data are generated under the user's control.

[0076] In one embodiment, a processor may refer to any logic circuit and / or generally a device configured to perform calculations or logical operations, configured to perform basic operations of a computer or system. In particular, a processor or computer processor may be configured to process basic instructions that drive a computer or system. This may be a semiconductor-based processor, a quantum processor, or any other type of processor configured to process instructions. For example, a processor may be or include a central processing unit ("CPU"). A processor may be a graphics processing unit ("GPU"), a tensor processing unit ("TPU"), a composite instruction set computing microprocessor ("CISC"), a reduced instruction set computing ("RISC") microprocessor, a very long instruction word ("VLIW") microprocessor, or a processor implementing a processor or combination of other instruction sets. The processing means may also be one or more dedicated processing units such as application-specific integrated circuits ("ASIC"), field-programmable gate arrays ("FPGA"), composite programmable logic circuits ("CPLD"), digital signal processors ("DSP"), and network processors. The methods, systems, and apparatus described herein may be implemented as software within a DSP, microcontroller, or any other subprocessor, or as hardware circuitry within an ASIC, CPLD, or FPGA. The term “processor” may also refer to one or more processing units, such as a distributed system of processing units deployed across multiple computer systems (e.g., cloud computing), and should be understood that it is not limited to a single device unless otherwise specified. A processor may also be an interface to a remote computer system, such as a cloud service. A processor may include, or may be, a secure enclave processor (SEP). An SEP may be a secure circuit configured to process a spectrum. A “secure circuit” is a circuit that prevents isolated internal resources from being directly accessed by external circuits.The processor may be an image signal processor (ISP) and may include circuitry suitable for processing images, particularly images containing personal and / or confidential information.

[0077] In one embodiment, memory may be a physical system memory that is volatile, non-volatile, or a combination thereof. Memory may include non-volatile mass storage devices such as physical storage media. Memory may be computer-readable storage media such as RAM, ROM, EEPROM, CD-ROM or other optical disk storage devices, magnetic disk storage devices or other magnetic storage devices, non-magnetic disk storage devices such as solid-state disks, or any other physical and tangible storage media accessible by a computing system that can be used to store desired program code means in the form of computer-executable instructions or data structures. Furthermore, memory may be a computer-readable medium (also called a transmission medium) that carries computer-executable instructions. Furthermore, upon reaching various computing system elements, program code means in the form of computer-executable instructions or data structures may be automatically transferred from the transmission medium to the storage medium (or vice versa). For example, computer-executable instructions or data structures received via a network or data link may be buffered in RAM within a network interface module (e.g., "NIC") and then finally transferred to computing system RAM and / or a less volatile storage medium in the computing system. Thus, it should be understood that storage media can be included in computing elements that also utilize (or rather exclusively utilize) transmission media.

[0078] In one embodiment, multiple chemical product data templates may be provided via a database, and selecting at least one chemical product data template based on at least a portion of the chemical product data and / or associated elements includes providing a query to the database based on the chemical product data template and / or receiving at least one chemical product data template based on the query. The query may define categories associated with searching for chemical product data templates. This enables predictable, robust, and human-relevant searches. Furthermore, users can define and refine searches as needed.

[0079] In one embodiment, a chemical product template and / or contextualized chemical product data may include a data generation task instruction for triggering a data-driven model to generate environmental characteristic data. The data generation task instruction may be associated with unstructured data, particularly string data. The data generation task instruction may be associated with and / or configured to trigger a data-driven model. The task instruction may be human-interpretable. Thus, the task instruction may be tailored, for example by a user, thereby enabling improved human-machine interaction.

[0080] In one embodiment, a plurality of chemical product data templates may include a plurality of types of chemical product data templates. Selecting at least one chemical product data template based on at least a portion of the chemical product data may mean selecting at least one chemical product data template based on the type of chemical product data template. The type of chemical product data template may relate to the production and / or processing of a chemical product and / or one or more properties of a chemical product. One or more properties of a chemical product may include one or more material properties and / or one or more environmental properties. In particular, the type of chemical product data template may be at least one of the following: chemical product producer data, chemical product production orders, chemical product processor data, chemical product processing orders, one or more material properties associated with a chemical product, one or more material properties associated with a product based on a chemical product, one or more environmental properties associated with a chemical product, one or more environmental properties associated with a product based on a chemical product, one or more environmental properties associated with one or more chemical compounds on which the chemical product is based, or a combination thereof.

[0081] In one embodiment, a data-driven model may be a pre-trained data-driven model. A pre-trained data-driven model may be trained on unstructured data to provide multiple different output datasets based on the provision of multiple different input datasets. A pre-trained data-driven model may be configured to perform multiple different tasks according to multiple different task instructions. A pre-trained data-driven model may be parameterized and / or trained on unstructured data, particularly text data and optionally tabular data or numerical data such as image data. A pre-trained data-driven model may be configured to perform multiple tasks. A pre-trained data-driven model may be configured to perform tasks according to provided task instructions. Therefore, a pre-trained data-driven model may be configured to provide multiple different task instructions and / or to provide multiple different types of output data upon receiving different task instructions.

[0082] In one embodiment, the data-driven model may be a fine-tuned data-driven model. The fine-tuned data-driven model may be a pre-trained data-driven model trained to provide environmental characteristic data based on what is provided by contextualized chemical product data. The pre-trained data-driven model may be trained on unstructured data to provide multiple different output datasets based on what is provided by multiple different input datasets. The multiple different input datasets may include and / or be independent of the training of contextualized chemical product data. The multiple different output datasets may include and / or be independent of the training environmental characteristic data. The fine-tuned data-driven model may be obtained by training a pre-trained data-driven model configured to perform multiple tasks according to multiple task instructions. The fine-tuned data-driven model may be further trained with a training dataset containing multiple task instructions of one type and corresponding output data. The fine-tuned data-driven model may be further trained to provide output data of a predefined type according to a training dataset containing multiple task instructions of one type and corresponding output data. A finely tuned data-driven model may be configured to provide multiple different task instructions and / or to provide multiple different types of output data when it receives different types of task instructions. Furthermore, a finely tuned data-driven model may be configured to provide a certain type of output data when it receives a certain type of task instruction with greater precision than it would when it receives other types of task instructions and provides other types of output data.

[0083] In one embodiment, multiple chemical product data templates may be provided via a database. Selecting at least one chemical product data template based on at least a portion of the chemical product data may include providing a query to the database to provide at least one chemical product data template based on the chemical product data and / or providing at least one chemical product data template based on the query. The query may be configured to trigger the database to provide at least one chemical product data template. The query may relate to the type of chemical product data and / or the chemical product data template. The type of chemical product data template relates to the production and / or processing of the chemical product and / or one or more properties of the chemical product. The type of chemical product data template may be provided, for example, together with the chemical product data. The type of chemical product data template may be linked to the chemical product data. Therefore, by providing the chemical product data, it may be possible to provide the type of chemical product data template.

[0084] In one embodiment, multiple chemical product data templates may be provided via a database. Selecting at least one chemical product data template based on at least a portion of the chemical product data may include generating embedded chemical product data. The embedded chemical product data may be a numerical representation of the chemical product data. Selecting at least one chemical product data template based on at least a portion of the chemical product data may further include generating one or more embedded chemical product data templates. One or more embedded chemical product data templates may be numerical representations of one or more chemical product data templates. Selecting at least one chemical product data template based on at least a portion of the chemical product data may further include selecting at least one chemical product data template based on a distance score indicating the distance between the embedded chemical product data and the embedded chemical product data template. The embedded chemical product data and / or embedded chemical product data template may be tensors associated with the chemical product data and / or chemical product data template. The distance score may be determined by determining the tensor product of the tensors associated with the chemical product data and the chemical product data template, in particular by determining the vector product by, for example, cosine or Euclidean distance.

[0085] In one embodiment, providing chemical product data may include retrieving chemical product data by providing a request to retrieve chemical product data, and providing chemical product data in response to a request to retrieve chemical product data. Chemical product data may be provided and / or retrieved by a database. A request to retrieve chemical product data may include instructions for one or more chemical products. Instructions for one or more chemical products may include digital identifiers for identifying the chemical products. Digital identifiers may be related to numerical identifiers and / or text identifiers. Additionally or alternatively, instructions for one or more chemical products may be related to digital representations of the chemical products and / or the properties of the chemical products. Instructions for one or more chemical products may be provided via a user interface. Providing instructions for one or more chemical products may trigger the provision of a request to retrieve chemical product data. By retrieving chemical product data to determine environmental property data, it becomes possible to retrieve chemical product data in real time. Furthermore, chemical product data can be obtained directly from the source. Therefore, there is no need to establish any further data sources. This saves resources and reduces potential errors.

[0086] In one embodiment, the chemical product is • One or more material flows associated with the processing and / or production of at least one chemical product, and / or • One or more production and / or processing conditions associated with the processing and / or production of at least one chemical product, • Transportation data showing the transport of at least one chemical product and / or material for the production and / or processing of at least one chemical product. It can be associated with this.

[0087] The material flow may represent the material flow associated with one or more materials for producing and / or processing at least one chemical product. The production and / or processing conditions may include the consumption of resources for producing and / or processing at least one chemical product. Examples of resources may include water, energy, etc. The transportation of at least one chemical product and / or materials for producing and / or processing at least one chemical product is: - Transporting at least one chemical product and / or materials for producing and / or processing at least one chemical product to one or more locations for initiating the production and / or processing of materials, and / or • One or more transportation of materials for producing and / or processing at least one chemical product from one or more locations where they are processed and / or produced. It may include.

[0088] In one embodiment, the chemical product data may include two or more chemical product datasets. • One or more material flows associated with the processing and / or production of at least one chemical product, • One or more production and / or processing conditions associated with the processing and / or production of at least one chemical product, • Transportation data showing the transportation of at least one chemical product and / or materials for producing at least one chemical product, or a combination thereof. At least one of them may be different.

[0089] The generated environmental characteristics data may include at least one environmental characteristics dataset for each chemical product dataset. Two or more environmental characteristics datasets may be provided to monitor and / or control the processing and / or production of at least one chemical product. Furthermore, two or more environmental characteristics datasets may be provided to select at least one of the environmental characteristics datasets. This makes it possible to compare two different scenarios for producing and / or processing at least one chemical product.

[0090] In one embodiment, chemical product data may relate to the production and / or processing of two or more chemical products. The production and / or processing of two or more chemical products may differ in at least one of the two or more chemical products. The two or more chemical products are • Two or more material flows associated with the processing and / or production of two or more chemical products, • Two or more production and / or processing conditions associated with the processing and / or production of two or more chemical products, • Two or more transport datasets showing the transport of materials for the production of two or more chemical products, or combinations thereof. It can be associated with this.

[0091] The generated environmental characteristics data may include at least one environmental characteristics dataset for each chemical product. Two or more environmental characteristics datasets may be provided to monitor and / or control the processing and / or production of at least one chemical product. Furthermore, two or more environmental characteristics datasets may be provided to select at least one of the environmental characteristics datasets. This makes it possible to compare two different chemical products according to their environmental characteristics.

[0092] In one embodiment, providing chemical product data involves retrieving chemical product data from a data source by providing a request to a data source to receive chemical product data and receiving the chemical product data from the data source in response to the provision of the request. In this embodiment, the data source may be configured to provide the requested data upon receiving a request to receive the data. In this way, real-time data can be retrieved. This allows for retrieval of up-to-date data related to the production and / or processing of chemical products. This improves the reliability of chemical production and / or processing data generated based on chemical product data. Ultimately, this improves the production and / or processing of chemical products.

[0093] In one embodiment, any one of the methods may further include retrieving further chemical product data from a data source by providing a request to a data source for receiving further chemical product data and receiving chemical product data from the data source in response to the provision of the request. The data source may be configured to provide the requested data upon receiving a request to receive data. Further chemical product data may be related to chemical product data. For example, chemical product data may include a first portion of related data points, and / or further chemical product data may include a second portion of chemical product data, in particular a plurality of data points corresponding to the chemical product data. Contextualized chemical product data may be generated by combining at least a portion of chemical product data, at least a portion of further chemical product data, and at least one selected production and / or processing data template.

[0094] In one embodiment, a digital identifier relating to one or more data points associated with chemical product data and / or further chemical product data may be provided, for example, via a user interface and / or together with contextualized chemical production and / or processing data. The digital identifier may link to one or more data points associated with chemical product data and / or further chemical product data. For example, the digital identifier may include and / or may point to one or more datasets associated with a data source. The chemical product data and / or further chemical product data may be accessed based on the digital identifier. This allows access to the original data stored in a database or website to check the chemical production and / or processing data generated by the data-driven model. This makes it possible to evaluate the quality of the responses from the data-driven model. This gives the user complete control over where the data was acquired and how the data was processed by the data-driven model. As a result, the user can interact with the data-driven model when errors are detected. This improves the accuracy of the chemical production and / or processing data generated by the data-driven model. Ultimately, this improves the production and / or processing of chemical products.

[0095] In one embodiment, contextualized chemical product data may be suitable for triggering a data-driven model to generate chemical product production and / or processing data.

[0096] In one embodiment, environmental characteristics data may be provided for monitoring the production of chemical products and / or for producing chemical products. In particular, a method for generating chemical product production and / or processing data related to the production and / or processing of chemical products may be a method for producing and / or processing chemical products.

[0097] A brief explanation of some of the figures in the drawing. The present disclosure will be further described below with reference to the attached drawings. The drawings and the same reference numerals in this disclosure are intended to refer to the same or similar elements, components and / or parts. [Brief explanation of the drawing]

[0098] [Figure 1] This shows the production and processing of chemical products. [Figure 2] This illustrates one embodiment of chemical production. [Figure 3] This invention provides a system for generating environmental characteristics data related to the production and / or processing of chemical products, or for generating second chemical product data related to the processing and / or production of chemical products. [Figure 4A] One embodiment of one or more provided chemical product data templates 406 is shown. [Figure 4B] This document presents one embodiment for providing a chemical product data template. [Figure 5] One embodiment of a method for monitoring and / or controlling the production of chemical products is shown. [Figure 6] One embodiment of a method for monitoring and / or controlling the production of chemical products is shown. [Figure 7A] One embodiment of contextualized chemical product data 702 and environmental characteristics data 704 is shown. [Figure 7B] One embodiment of contextualized chemical product data 706, 710 and environmental characteristics data 708, 712 is shown. [Figure 8] One embodiment for training the embedding layer is shown. [Figure 9A] An embodiment of a transformer encoder structure is shown. [Figure 9B] An embodiment of the transformer decoder architecture is shown. [Figure 9C] This document shows one embodiment of a transformer encoder decoder architecture. [Figure 10]One embodiment for training and / or deploying a transformer encoder, transformer decoder, and / or transformer encoder decoder is shown. [Figure 11] One embodiment of input embedding is shown. [Figure 12] This shows one embodiment of the Mamba architecture. [Modes for carrying out the invention]

[0099] Detailed explanation The following embodiments are merely examples for implementing the methods, systems, or devices disclosed herein and should not be considered as limiting the invention.

[0100] Figure 1 illustrates the production and processing of chemical products. Chemical products may be produced by a chemical product producer 104. Typically, several steps and / or synthesis may be required to produce chemical products from recycled materials based on raw materials provided by an input material supplier 102 or chemical products provided by a recycling system operator 114. Thus, chemical products may be produced within a chemical production network including multiple plants. Chemical products may then be supplied to a chemical product user 106. Chemical product user 106 may process the chemical products to produce intermediate products, which are products based on the chemical products. Intermediate products may be further supplied and / or processed until an end product producer 108 can process one or more intermediate products to produce an end product. This end product may be based on the chemical products. The end product may be supplied to an end product user 110. The end product user 110 may be an end product consumer. When the end product becomes an end product, it may be supplied to a waste collection device and / or sorter 112. The end product may be separated based on multiple fractions to enable recycling. The fractions may be supplied to a recycling system operator 114. Nevertheless, end-of-life products and / or fractions thereof may be based on chemical products. Subsequently, the treatment of chemical product 118 may include all actions performed on chemical products, products based on chemical products, and waste and waste fractions based on chemical products.

[0101] Producing chemical products, intermediate products, and final products requires the processing of waste after the use of the final product, processing of waste fractions, recycling one or more waste fractions into raw materials, and / or obtaining raw materials, which in turn requires resources such as energy and water and releases carbon dioxide or equivalent climate-active substances. Therefore, all components of a supply chain can be associated with one or more environmental characteristics, and the environmental characteristic data associated with a component of a supply chain can depend on the environmental characteristic data associated with a previous component of the supply chain. Thus, a decrease in environmental characteristic data associated with chemical products, which are usually located upstream in the supply chain, has a significant impact on various supply chains, especially due to the resource-intensive production of chemical products. Furthermore, material recycling enables chemical product producers 104 to have predetermined and more favorable environmental characteristics associated with their chemical products.

[0102] Figure 2 shows one embodiment of chemical production. Chemical production may include one or more chemical production facilities 202. One or more chemical production facilities 202 may be configured to process raw materials 224 into one or more chemical products 204. One or more chemical production facilities 202 may be operated by an operating system 206. The operating system 206 may comprise a control and / or monitoring engine 208, a model engine 210, one or more data sources 214 containing chemical product data templates 218 and / or chemical product data 220, and an intake interface 212. The control and / or monitoring engine 208 may be configured to receive and / or provide machine-readable commands to one or more chemical production facilities 202. Furthermore, the control and / or monitoring engine 208 may be configured to monitor the production of one or more chemical products 204 by one or more chemical production facilities 202.

[0103] Environmental characteristics are important for monitoring and / or controlling one or more chemical production facilities 202. Environmental characteristics are a measure of resource investment and describe the impact of production on the environment. This may include workers and machinery within the work area, as well as the surrounding environment outside the work area. To enable the sustainable production of chemical products, it is necessary to monitor environmental characteristics and adapt production to those characteristics.

[0104] Environmental properties may not be tangible. Therefore, environmental properties may not be obtained through material analysis of chemical products. Consequently, environmental properties need to be monitored in other ways to enable control over the production of one or more chemical products 204 with individualized environmental impacts. Environmental properties can be obtained as described in this disclosure to monitor and / or control the production and / or processing of one or more chemical products 204. By providing contextualized chemical product data obtained by merging chemical product data associated with one or more chemical products with chemical product data templates related to the structure of at least some of the chemical product data, a data-driven model can focus on meaningful parts of the contextualized chemical product data to provide accurate environmental properties. Furthermore, similar searches of environmental property data can utilize the same chemical product data templates. This makes it possible to efficiently generate contextualized chemical product data. Moreover, since the user does not need to command the data-driven model but can efficiently command it via the chemical product data templates, improved human-machine interaction is achieved by deploying chemical product data templates to generate contextualized chemical product data.

[0105] In one embodiment, chemical product data may be provided via an intake interface 212, for example, a user interface. Additionally or alternatively, instructions regarding chemical products may be provided via the intake interface 212. Instructions regarding one or more chemical products may trigger environmental characteristic data 310 to retrieve chemical product data associated with the chemical products.

[0106] The control and / or monitoring engine 208 may be provided with environmental characteristics data by the model engine 210. The model engine 210 may be configured to receive chemical product data 220 and chemical product data templates 218. In particular, the model engine 210 may be configured to be provided with at least one selected chemical product data template. The at least one selected chemical product data template may be selected by determining at least one chemical product data template that corresponds to at least a portion of the chemical product data. Thus, the model engine 210 may be configured to receive chemical product data and, in response to the receipt of chemical product data and / or based on the chemical product data, select at least one chemical product data template. The model engine may be configured to generate contextualized chemical product data by merging at least a portion of the chemical product data with at least one selected chemical product data template. Furthermore, the model engine 210 may be configured to provide the contextualized chemical product data to a data-driven model that generates environmental characteristics data related to environmental characteristics associated with the processing and / or production of chemical products. The model engine 210 may be connected to the data-driven model, for example, via an interface. A data-driven model may be configured to provide environmental characteristics data in response to contextualized chemical product data. The data-driven model may be a pre-trained data-driven model. A pre-trained data-driven model may be trained on unstructured data, particularly text data. A pre-trained data-driven model may be available in the current state of the technology. A pre-trained data-driven model may operate by providing it with requests, such as contextualized chemical product data, via an interface. The model engine 210 may be provided with environmental characteristics generated from the chemical product data contextualized by the data-driven model, particularly via an interface to the data-driven model.The model engine 210 may be configured to provide environmental characteristics data to the control and / or monitoring engine 208.

[0107] Figure 3 shows a system for generating environmental characteristics data related to the processing and / or production of chemical products, or for generating second chemical product data.

[0108] The system may include a user interface 306 for receiving chemical product data 308 and / or providing environmental characteristics data 310. Furthermore, the user interface 306 may be configured to provide instructions for one or more chemical products. Chemical product data may be provided based on instructions for one or more chemical products, as described in the context of Figure 2. One or more chemical product data templates 406 may be received based on the chemical product data 308. Thus, one or more chemical product data templates 406 may be selected from multiple chemical product data templates or from a set of chemical product data templates. Multiple chemical product data templates may be contained by one or more databases 302. One or more chemical product data templates may be received from one or more databases 302, as described in the context of Figures 4A and 4B. Contextualized chemical product data may be generated based on receiving the chemical product data 308 and one or more chemical product data templates. Contextualized chemical product data may be provided to a data-driven model 304 via an application programming interface (API) 312. The data-driven model 304 may provide environmental characteristics data 310 based on receiving contextualized chemical product data. The environmental characteristics data 310 generated by the data-driven model 304 may be provided via API 314. API 312 may be the same API as 314 or a different API. The received chemical product data 308 may be provided via the user interface 306 in response to being received and / or provided via API 314.

[0109] Figure 4A shows one embodiment of one or more chemical product data templates 406 provided.

[0110] One or more chemical product data templates 406 may be provided by a database containing chemical product data templates 418. One or more data sources 214 may include and / or be database 302, as described with respect to Figure 2. In Figure 4A, chemical product data 402 may point to and / or link to one or more chemical product data templates 418. Specifically, chemical product data 402 may point to and / or link to at least one query associated with chemical product data 404. Preferably, the query associated with chemical product data 404 may point to and / or link to one or more chemical product data templates 418. The query associated with chemical product data 404 may be suitable for retrieving one or more chemical product data templates 418 from a database containing multiple chemical product data templates 406. Based on the query 404 associated with the chemical product data, one or more chemical product data templates 406 may be received. Therefore, receiving one or more chemical product data templates 406 may include providing a request to a database containing multiple chemical product data templates 418 to receive one or more chemical product data templates 406 containing queries associated with chemical product data 404. Furthermore, receiving one or more chemical product data templates 406 may also include providing one or more chemical product data templates 406 to the database based on the provision of a request to receive one or more chemical product data templates 406. This is advantageous because the computational resources required to receive one or more chemical product data templates 406 are low.

[0111] Figure 4B shows one embodiment for providing a chemical product data template. One or more chemical product data templates 412 may be provided by a database containing multiple chemical product data templates 414. One or more data sources 214 may include and / or be database 302, as described with respect to Figure 2. To provide one or more chemical product data templates 412, chemical product data 408 may be embedded via an embedding layer, as described in the context of Figure 8. By embedding the chemical product data 408, embedded chemical product data 410 may be obtained. The embedded chemical product data 410 may be a lower-dimensional representation of the chemical product data and may be machine-processable. This representation may similarly represent similar portions of the chemical product data 408. Chemical product data templates 414 may be embedded via an embedding layer, as described in the context of Figure 8. By embedding the chemical product data template 414, an embedded chemical product data template 416 may be obtained. The embedded chemical product data 410 and the embedded chemical product data template 416 may be compared in terms of similarity. The more similar the embedded chemical product data 410 and the embedded chemical product data template 416 are, the closer the tensors associated with the embedded chemical product data 410 and the embedded chemical product data template 416 may be. Therefore, to select one or more chemical product data templates 414 corresponding to the chemical product data 408, the embedded chemical product data template 416 and the embedded chemical product data 410 are compared, for example, by calculating the dot product between two or more tensors associated with the embedded chemical product data template 410 and the embedded chemical product data template 416. By selecting one or more chemical product data templates 412 from the chemical product data templates 414 through the similarity between the embedded chemical product data 410 and the embedded chemical product data template 416, one or more chemical product data templates 412 may be received, for example, based solely on their association with the chemical product data 408, and not on categories, as specified by a query associated with the chemical product data 404.Therefore, using embeddings to search for data from the database improves the accuracy of finding one or more chemical product data templates 412 that match the chemical product data 408.

[0112] Figure 5 shows one embodiment of a method for monitoring and / or controlling the production of chemical products. The method may include, for example, providing chemical product data via a user interface 502, as described in the context of Figure 3. Additionally or alternatively, chemical product data may be provided by one or more data sources 214, as described in the context of Figure 2.

[0113] Users can select chemical product data and / or input chemical product data through the user interface. For example, users can select chemical product data from production data acquired during the production of chemical products, in particular from historical production data of chemical products.

[0114] Furthermore, the chemical product data template may be received as described in the context of Figures 4A and 4B (504).

[0115] To provide contextualized chemical product data, two or more data-driven models may be available. These two or more data-driven models may include different model parameters that result in different outputs and / or different times associated with providing chemical product production and / or processing data. For example, a first data-driven model among the two or more data-driven models may include fewer model parameters than a second data-driven model among the two or more data-driven models. Therefore, the first data-driven model may provide chemical product production and / or processing data faster than the second data-driven model. On the other hand, the second data-driven model may provide more accurate chemical product production and / or processing data than the first data-driven model. Depending on the desired quality of output, a data-driven model may preferably be selected from among multiple data-driven models. For this purpose, a user selection of one or more data-driven models may be received, for example, via a user interface (506).

[0116] Furthermore, contextualized chemical product data may be generated by combining at least a portion of the chemical product data template with the chemical product data described within the context of Figure 6.

[0117] Contextualized chemical product data can be provided to a data-driven model for generating environmental characteristics data (510). The data-driven model can be trained and / or parameterized as described in the context of Figures 6 and 10. The data-driven model can process the contextualized chemical product data into environmental characteristics data as described in the context of Figures 6 to 11.

[0118] Environmental characteristics data may be provided via a user interface, for example, as described in the context of Figure 3, in response to receiving environmental characteristics data from a data-driven model (512).

[0119] Figure 6 shows one embodiment of a method for monitoring and / or controlling the production of chemical products. The received chemical product data 602 is used to generate contextualized chemical product data 606 based on the received chemical product data template. The chemical product data template may indicate a structure associated with at least a portion of the chemical product data 602. For example, the chemical product data template may specify one or more parts of one or more sequences and one or more production parameters adjacent to one or more parts of one or more sequences. The production parameters may be placeholders for the production parameter values ​​to be inserted. To generate contextualized chemical product data, at least a portion of the chemical product data, preferably production parameter values, may be inserted into the chemical product data template, including and / or relating to the chemical product data. The generated contextualized chemical product data may be input data to a data-driven model 610. The data-driven model may be and / or include one or more architectures, as described in the context of Figures 9A-9C and / or Figure 11. The data-driven model may be trained as described with respect to Figure 10.

[0120] The generated contextualized chemical product data 606 is provided to and / or received by the data-driven model 610. The data-driven model can be trained to process, in particular, embed the received contextualized chemical product data 606. For example, the contextualized chemical product data 606 may include one or more sequences, such as at least part of a sentence, one or more digits, or a combination thereof. At least part of a sentence may be received via a user interface, for example, in text format, audio format, etc.

[0121] To process the contextualized chemical product data 606 by the data-driven model 610, the data-driven model 610 can convert the contextualized chemical product data 606 into embedded contextualized chemical product data 618. The embedded contextualized chemical product data 618 may be a second-order tensor. The embedded contextualized chemical product data 618 may and / or may represent the contextualized chemical product data 606 in a machine-processable format. If the contextualized chemical product data 606 may contain at least part of a sentence, words, digits, and / or parts of words associated with the contextualized chemical product data 606 may be embedded. Words, digits, and / or parts of words associated with the contextualized chemical product data 606 may be examples of elements associated with the input data. A further example of elements associated with the input data may be digits. The embedded contextualized chemical product data 618 may contain one or more data points. Embedding the contextualized chemical product data 606 enables computational processing of the contextualized chemical product data 606. Furthermore, embeddings, especially word embeddings, are efficient representations of data that use less storage and require fewer computational resources to process each piece of data.

[0122] In particular, if the contextualized chemical product data 606 may contain at least a portion of the sequence, the positions of one or more elements associated with the contextualized chemical product data 606 may be considered as described within the context of Figures 9A to 9C. By applying position coding to the contextualized chemical product data 606, sequence-specific embedded contextualized chemical product data 618 can be obtained.

[0123] In one embodiment, contextualized chemical product data 606 can be embedded in parts. For example, a first part of contextualized chemical product data 606 may be embedded to obtain a first embedded contextualized chemical product data 618, and a second part of contextualized chemical product data 606 may be embedded to obtain a second embedded contextualized chemical product data 618. The first embedded contextualized chemical product data 618 and the second embedded contextualized chemical product data 618 can be combined to obtain embedded contextualized chemical product data 618. The first embedded contextualized chemical product data 618 and the second embedded contextualized chemical product data 618 can be joined by concatenating a first vector associated with the first embedded contextualized chemical product data 618 and a second vector associated with the second embedded contextualized chemical product data 618. By concatenating the first embedded contextualized chemical product data 618 and the second embedded contextualized chemical product data 618, embedded contextualized chemical product data 618 can be obtained. Additionally or alternatively, the first embedded contextualized chemical product data and the second embedded contextualized chemical product data may be embedded together in the second rank tensor.

[0124] The embedded contextualized chemical product data 618 may be a second-order tensor representing the contextualized chemical product data 606. The second-order tensor may be combined with itself and / or transformed by a tensor product to yield a chemical product data context tensor 620, as described in the context of Figures 9A-9C and Figure 10. The chemical product data context tensor 620 may contain a second-order tensor and / or represent relationships between one or more elements associated with the contextualized chemical product data and / or the chemical product data template.

[0125] If the contextualized chemical product data 606 is sentence-based, the chemical product data context tensor 620 may represent the words associated with the chemical product data context tensor 620 and the relationships between those words. Similarly, in the case of numerical input, as illustrated in the context of Figure 11. By doing so, the relationships between two or more components of the contextualized chemical product data are taken into consideration when generating chemical product production and / or processing data. This allows for more accurate determination of chemical product production and / or processing data.

[0126] The chemical product data context tensor 620 can be transformed into environmental characteristics data 612 via one or more machine learning architectures, such as a linear layer and a classification layer. In particular, the first portion of the environmental characteristics data 612, referred to herein as the first environmental characteristics data 612, can be generated based on the embedded contextualized chemical product data 618. As illustrated in the context of Figure 10, the second portion of the environmental characteristics data 612, the second chemical product production and / or processing data, in particular the second portion of sequential chemical product production and / or processing data, can be generated and / or provided based on the embedded contextualized chemical product data 618 and the embedded first chemical product production and / or processing data. The embedded first chemical product production and / or processing data may refer to the first chemical product production and / or processing data which is embedded in the same way as the embedded contextualized chemical product data 618.

[0127] The embedded contextualized chemical product data 618 and the embedded first chemical product production and / or processing data can be combined, for example, by concatenation to obtain a chemical product data context tensor 620 in a second time step for generating second chemical product production and / or processing data. The second chemical product production and / or processing data can be generated in a second time step based on the chemical product data context tensor 620. This can be repeated for the remainder of the chemical product production and / or processing data. If the chemical product production and / or processing data and / or chemical product data can be sentence-based, then parts of the chemical product production and / or processing data, chemical product data, and / or contextualized chemical product data may refer to words and / or parts of words, in particular tokens associated with words. Parts of the chemical product production and / or processing data can be generated based on previously generated parts of the chemical product production and / or processing data until complete chemical product production and / or processing data is generated, typically indicated by a termination token.

[0128] In one embodiment, embedded contextualized chemical product data 618 can be transformed into a first chemical product data context tensor 620 and a second chemical product data context tensor 620. The first chemical product data context tensor 620 and the second chemical product data context tensor 620 can be joined, for example, by concatenation to form a chemical product data context tensor 620. The chemical product data context tensor 620 can be processed as described above to yield environmental characteristic data 612. By taking into account two or more sets of chemical product data context tensors 620, the computational resources and / or time required to generate the environmental characteristic data 612 are reduced.

[0129] The data-driven model 610 may include one or more of the following: a transformer decoder, a transformer encoder, a transformer encoder decoder architecture, etc. This data-driven model 610 may be self-supervised based on a training dataset. The training dataset may include historical data related to chemical product production data and environmental characteristics data. The data-driven model may be trained to sequentially generate environmental characteristics data 612 based on chemical product data 602 and previously generated portions of chemical product production and / or processing data. For this purpose, the data-driven model may be provided by contextualized chemical product data 606 to generate a first portion of chemical product production and / or processing data, and by contextualized chemical product data 606 and a first portion of chemical product production and / or processing data as specified by the training dataset. During training, the data-driven model may be provided with ground truth of previous portions of chemical product production and / or processing data as specified by the training dataset. This self-supervised training enables human-independent learning of the data-driven model, resulting in efficient embedding that does not require human interaction. Therefore, time is saved by providing efficient learning while simultaneously achieving highly accurate results associated with the environmental characteristics data 612.

[0130] The generated environmental characteristics data may be provided, for example, by a data-driven model as described in the context of Figure 2, and / or via the model engine 210, and / or provided to the control and / or monitoring engine 208, for monitoring and / or controlling the processing and / or production of chemical products.

[0131] Figure 7A shows one embodiment of contextualized chemical product data 702 and environmental characteristics data 704.

[0132] Users may wish to determine the carbon footprint, recycled content, bio-based content, renewable content, or environmental aspects, and social and / or corporate governance (ESG) data is an example of the environmental characteristics of chemical product A by providing chemical product data through the user interface. Furthermore, environmental characteristics may include and / or refer to measures of carbon footprint, water consumption, waste generation, energy efficiency, chemical use and emissions, resource consumption, supply chain emissions, water quality, worker safety, local impact, recycling and reuse, supply chain transparency, or a combination thereof. Measures of carbon footprint may measure the amount of gas emissions produced during the production stage, particularly greenhouse gas emissions, including emissions from energy consumption, raw material extraction, and transport. Measures of water consumption may measure the amount of water used during the production stage, including water used for treatment, washing, and cooling. Measures of waste generation may measure the amount of waste generated during the production stage, including hazardous and non-hazardous waste. Measures of energy efficiency may measure the amount of energy used per unit of production output, indicating the efficiency of energy use during the production stage. The material efficiency measure can measure the amount of material used per unit of production output, indicating the efficiency of material use in the production stage. The chemical use and emission measure can measure the amount of chemicals used and emitted during the production stage, including chemicals that may have adverse effects on the environment. The resource consumption measure can measure the amount of natural resources used in the production stage, including wood, metals, and minerals. The supply chain emission measure can measure greenhouse gas emissions associated with the production stage, including emissions from the transport, storage, and distribution of raw materials and finished products. The water quality measure can measure the impact of the production stage on water quality, including the amount of pollutants released into waterways. The worker safety measure can measure the impact of the production stage on worker safety, including exposure to hazardous chemicals and materials. The community impact measure can measure the impact of the production stage on the local community, including impacts on air quality, noise pollution, and socioeconomic well-being. The recycling and reuse measure can measure the amount of waste recycled or reused during the production stage, indicating the production stage's contribution to the circular economy.Measures of supply chain transparency can measure the level of transparency and accountability in the supply chain, including raw material traceability, the use of conflict minerals, and the presence of forced labor.

[0133] A corresponding chemical product data template may be received, and contextualized chemical product data 702 may be obtained by inserting chemical product data into the chemical product data template. Environmental characteristic data 704 may be generated by a data-driven model in response to the provision of contextualized chemical product data. For example, the environmental characteristic data may be "1.49 kg of carbon dioxide equivalent per kilogram of chemical product A".

[0134] Figure 7B shows one embodiment of contextualized chemical product data 706, 710 and environmental characteristics data 708, 712.

[0135] A user may wish to determine chemical product production and / or processing data through a user interface. Therefore, chemical product data may be received. A corresponding chemical product data template may also be received. Contextualized chemical product data may be generated by inserting chemical product data into the chemical product data template. Environmental characteristics data 708 may be generated by the data-driven model in response to the provision of contextualized chemical product data. The user may desire further information and provide second contextualized chemical product data 710. Based on the provision of the second contextualized chemical product data 710, the data-driven model may provide chemical product production and / or processing data 712. Chemical product production and / or processing data 712 may specify changes to production orders related to the production of chemical products. Therefore, chemical product production and / or processing data 712 may be provided to the analysis engine as specified in Figure 2. Environmental characteristics data 708 may be provided to the data-driven model to generate the chemical product production and / or processing data 712. Therefore, the data-driven model can generate chemical product production and / or processing data 712 based on environmental characteristics data 708.

[0136] Figure 8 shows one embodiment of obtaining the embedded layer. The embedding layer can be obtained, for example, by training a continuous bag-of-word model (CBOW) or a skipgram model. The embedding layer may be suitable for generating embedding input data based on the input data. Generating embedded input data can refer to embedding the input data. By embedding the input data, a representation associated with the input data can be obtained. Thus, the embedding input 814 may be a representation associated with the input data. The input data may contain one or more elements. One or more elements may be represented by the input vector 806. In particular, the embedding input 814 and / or the input vector 806 may be machine-readable and / or processable by a processor. For this purpose, the embedding input 814 and / or the input vector 806 may be a tensor, in particular a first-order tensor. Specifically, the input vector 806 may be a one-hot vector or a sum of several one-hot vectors. A one-hot vector may be a vector with one entry that is not equal to zero. Examples of one-hot vectors may be 808, 810, and 812. Non-zero entries in the one-hot vector and / or input vector 806 can represent elements. For example, a lookup table can define the relationship between the position of a non-zero entry and the element represented by the one-hot vector. A lookup table can specify multiple distinct elements. The number of distinct elements may be equal to the number of entries in the one-hot vector. The number of distinct elements may be called the vocabulary size. In one example, an element may be represented by tokens, and a sequence of elements may refer to at least part of a sentence. At least part of a sentence may be represented by multiple tokens. Tokens may represent elements and / or at least part of a word. For example, if one element is associated with only one word, words such as "embeddings," "embedding," or "embed" constitute distinct elements. The first token may represent the stem "embed," and the suffix, which typically appears in multiple words, may be represented by a second, third, and fourth token.The second, third, and fourth tokens may be used, preferably together with a fifth token representing the stem "look," to represent other words such as "look" and "looking." Ultimately, this tokenization of elements associated with multiple stems and multiple suffixes reduces the number of tokens used to represent multiple elements, and therefore reduces the computational resources used.

[0137] For example, a lookup table specifying a subset of the English vocabulary size may contain more than 10,000 words. An embedding input 814 can be a lower-dimensional representation than an input vector 806. For example, a typical embedding input 814 may contain hundreds of different entries. Thus, an embedding input 814 can constitute a high-density representation of one or more elements using fewer computational resources. Furthermore, an embedding input 814 can represent a relationship between two or more elements. For example, the words "Italy" and "Germany" may be similar or more closely related, as both define European countries, while the word "embodiment" may be very different from each of the two words. The smaller the dot product between two embedding inputs 814, the more similar the two elements associated with the embedding input 814 may be. Thus, an embedding input 814 can accurately represent one or more elements, and processing the embedding input 814 can lead to accurate results.

[0138] To transform the input vector 806 into the embedding input 814, the embedding layer may contain a number of neurons equal to the number of entries in the embedding input 814. Based on the embedding input 814, the output layer may generate the output vector 816. The output vector may be a vector and / or represent one or more elements. The output vector 816 may represent one or more elements different from the input vector 806 and / or the one-hot vector associated with the input vector 806. For this purpose, the output layer may contain a number of neurons equal to the number of entries in the input vector 806 and / or the output vector 816. The output layer may apply a softmax function to the embedding input 814. By doing so, the output vector may contain probabilities related to the elements associated with the non-zero entries in the output vector 816. Thus, one or more elements may be obtained from the output vector 816 with corresponding probabilities. If the input vector 806 may specify one or more element sequences, the output vector 816 may specify one or more elements corresponding to the sequence of elements specified by the input vector 806. In the example in Figure 8, the element associated with vector 818 can correspond to the input vector with a 71% probability. Additional or substitute elements can correspond to the input vector with a lower probability, as indicated by the output vector. By defining thresholds for comparing probabilities, the selection of corresponding elements can be tailored to user needs. The elements generated by the model, including the embedding layer 802 and the output layer 804, can point to the most likely element indicated by the output vector 816. Thus, the model shown in Figure 8 can generate the element associated with vector 818 with a 71% confidence score.

[0139] The model in Figure 8 may be a continuous word bag (CBOW) model. A CBOW model can be trained on a training dataset containing multiple input vectors and corresponding output vectors. Since the training dataset may not be labeled, training a CBOW model may be called self-supervised. Before training a CBOW model, it may be initialized by assigning random values ​​to the weights of its neurons. During training of the CBOW model, input vectors may pass through the initialized embedding and output layers, and the loss may be determined by comparing the output vector obtained by passing the input vector 806 through the model with the output vector corresponding to the input vector 806 specified by the training dataset. Based on the determined loss, backpropagation may be applied to determine the gradients associated with the neurons in the embedding layer 802 and output layer 804 to reduce the loss. According to the determined gradients, the weights of the neurons may be updated by using a gradient descent algorithm. If the CBOW model can achieve a given loss, 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 containing one or more elements. This embedding layer 802 can be used in other machine learning architectures that require the embedding layer 802, such as transformer encoders, transformer decoders, or transformer encoder decoder architectures, as described in the context of Figures 9A, 9B, and 9C. A trained embedding layer 802 may be required to train these architectures. Therefore, a model such as a CBOW model can be trained before training a transformer encoder, transformer decoder, or transformer encoder decoder architecture.

[0140] Figure 9A shows one embodiment of a transformer encoder kit. The transformer encoder comprises an encoder input 978, one or more encoder blocks 974, 914, and an encoder output. Transformer encoder kits are known in the art and can be derived from transformer encoder decoder architectures such as the one shown in Figure 9C. In particular, a transformer encoder may be called an X-former. Instead of directly connecting the encoder blocks to the decoder of a transformer encoder decoder architecture, the transformer encoder kit may correspond to an encoder architecture associated with a transformer encoder decoder architecture that has additional encoder outputs. Several transformer encoder kits are available in the art, such as bidirectional encoder representations from a transformer (BERT).

[0141] Input data may be received at encoder input 978. The input data may be contextualized chemical product data. Encoder input 978 may have input embedding 902 applied. Applying input embedding 902 may mean passing the input data through the embedding layer, for example, as described in the context of Figure 8. By applying input embedding 902 to contextualized chemical product data, embedded contextualized chemical product data may be obtained.

[0142] The encoder input 978 may have position coding 904 applied to it. Applying position coding 904 may mean adding a position coefficient to the embedding input obtained via the input embedding. Preferably, the input data may specify a sequence of multiple elements. The position coefficient Ppos may indicate the position of an element in the sequence. For example, the position coefficient Ppos may be obtained based on the following formula:

[0143]

number

[0144] If pos can refer to the position of an element in the sequence, i can refer to the dimension associated with the input embedding, and d can refer to the dimension of the model, e.g., transformer decoder, transformer encoder, or transformer encoder decoder. This may be called absolute position embedding. Alternatively, position coding may be based on rotational position embedding (RoPE). Position coding is beneficial because it enables processing of continuous data without requiring further dimensions to indicate the position of each element. Thus, position coding 904 reduces the computational resources required to embed the input data. By passing the input data to the encoder input, the input data can be transformed into a second-order tensor representing the sequence of elements. This second-order tensor may be called the embedding input data. The embedding input data may be processed by the encoder block. The embedding input data may be provided to layer normalization 908 by residual connection. Multi-head self-attention 906 may be applied to the embedding input data. Multi-head self-attention 906 may include two components: multi-head and self-attention. Self-attention can be understood as a filter applied to the embedding input data. By applying a filter to embedded input data, elements associated with the embedded input data that contribute to the generated output data can be identified for generating the output data. Thus, the filter can represent the degree to which elements associated with the embedded input data contribute to the generated output data. Applying a filter can be called weighting the elements associated with the embedded input data. This is particularly advantageous for long element sequences. Filters can be learned and improved during training by learning to identify the contributions of elements associated with the embedded input data. For example, in the subsentence "I went to the bakery to buy a," the last word can be generated by a data-driven model such as a transformer encoder.Self-attention can focus the transformer encoder on the words "bakery" and "buy" and, in most cases, generate the word "bread." Self-attention can refer to attention generated based on input data. Therefore, filters can be determined based on input data, preferably embedded input data. Embedded input data can function as a query Q, key K, and value V for the self-attention operation. Self-attention can refer to attention based on received input data. Therefore, filters can be calculated based on the following formula by inserting the respective tensors based on the embedded input data.

[0145]

number

[0146] In the formula, d K This corresponds to the key dimension. To further improve the efficiency of the transformer encoder, filters are applied using multiple heads, resulting in a multi-head self-attention 906. The multi-head self-attention 906 may involve applying filters to two or more parts of the embedded input data. Thus, a tensor may be divided into two or more parts, and filters may be applied separately to these two or more parts by two or more heads according to the following equation: Head i = Attention (QW i Q ,KW i K VW i V ) Here, the parameter matrix

[0147]

number

[0148] And, however, i may refer to the number of heads, d V dK and d Q may refer to dimensions of values, keys, and queries.

[0149] The results of two or more heads may be concatenated according to the following formula: Multi - head(Q, K, V)=Concat(head1, …, head h )W 0 where

[0150]

Number

[0151] and h may refer to the number of heads. The embedded input data can be transformed into a context tensor via multi-head self-attention 906. The context tensor can represent a sequence of elements and relationships between two or more elements of the input data. Transforming the embedded contextualized chemical product data may result in a chemical product and environmental properties data context tensor 622. Transforming the embedded contextualized chemical product data may yield a chemical product data context tensor 620. Therefore, the context tensor can be the chemical product data context tensor 620 and / or the chemical product and environmental properties data context tensor 622. The context tensor may be a second-order tensor and / or may contain one or more first-order tensors. After multi-head self-attention 906, layer normalization 908 may be applied based on the context tensor and / or the embedded input data from residual connections. Applying layer normalization 908 may mean normalizing the context tensor. Normalizing the context tensor can reduce the values ​​of the entries in the context tensor. This reduces the computational cost associated with processing the context tensor. After layer normalization 908, the context tensor can be passed back to the feedforward layer 910, followed by layer normalization 912 based on the context tensor and / or residual connections to the output of the feedforward layer 910. The feedforward layer 910 may be a feedforward neural network. A feedforward neural network may contain multiple fully connected neurons. By passing the context tensor to the feedforward neural network, the context tensor can be linearly transformed. Additionally or alternatively, the neural network may include one or more activation functions, such as a normalized linear unit (ReLU). Thus, the neural network may be configured to perform one or more nonlinear operations on the context tensor and / or to nonlinearly transform the context tensor.After the context tensor has been transformed and / or normalized by the feedforward layer 910 and layer normalization 912, the context tensor may be provided to one or more further encoder blocks 914. Passing the context tensor through the feedforward layer 910 allows it to be adapted for processing by further attention layers of one or more further encoder blocks 914 in order to apply a self-attention filter, preferably a multi-head self-attention filter 906. The context vector after being transformed by layer normalization 912 and the feedforward layer 910 may be called the hidden state.

[0152] The encoder output 976 comprises a linear layer 916 and a softmax layer 918. The linear layer 916 can convert a context vector into a logit vector. The linear layer may be fully connected. The logit vector obtained by passing the context tensor through the linear layer 916 can be passed through the softmax layer 918. Passing the logit vector to the softmax layer 918 may mean applying a softmax function to the logit vector. Applying a softmax function to the logit vector may yield a probability distribution of one or more elements corresponding to a sequence of elements in the input data. Based on predefined selection criteria, one or more elements may be selected from the probability distribution. The one or more selected elements may be called the one or more elements generated by the transformer encoder. The one or more generated elements may be provided to the encoder input to generate one or more further elements corresponding to a sequence of elements in the input data, and one or more elements generated by the transformer encoder as described in the context of Figure 10.

[0153] The output data from encoder output 976 may be chemical product production and / or processing data. Therefore, the result of transforming the chemical product data context tensor 620 may also be chemical product production and / or processing data.

[0154] Figure 9B shows one embodiment of the transformer decoder architecture. The input data, embedded input data, context tensor, and / or output data may be as defined in the context of Figure 9A.

[0155] The transformer decoder comprises a decoder input 984, one or more decoder blocks 980, 932, and a decoder output 992. The transformer decoder architecture is known in the art and can be derived from transformer encoder-decoder architectures such as the one shown in Figure 9C. The transformer decoder may be referred to as an X-former. The transformer decoder architecture may correspond to a decoder architecture associated with the transformer encoder-decoder architecture, independently of receiving one or more hidden states from the encoder of the transformer encoder-decoder. Several transformer decoder architectures, such as the Generative Pre-trained Transformer (GPT®), are available in the art.

[0156] Decoder input 984 can be subjected to input embedding 920 and position coding 922, as well as input embedding 902 and position coding 904, as described in the context of Figure 9A.

[0157] The decoder block 980 may include a layer normalization 926, a masked multi-head self-attention 924, a feedforward layer 928, and / or a layer normalization 930. Embedded input data obtained by passing input data through the decoder input 984 may be provided to the layer normalization 926 via a residual connection. Furthermore, a masked multi-head self-attention 924 may be applied to the embedded input data. The masked multi-head self-attention 924 corresponds to the multi-head self-attention 906 as described in the context of Figure 9A, and additionally masks a portion of the embedded input data associated with elements later in the sequence than the elements being generated. Additionally or alternatively, a portion of the input data associated with elements later in the sequence than the elements being generated may not be received and / or may not be converted into embedded input data. Thus, a transformer decoder may be suitable for generating elements following a sequence, while a transformer encoder may be suitable for generating missing elements within one sequence and / or between two or more sequences. Thus, a transformer encoder may be configured for classification tasks. A transformer decoder can be configured for text generation.

[0158] Similar to the transformer encoder described in the context of Figure 9A, a context tensor can be generated by applying a masked multi-head self-attention 924 and layer normalization 926. The context tensor can be provided to layer normalization 930 via residual connections. Furthermore, the feedforward layer 928 and layer normalization 930 may be analogous to the feedforward layer 910 and layer normalization 912 described in the context of Figure 9A. The context tensor can be provided to one or more further decoder blocks 932.

[0159] The decoder output 992 may include a linear layer 934 and a softmax layer 936. The linear layer 934 and softmax layer 936 may be analogous to the linear layer 916 and softmax layer 918 described in the context of Figure 9A.

[0160] Figure 9C shows one embodiment of a transformer encoder decoder architecture. The input data, embedded input data, context tensor, and / or output data may be as defined in the context of Figure 9A.

[0161] The transformer encoder decoder may include an encoder input 988, one or more encoder blocks 986, 964, a decoder input 994, a decoder block 990, and a decoder output 992. Encoder input 988 may correspond to encoder input 978 in Figure 9A. One or more encoder blocks 986, 964 may correspond to one or more encoder blocks 974, 914 in Figure 9A. Decoder input 994 may correspond to decoder input 984 in Figure 9B.

[0162] Decoder block 990 may include masked multi-head self-attention 970, layer normalization 972, feedforward layer 938, and layer normalization 940, similar to the masked multi-head self-attention 924, layer normalization 926, feedforward layer 928, and layer normalization 930 described in the context of Figure 9B. Decoder block 990 may further include multi-head self-attention 950 and layer normalization 948. As described in Figure 9B, the context tensor can be obtained from the masked multi-head self-attention 970 and layer normalization 972. Multi-head self-attention 950, similar to the multi-head self-attention 906 in Figure 9A, may be applied to the context vector obtained from layer normalization 972 and the hidden states of one or more encoder blocks 986, 964. Layer normalization 948 may be applied to the context vector obtained from multi-head self-attention 950 and the context vector obtained from layer normalization 972 provided via residual connections. The context vector obtained from layer normalization 948 can be processed via feedforward layer 938 and layer normalization 940, as described in Figure 9B. The context vector obtained from layer normalization 940 can be provided to a further decoder block 942, as well as to decoder block 990. Context vectors obtained from one or more decoder blocks 990, 942 can be provided to decoder output 992. Decoder output 992 may correspond to decoder output 982 in Figure 9B.

[0163] In the architecture described above, the transformer encoder decoder can receive and process input data at encoder input 988, one or more encoder blocks 986, 964, decoder block 990, and decoder output 992. Based on the input data, the transformer encoder decoder can partially or sequentially generate output data. The sequentially generated output data can be provided to and / or processed by decoder input 994, one or more decoder blocks 990, 942, and decoder output 992. Preferably, the sequence can be provided to encoder input 988 to generate at least a portion of the output data, and then at least a portion of the already generated output data can be provided to decoder input 994. Since more data can be received over time by the transformer encoder decoder, the next element of the output data can be generated with greater precision by taking into account the input data and the generated output data.

[0164] A transformer encoder-decoder architecture allows a transformer encoder-decoder to be configured to convert a sequence into another representation of the sequence. One example of converting a sequence into another representation is translating a sentence into another language. Several transformer encoder-decoders, such as BART and T5, are available in the art.

[0165] In one embodiment, layer normalization 908, 912 may be applied before the multi-head self-attention 924, multi-head self-attention 906, and / or feedforward layer 910 that are masked in the transformer decoder, transformer encoder, and / or transformer encoder decoder. By doing so, the computational resources required to apply the multi-head self-attention 906 and / or feedforward layer 910 to the embedded input data and / or context tensor may be reduced because the entries in each tensor may be lower after normalization.

[0166] In one embodiment, the decoder output 992 may include a classification neural network, further feedforward layers, convolutional layers, fully connected layers, and so on. For example, the transformer encoder decoder may be configured to select from multiple options. For this purpose, the transformer encoder decoder may be provided with three different input datasets, and the transformer encoder decoder may classify context vectors obtained from one or more decoder blocks 990 via one or more linear layers. Thus, the architecture may be extended depending on the use case to be solved.

[0167] Figure 10 shows one embodiment of training and / or deploying a transformer encoder, a transformer decoder, and / or a transformer encoder decoder.

[0168] The encoder / decoder / encoder-decoder architecture 1002 may correspond to transformer decoders, transformer encoders, and / or transformer encoder decoders as described in the context of Figures 9A to 9C. The input data, embedded input data, context tensor, and / or output data may be as defined in the context of Figure 9A.

[0169] The output data generated by the encoder / decoder / encoder-decoder architecture 1002 may contain one or more elements, in particular a sequence of elements. Previously generated elements of the output data may be provided as input for generating the next element in the sequence of output data.

[0170] In the example in Figure 10, the input data may include N elements, in particular input tokens. The input tokens may be tokens specifically for input to data-driven models such as transformer decoders, transformer encoders, or transformer encoder decoders. The generated output data may include M elements. The encoder / decoder / encoder-decoder architecture 1002 may generate one element of the output data based on receiving the input data and optionally previously generated elements of the output data in a time step. Therefore, M time steps are required to generate M elements. A time step includes providing inputs 1010, 1012, and 1014 to the encoder / decoder / encoder-decoder architecture 1002 and receiving output data 1004, 1008, and 1006 from the encoder / decoder / encoder-decoder architecture 1002. In the first time step, input 1010 may include N input tokens. The N input tokens may be associated with, for example, N words, stems, or suffixes. Preferably, the N input tokens may specify a question. One or more input tokens may specify the beginning and / or end of a sequence of tokens. Input 1010 may be processed by an encoder / decoder / encoder-decoder architecture 1002. Based on input 1010, at least a portion of output data 1004 may be generated. At least a portion of the output data may include a first output token. In the next time step, the generated first output token may be provided together with input 1012. Specifically, if input 1012 may be received by a transformer encoder decoder, the input token may be received at encoder input 988, and the first output token may be received at decoder input 994. If input 1012 may be received by a transformer encoder, input 1012 may be received by encoder input 978, and similarly with respect to the transformer decoder and decoder input 984. Based on input 1012, output data 1008 including the first and second output tokens may be generated.Generating output data 1008 based on input 1012 may mean generating a first token and a second token based on N input tokens, where the first token may be generated based on N input tokens. This process may be repeated until the last token in the sequence of output data 1006 can be generated. Preferably, the last token may be a termination token. The termination token can terminate the generation of further output tokens.

[0171] Similar to data processing during the deployment of the encoder / decoder / encoder-decoder architecture 1002, the encoder / decoder / encoder-decoder architecture 1002 can be trained. The training dataset may include multiple sequences containing multiple elements. The sequences may be associated with input data and / or output data. Additionally or alternatively, the sequences may be independent of the input data and / or output data. For example, if the input and output data may refer to chemical compositions represented via text, the training dataset may include sequential text data unrelated to chemical compositions. In this example, the training dataset may include a sequence of words resulting from a conversation. In one embodiment, the training dataset may include, at least partially, the input dataset and / or output dataset.

[0172] Training can be initiated by initializing the encoder / decoder / encoder-decoder architecture 1002. In one embodiment, the parameters associated with the encoder / decoder / encoder-decoder architecture 1002 may be initialized randomly. Additionally or alternatively, the input embeddings of the encoder / decoder / encoder-decoder architecture 1002 may be obtained by training a CBOW model or a skipgram model, as described in the context of Figure 8. The trained embedding layers may be used during training. The parameters associated with the embedding layers may be kept constant and / or updated after a predetermined number of training epochs. Doing so allows for a smaller number of parameters to be updated, enabling faster training with less computational resource consumption. Furthermore, the accuracy associated with the embedding layers may be constant and / or improved by avoiding error compensation with respect to a just-initialized encoder / decoder / encoder architecture 1002.

[0173] During training of the encoder / decoder / encoder-decoder architecture 1002, at least a portion of the sequence of the training dataset may be provided to the encoder / decoder / encoder-decoder architecture 1002 one by one, and one or more elements may be generated one by one based on the sequence of the training dataset. The elements generated based on the sequence may follow elements of the portion of the sequence provided to the encoder / decoder / encoder-decoder architecture 1002. The generated one or more elements may be compared to one or more elements following at least a portion of the sequence provided to the encoder / decoder / encoder-decoder architecture 1002, as specified by the training dataset. Thus, during training, the encoder / decoder / encoder-decoder architecture 1002 may generate inferences about the next element, and the inferences about the next element in the sequence may be compared to ground truth, which specifies the actual next element according to the training dataset. Based on the inferences about the next element and the ground truth, the loss can be determined. The loss may define the similarity between the inferences about the next element and the ground truth. The loss may be determined by forming a vector dot product between tokens associated with one or more elements and tokens associated with the ground truth. Losses that are not equal to zero may result in updates to parameters associated with the encoder / decoder / encoder-decoder architecture 1002. Preferably, the parameters associated with the encoder / decoder / encoder-decoder architecture 1002 may be independent of the embedding layer. For example, the parameters associated with the encoder / decoder / encoder-decoder architecture 1002 may be the weights of the neurons in the encoder / decoder / encoder-decoder architecture 1002.

[0174] Based on the determined loss, backpropagation can be applied to determine the gradient associated with the parameters of the encoder / decoder / encoder-decoder architecture 1002 in order to reduce the loss. According to the determined gradient, the parameters associated with the encoder / decoder / encoder-decoder architecture 1002, preferably the weights of the neurons associated with the encoder / decoder / encoder-decoder architecture 1002, can be updated by using a gradient descent algorithm.

[0175] The training dataset does not need to be labeled. The sequence of elements in the training dataset can, in essence, contain the ground truth for determining the loss for one or more elements generated during the training of the encoder / decoder / encoder architecture 1002. Thus, the encoder / decoder / encoder-decoder architecture 1002 can be a pre-trained self-supervised type. This is advantageous because it can save time and resources in creating a labeled training dataset. Furthermore, this allows for the use of large training datasets, associated with sizes of several terabytes. As a result, the data-driven model can be accurate in generating elements in a sequence. In addition, large training datasets can even eliminate shot predictions little to no. Thus, a data-driven model trained as described above can be used for multiple purposes, contributing to the limitation of resources required to train and / or host multiple purpose-driven models such as CNNs. The training described above can be called pre-training. The data-driven model can be configured to eliminate little to no shot predictions to perform for multiple use cases after pre-training. The performance of the data-driven model can be further improved by additional training called fine-tuning.

[0176] Figure 11 shows one embodiment of input embedding. The input data, embedding input data, context tensor, and / or output data may be as defined in the context of Figure 9A.

[0177] Preferably, if the sequence of elements associated with the input data contained in the input data can be of one type, then input embeddings 902, 920, 952, and 966, as described in the context of Figures 9A to 2C, may be used. For example, the type of input data may be text, in which case elements may be associated with at least part of a word, punctuation, a start token specifying the beginning of one or more sequences associated with the input data, and / or an end token. In another example, the input data may be at least partially numeric. Thus, the input data may contain multiple digits. Numeric input data may be, for example, tabular data. Tabular data may specify one or more rows and / or one or more columns. Thus, tabular data may contain one or more cells, and cells may be associated with one or more numbers.

[0178] Numeric input data may require different embedding than text input data. Input embedding for numeric input data may include token embedding, positional embedding, column embedding, row embedding, or a combination thereof.

[0179] By applying token embedding to one or more elements, particularly tokens associated with input data, a machine-readable representation of one or more elements, especially tokens, can be obtained. Applying token embedding to one or more elements may mean passing one or more elements through an embedding layer, as illustrated, for example, in the context of Figure 8. Thus, token embedding can specify one or more elements, particularly tokens, in a machine-readable representation. For example, token embedding can convert numerical values ​​into vectors. This representation is advantageous because it can be enhanced by further information, such as the location of tokens in a sequence and / or in a table associated with a sequence of tokens. Location embedding may be similar to location embedding illustrated in the context of Figures 8, 9A to 2C. Column embedding may be applied when the input data may be tabular data. By applying column embedding to one or more elements, particularly tokens associated with input data, a machine-readable representation can be obtained that specifies the location of one or more elements within table 1102, preferably in the columns of table 1102. Applying column embedding may mean adding column coefficients to the input data embedded via token embedding, particularly the embedded input data. Column coefficients may be the same for elements associated with the same column and / or may differ between two or more elements associated with different columns. Similarly, row embedding may be applied if the input data may be tabular data. By applying row embedding to one or more elements, in particular tokens associated with the input data, a machine-readable representation may be obtained that specifies the location of one or more elements within table 1102, preferably within rows of table 1102. Applying row embedding may mean adding column coefficients to input data embedded via token embedding, in particular embedded input data. Row coefficients may be the same for elements associated with the same row and / or may differ between two or more elements associated with different rows.

[0180] In one embodiment, the input data may be at least partially numeric and at least partially text. Therefore, the input data may contain two or more types of data. The data type may refer to a modality. Therefore, different embeddings may be applied to the input data. The portion of the input data containing text may be subject to the input embeddings referenced in Figures 8, 9A to 2C. The portion of the input data that is numeric token embedding may be subject to positional embedding, column embedding, and row embedding. Furthermore, segment embedding may be applied to the input data regardless of the type of input data. Segment embedding may specify a type of input data to which one or more elements may be associated. For example, if the input data contains text and numbers, the input data may consist of two types of input data. Applying segment embedding to input data may refer to adding a segment coefficient to the input data, preferably the embedded input data and / or the input data, after applying token embedding. A segment coefficient may specify a type of data to which one or more elements may be associated. A segment coefficient may be the same for one or more elements associated with the same type of input data, and / or may differ for two or more elements associated with different types of input data.

[0181] Applying token embedding, positional embedding, segment embedding, column embedding, row embedding, or a combination thereof may generate embedded input data and / or an output of any one of the encoder inputs 978, 984, 988, or decoder inputs 984, 994. The data obtained by applying token embedding, positional embedding, segment embedding, column embedding, row embedding, or a combination thereof may be processed by encoder blocks 974, 986, decoder blocks 980, 990, encoder output 976, decoder output 992, 982.

[0182] Figure 12 shows one embodiment of the Mamba architecture. The Mamba architecture can be used as a data-driven model. The Mamba architecture can improve inference speed in relation to transformer-based models.

[0183] A layered Mamba architecture may be similar to the transformer-decoder architecture discussed in relation to Figure 12. However, instead of decoder blocks, Mamba blocks 1232 and 1204 are stacked. Mamba block 1232 may be based on a selective spatial state sequence model (S6).

[0184] The input tokens are linearly projected into an extended latent space (which may allow for capturing more information during processing in the selective state-space layer 1210) via linear layers 1212 and 1220, followed by convolution via convolution layer 1214 and a nonlinear function (e.g., a sigmoid linear unit (SiLu) or a SWISH activation function). The convolution prior to the selective state-space layer 1210 may interfere with independent token computation. The selective state-space layer 1210 performs selective state-space operation. Furthermore, learnable skip connections may be provided via linear layer 1220, which may map inputs to outputs using linear transformations, which may help mitigate the vanishing gradient phenomenon, similar to residual connections in transformer models.

[0185] The selective state-space layer 1210 may be a linear recurrent network that selectively processes data based on input tokens, which may allow for focusing on relevant data and discarding irrelevant data. For example, a separate weight vector can be determined at each step based on each input token. The determined weight vectors can then be used for selective scanning.

[0186] The selective state-space layer 1210 can be used in convolution mode, for example, for parallelizable training, and in recurrent mode, for nearly constant generation of output data. State-space operations can be based on solving state and output equations, where the state equation can describe how the state changes based on how the input affects the state, and the output equation can describe how the state is transformed into an output. Furthermore, how the input affects the output can be represented by a learnable linear transformation, such as matrix D, used in learnable skip connections.

[0187] The state equation for the hidden state can be as follows (in discretized form): h k =Ah k-1 +Bx k The output can be expressed (in discretized form) as follows:

[0188] y k =Ch k This discretized space-state model can be expanded into a recurrent form similar to a recurrent network, illustrating that a selective state-space model can be or may contain a linear recursive model. However, here matrices A, B, and C can also be used as kernels for the convolution of the state-space model. The kernel K for this purpose is, for example, K=(CA 2 B, CAB, CB) This is possible, and as a result, the output: y k+1 =CA 2 Bx k-1 +CABx k +CBx k+1 It becomes possible to determine this. Therefore, in this representation of the state-space model, training can be performed in parallel, as in a convolutional neural network.

[0189] Matrix A can be a matrix that sufficiently represents recent tokens and decays older tokens, and HiPPO:

[0190]

number

[0191] It can be initialized using, where every entry below the diagonal is set to 0. This may make it possible to create long-term memory for a selective state-space model.

[0192] For Mamba block 1232, matrices B and C, as well as the step size Δ used for matrix discretization, may depend on the input token and may be trained during training, resulting in different matrices B and C being determined for each input token, thereby enhancing content recognition and potentially acting similarly to multi-head self-attention in transformer models. However, unlike spatial state models with fixed matrices A, B, and C, the convolutional representation may not be readily determined here. Therefore, to operate the selective state space layer 1210 in convolutional mode, selective scanning may be applied by leveraging the relevant properties of hidden state computation, and parallel training may be used by enabling parallel determination of partial sequences and iteratively combining them. Further read and write operations can be reduced by using kernel fusion with the described step size, selective scanning, and multiplication with C. The linear layer 1202 can project the generated output back to the same dimension as the input.

[0193] Mamba blocks may be used with a mixture of transformer decoder blocks or expert blocks (for example, a decoder block in which a feedforward layer is swapped with a gating network and several parallel feedforward layers, and the gating network switches between feedforward layers depending on the input), which may allow the advantages of different architectures to be leveraged.

[0194] An example of the Mamba block architecture can be found in “Mamba: Linear-Time Sequence Modeling with Selective State Spaces”, Albert Gu and Tri Dao arXiv:2312.00752v2[cs.LG] May 31, 2024, which is incorporated herein by reference.

[0195] The publication Prior Art Disclosure; No. 684; paragraphs

[1000] to

[8005] ; ISSN: 2198-4786; published February 12, 2024, is considered Reference RF1, which is incorporated herein by reference in its entirety. Preferably, the (chemical) product is a 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 the product.

[0196] The conversion process for obtaining the product preferably includes one or more steps described below and can be carried out by conventional methods well known to those skilled in the art. The conversion process preferably includes, Recycling, preferably by depolymerization, gasification, pyrolysis, and / or vapor decomposition, and / or Purification, preferably crystallization, (solvent) extraction, distillation, evaporation, hydrogenation, absorption, adsorption, and / or subjecting to ion exchange, and / or Assembly, preferably foaming, synthesis, chemical transformation, polymerization, and / or compounding, and / or Formation, preferably foaming, extrusion and / or molding, and / or Finishing, preferably coating and / or smoothing It includes one or more steps selected from the following.

[0197] In addition, one or more steps are described in detail in reference RF1; paragraphs

[1000] to

[8005] .

[0198] This disclosure has been described with reference to several examples, along with preferred embodiments. However, by examining the drawings, this disclosure, and the claims, other variations for carrying out the subject matter described in the claims can be understood and implemented by those skilled in the art. In particular, any of the presented steps can be performed in any order, i.e., the present invention is not limited to any particular order of these steps. Furthermore, it is not required that the different steps be performed in a specific location or on one node of a distributed system, i.e., each step can be performed on different nodes using different equipment / data processing devices.

[0199] All sequences of method steps presented above are not mandatory, and alternative sequences may also be possible. Nevertheless, any particular sequence of method steps shown as an example in the figures should be considered one possible sequence of method steps for each embodiment described by each figure, or for embodiments including at least some of the steps described by each figure.

[0200] In this specification, it should be understood that any connection presented in the embodiments described herein is such that the components involved are operably coupled. Thus, the connection may be direct or indirect, involving any number or combination of intervening elements, and merely a functional relationship may exist between the components.

[0201] As used herein, “determine” may also include “initiate or cause a decision,” “generate” may also include “initiate and / or cause a generation,” “provide” may also include “initiate or cause a decision, generation, selection, transmission and / or transmission,” and “receive” may also include “initiate or cause a decision, generation, selection, retrieval and / or reception.” “Initiate or cause the performance of an action” may include any processing signal that triggers a computing node or device to cause each action to be performed.

[0202] The terms "comprising" or "including" should be understood in an open sense, meaning that an object "comprising element A" may also possess additional elements in addition to element A. Furthermore, the terms "comprising" or "including" can also be limited to "consisting of," that is, consisting only of specific elements.

[0203] The indefinite article “a” or “an” should not be understood as “one,” meaning that the use of the expression “an element” does not preclude the existence of further elements. A single element or other unit may perform the function of several subjects or items described in the claims. The mere fact that certain means are described in different dependent claims does not mean that combinations of these means cannot be used in a favorable implementation or that further elements may be included.

[0204] The expressions "A and / or B" and "at least one of A or B" are considered interchangeable and mean any one of the following three scenarios: (i) A, (ii) B, or (iii) A and B. More generally, the expressions "at least one of the following list of two or more elements" and "at least one of the following list of two or more elements" and similar expressions mean at least one of the elements, or at least any two or more of the elements, or at least all of them, when the list of two or more elements is joined by "and" or "or".

[0205] Providing information within the scope of this disclosure may include any interface configured to provide data. This may include application programming interfaces, human-machine interfaces such as displays, and / or software module interfaces. Providing information may include the communication of data or the transmission of data to an interface, in particular display to a user or use of the data by a receiving entity.

[0206] Within the scope of this disclosure, acquisition may include any interface configured to acquire or receive data. This may include application programming interfaces, human-machine interfaces such as displays, and / or software module interfaces. Acquisition may include the communication of data or the submission of data from an interface, in particular the use of data by a receiving entity. Any acquisition of data, data structures, datasets, etc. may include receiving data, data structures, datasets, etc. from a server that provides (e.g., hosts) a database containing data, data structures, datasets, etc.

[0207] Various units, circuits, entities, nodes, or other computing components may be described as “configured to” perform one or more tasks. “Configured to” means having “circuits” that perform one or more tasks of operation. A unit, circuit, entity, node, or other computing component may be configured to perform tasks even when the unit / circuit / component is not operating. A unit, circuit, entity, node, or other computing component forming a structure corresponding to “configured to” may include hardware circuitry and / or memory that stores executable program instructions to perform operations. For convenience in this description, a unit, circuit, entity, node, or other computing component may be described as performing one or more tasks. Such descriptions shall be interpreted as including the phrase “configured to.” Any description of “configured to” is expressly intended not to invoke the interpretation of § 112(f) of the U.S. Patent Act.

[0208] Generally, the methods, apparatus, systems, computer elements, nodes, or other computing components described herein may include memory, software components, and hardware components. Memory may include volatile memory such as static or dynamic random access memory and / or non-volatile memory such as optical or magnetic disk storage, flash memory, and programmable read-only memory. Hardware components may include any combination of combinational logic circuits, clock memory devices such as flops, registers, and latches, finite state machines, memory such as static random access memory or embedded dynamic random access memory, custom-designed circuits, and programmable logic arrays.

[0209] In this specification, it should be understood that any connection presented in the embodiments described herein is such that the components involved are operably coupled. Thus, the connection may be direct or indirect, involving any number or combination of intervening elements, and merely a functional relationship may exist between the components.

[0210] Furthermore, any of the methods, processes, and actions described or illustrated herein may be performed using executable instructions within a general-purpose or dedicated processor and stored in a computer-readable storage medium (e.g., disk, memory) executed by such a processor. The reference to “computer-readable storage medium” should be understood to include dedicated circuits such as signal processing devices and other devices.

[0211] Any disclosures and embodiments described herein relate to the methods, systems, devices, and computer program elements outlined above, and vice versa. Advantageously, any benefits derived from any embodiment and example are equally applicable to all other embodiments and examples, and vice versa.

[0212] All terms and definitions used herein are to be understood in a broad sense and, unless otherwise indicated, have their general meanings.

[0213] All presented embodiments are illustrative, and it should be understood that any feature presented for a particular exemplary embodiment may be used in any form of itself, or in combination with any feature presented for the same or another particular exemplary embodiment, and / or in combination with any other feature not mentioned herein. In particular, the exemplary embodiments presented herein should also be understood to be disclosed in all possible combinations with each other, provided that they are technically reasonable and the exemplary embodiments are not substitutes for each other. Furthermore, it should be understood that any feature presented for an exemplary embodiment in a particular category (method / apparatus / computer program / system) may also be used in a corresponding manner in any other exemplary embodiment in any other category. It should also be understood that the presence of a feature in a presented exemplary embodiment does not necessarily mean that the feature constitutes an essential feature and cannot be omitted or replaced.

Claims

1. A method for monitoring and / or controlling the production of at least one chemical product, To provide chemical product data related to the production and / or processing of the aforementioned chemical products, To provide a plurality of chemical data templates that show the structure associated with at least a portion of the aforementioned chemical product data, Selecting at least one chemical product data template by determining that at least one chemical product data template corresponds to at least a portion of the chemical product data, To generate contextualized chemical product data by merging at least a portion of the aforementioned chemical product data and at least one selected chemical product data template, Providing the contextualized chemical product data to a data-driven model for generating environmental characteristics data relating to the processing and / or production of the chemical product, wherein the data-driven model is configured to provide the environmental characteristics data in response to what is provided by the contextualized chemical product data; To provide the environmental characteristics data for monitoring and / or controlling the processing and / or production of at least one of the chemical products, Methods that include...

2. The aforementioned chemical product data is One or more material flows associated with the processing and / or production of the at least one chemical product, and / or One or more production conditions and / or processing conditions associated with the processing and / or production of the at least one chemical product, Transportation data indicating the transportation of the at least one chemical product and / or the materials for producing the at least one chemical product. The method according to claim 1, which can be associated with the following.

3. The aforementioned chemical product data may relate to the production and / or processing of two or more chemical products, and the production and / or processing of the two or more chemical products is Two or more material flows associated with the processing and / or production of the two or more chemical products, Two or more production conditions and / or processing conditions associated with the processing and / or production of the two or more chemical products, A transport dataset comprising two or more transport datasets representing the transport of two or more chemical products and / or the transport of materials for producing the two or more chemical products, or a combination thereof, wherein the environmental characteristics data includes at least one environmental characteristics dataset for each chemical product. The method according to claim 1 or 2, which may differ in at least one of the following:

4. The method according to any one of claims 1 to 3, wherein the data-driven model is a pre-trained data-driven model, and the pre-trained data-driven model is trained on unstructured data to provide a plurality of different output datasets based on the provision of a plurality of different input datasets.

5. The method according to any one of claims 1 to 4, wherein the data-driven model is a fine-tuned data-driven model, the fine-tuned data-driven model is a pre-trained data-driven model trained to provide environmental characteristic data based on what is provided by contextualized chemical product data, and the pre-trained data-driven model is trained on unstructured data to provide a plurality of different output datasets based on what is provided by a plurality of different input datasets.

6. The method according to any one of claims 1 to 5, further comprising selecting a data-driven model from a plurality of data-driven models based on the chemical product data, and / or providing instructions for selecting a data-driven model from a plurality of data-driven models, and selecting a data-driven model from a plurality of data-driven models based on the instructions.

7. The method according to any one of claims 1 to 6, wherein the plurality of chemical product data templates are provided via a database, and selecting the at least one chemical product data template based on at least a portion of the chemical product data includes providing a query to provide the at least one chemical product data template to the database based on the chemical product data and / or providing the at least one chemical product data template based on the query.

8. The method according to any one of claims 1 to 7, wherein the plurality of chemical product data templates are provided via a database, and the method includes: generating embedded chemical product data, wherein the embedded chemical product data is a numerical representation of the chemical product data; generating one or more embedded chemical product data templates, wherein the one or more embedded chemical product data templates are numerical representations of the one or more chemical product data templates; and selecting the at least one chemical product data template based on a distance score indicating the distance between the embedded chemical product data and the embedded chemical product data templates.

9. The method according to any one of claims 1 to 8, wherein the chemical product template and / or the contextualized chemical product data includes a data generation task command for triggering the data-driven model to generate the environmental characteristics data. The aforementioned data generation task instruction may be associated with unstructured data, particularly string data. The data generation task instruction may be associated with and / or configured to trigger the data-driven model.

10. The method according to any one of claims 1 to 9, wherein the contextualized chemical product data includes one or more chemical data elements associated with the chemical product data and one or more template elements associated with the at least one selected chemical product data template.

11. The method according to any one of claims 1 to 10, wherein the chemical product data template indicates a sequence of one or more chemical data elements and one or more template elements.

12. The method according to any one of claims 1 to 11, wherein the plurality of chemical product data templates include a plurality of types of chemical product data templates, and selecting at least one chemical product data template based on at least a portion of the chemical product data means selecting at least one chemical product data template based on the type of the chemical product data template, wherein the type of the chemical product data template relates to the production and / or processing of the chemical product and / or one or more properties of the chemical product.

13. The method according to any one of claims 1 to 12, wherein providing the chemical product data includes searching for the chemical product data by providing a request to search for the chemical product data, and providing the chemical product data in response to providing the request to search for the chemical product data.

14. Apparatus for monitoring and / or controlling the production of chemical products, Processor and A memory that stores instructions that, when executed by the processor, constitute the device to perform any one step of the method according to any one of claims 1 to 13, A device equipped with the following features.

15. Use of environmental characteristics data relating to the processing and / or production of one or more chemical products, obtained by any one of the methods described in any one of claims 1 to 13, for controlling and / or monitoring the production and / or processing of one or more chemical products.

Images