Orchestrator-based research and development system and method for operating the system

Abstract

The present invention relates to a research and development system (100) suitable for researching and / or developing products and methods for manufacturing products, in particular energetic materials, comprising a database (110) and an interface (120) for inputting and outputting data, a data processing unit (130) and an execution unit (140). The present invention further relates to a method.

Description

[Technical Field]

[0001] The present invention relates to a research and development system for researching and / or developing products and methods for manufacturing (producing) products, in particular energetic materials. The system comprises a graph database, a data processing unit and an execution unit. The system is configured to connect work steps of research and development projects in an optimized workflow, to work with heterogeneous systems, and to aggregate and store heterogeneous resulting data according to a common ontology. The present invention further relates to a method for operating the research and development system according to the present invention.

[0002] The present invention relates to a research and development system for conducting and integrating research and / or development (R&D) on products and methods for producing products, in particular energetic materials. The present invention further relates to a method for operating an R&D system according to the invention. [Background technology]

[0003] Current systems and methods for supporting research and development of products and methods for manufacturing products, particularly energetic materials, are time- and labor-intensive, prone to errors and failures, and often produce suboptimal results. [Objective of the present invention]

[0004] It is therefore an object of the present invention to provide an improved research and development system and method for operating a research and development system for researching and / or developing products and methods for manufacturing products, in particular energetic materials. [explanation]

[0005] The above problem is solved by an R&D system for research and / or development according to claim 1 and a method for operating an R&D system according to claim 8. Preferred configurations of the invention are set out in the dependent claims, which help to provide an improved R&D system and an improved method for operating said system.

[0006] The research and development system according to the present invention is suitable for researching and / or developing products and methods for manufacturing products, in particular energy materials. The system comprises a database, an interface for inputting and outputting data, a data processing unit, and an execution unit. The database is a graph database configured to store data according to a data model that maps a well-defined ontology.

[0007] The data processing unit may be configured to normalize (standardize), complement, and / or enrich the data stored in the graph database. Additionally, the data processing unit may be configured to identify statistical and / or causal relationships between the data in the database and to model these relationships using statistical and / or physico-chemical models. Furthermore, the data processing unit may be configured to identify research and development goals in the stored data and / or to generate and / or adapt appropriate workflows and / or work steps to achieve the research and development goals.

[0008] In this case, the execution unit may be configured to select a workflow for achieving the development goal depending on the development goal, and in addition, the execution unit may be configured to select a next work step from the sequence of work steps of the selected workflow depending in particular on the preceding work step and / or the result of the preceding work step.

[0009] In one implementation, the execution unit can be configured to select a terminal (terminal device) for performing a work step, particularly depending on the type, scope and / or time of the work step to be performed, and / or the type, scope, time and / or availability of the terminal depending on whether certain conditions are met.

[0010] In a further embodiment, the research and development system may comprise a data acquisition unit configured to localize, capture, and / or store in the graph database information related to the selected research goal, in particular expert articles, publications, test series, lectures, comments, and / or other related records and / or documents. The data acquisition unit may further be configured to localize, capture, label with respect to origin and provenance, and / or store in the graph database results of execution of work steps by the terminals.

[0011] In one implementation, the data acquisition unit can be configured to capture information about the type, range, and / or availability and / or non-availability of the terminal and / or store the information in the graph database. In addition, the data acquisition unit can be configured to capture information about the technical, operational, legal, contractual, and / or other availability conditions of the terminal and the fulfillment status of the availability conditions and / or store the information in the graph database.

[0012] In one implementation, the research and development system includes an interactive human-machine interface that can be configured to graphically display work steps and / or results of work steps of a workflow and / or display availability of workflows, work steps, and terminals for performing the work steps. Similarly, the interactive human-machine interface can be configured to receive input from a human user for selecting a workflow to perform, a work step to perform, and / or a terminal for performing the work step.

[0013] In one implementation, the research and development system includes a server, a cloud system, a terminal, an edge computing unit, and / or a fog computing unit. In a preferred embodiment, the server, the cloud system, the terminal, the edge computing unit, and / or the fog computing unit are IoT-enabled.

[0014] In one implementation, the graph database, the data processing unit, the execution unit, the data acquisition unit, the edge computing unit, the fog computing unit and / or the interactive human-machine interface are configured to use artificial intelligence (AI) methods, in particular machine learning (ML) and / or deep learning (DL), and / or to provide results in real time.

[0015] A method for operating an R&D system configuration according to the present invention may comprise one or more of the following steps a) to k), which in preferred configurations are performed using artificial intelligence methods and / or in real time:

[0016] storing data in a graph database according to a data model representing a well-defined ontology;

[0017] centralizing, complementing, and / or enriching the data stored in the graph database;

[0018] identifying statistical and / or causal relationships between the data stored in the graph database and / or modeling these relationships using statistical and / or physico-chemical models;

[0019] identifying research and development objectives contained in the stored data and / or generating workflows and / or work steps suitable for achieving the research and development objectives;

[0020] selecting a workflow for achieving the development goal depending on the development goal and / or selecting a work step from the sequence of work steps of the selected workflow, in particular depending on the preceding work step and / or the result of the preceding work step;

[0021] selecting a terminal for performing a work step, in particular depending on the type, extent and / or time of the work step to be performed and / or the fulfillment of the type, extent, time and / or availability conditions of the terminal;

[0022] analyzing the results of the work steps by the edge computing unit and / or the fog computing unit and / or forwarding the analysis results as results of the work steps;

[0023] taking the results of the work steps, labelling them with respect to origin and provenance, and / or storing them in a graph database;

[0024] localizing, capturing and / or storing in a graph database information relevant to the research goal, in particular expert articles, publications, test series, lectures, comments and / or other relevant records and / or documents;

[0025] localizing, capturing and / or storing in a graph database information about the type, range, availability and / or non-availability of the terminals, about the technical, operational, legal, contractual and / or other conditions of the terminals' availability and / or about the fulfilment of the availability conditions;

[0026] graphically displaying the results of the work steps, the availability of certain work steps of the workflow, the workflow, the work steps and / or the terminals for performing the work steps; and / or

[0027] Receiving from a human user a selection of a workflow or work step to be performed and / or a terminal for performing the work step.

[0028] The method according to the invention and / or the individual steps of the method according to the invention can be implemented as a computer program product, which is capable of performing the method and / or the individual steps of the method when executed by a suitable computing unit. Such a computer program product can be stored on a storage medium and / or can be provided for installation, storage and / or download to a computing unit, such as a server or a cloud system. [Brief explanation of the drawings]

[0029] For a fuller understanding of the features and advantages of the present invention, several embodiments of the present invention will now be described in more detail with reference to the accompanying drawings 1 to 14. However, the illustrations in the drawings merely serve to illustrate several embodiments and should not be considered limiting, as there may be other equally effective embodiments.

[0030] [Figure 1] 1 is a schematic diagram illustrating an implementation of a research and development system according to the present invention. [Figure 2] 1 shows an example of a series of steps of a method according to the invention for designing, initializing and continuously using a research and development system according to the invention; [Figure 3] 1 shows an example of a series of steps of a method according to the invention for designing, initializing and continuously using a research and development system according to the invention; [Figure 4] A simplified representation of the European Materials Modelling (EMMO) ontology. [Figure 5] FIG. 1 shows a simplified example of an ontology according to the present invention. [Figure 6]FIG. 1 illustrates the main advantages of the cross-functional and / or cross-user ontology according to the present invention. [Figure 7] 1 is a diagram showing the main components of the Resource Description Framework (RDF) concept that can be used in an arrangement according to the invention, for example in the field of incorporating materials into components for H2 technology. [Figure 8] FIG. 1 shows an example of the integration of a graph database into an R&D system according to the present invention, with a focus on decentralization of terminals and a highly heterogeneous data structure with a complex data pipeline. [Figure 9] FIG. 10 illustrates an example of importing raw data in tabular format, e.g., as an .out file, according to the present invention. [Figure 10] 10A and 10B show examples of graphical representations of imported raw data according to the present invention. [Figure 11] 11 shows an example of a top-level visualization of a manufacturing process based on metrology and simulation data in accordance with the present invention. Figure 11 shows a simplified representation of the data infrastructure and the steps for training and applying machine learning models in accordance with the present invention. [Figure 12] 1 is a simplified schematic diagram of an embodiment according to the present invention; [Figure 13] FIG. 1 shows a schematic example of the use of statistical and causal models according to the present invention. [Figure 14] Diagram showing the process and architecture. [Detailed Description]

[0031] Current scientific research and development (R&D), especially in the field of energy materials, requires enormous efforts and highly skilled personnel. The technological and organizational infrastructure, especially systems, devices and processes, is highly heterogeneous, decentralized, poorly connected, and standardized to a limited extent. This has a negative impact on the efficiency and quality of R&D.

[0032] To solve the above drawbacks, an improved research and development system 100 is proposed for the development of products and their manufacturing methods, especially for energy materials, through to device integration. This research and development system 100 is particularly helpful for ensuring data connectivity between decentralized data and terminal nodes, and optimizing the research and development process.

[0033] Research and / or development is understood to mean all planned and / or systematic activities based on scientific methods with the aim of acquiring new knowledge, where "new" is to be understood in relation to the respective organizational unit carrying out the research and / or development. Research and development system refers to a system for supporting research and / or development activities. Product refers to an object having tangible components.

[0034] The research and development system 100 according to the present invention may also be referred to as an orchestrator. An orchestrator refers to a hardware- and software-based unit for automatically managing tasks on one or more devices. The orchestrator can orchestrate the execution of tasks, i.e., connect and / or automate tasks in a coherent workflow, to achieve a predetermined goal. In particular, this can include making devices available and automatically initiating task execution on devices, reserving or allocating processing power, working with heterogeneous systems, and / or implementing deployments in different geographic locations and with various device operators. In addition, orchestration can include the performance of other management and control functions, such as authorization monitoring and / or policy enforcement when using devices.

[0035] A distinction should be made between orchestration and simple automation. Automation is a subfield of orchestration. Automation focuses on making tasks rapidly repeatable with little or no manual intervention. Orchestration allows for coordination between and across multiple automated activities and is environmentally responsible.

[0036] A workflow is a process consisting of individual parallel and / or sequential work steps and / or activities, especially for research and development of products or manufacturing methods. A workflow describes the operational and technical perspective of the work steps and / or activities to be performed. Ideally, this description is precise enough that subsequent work steps or activities depend on the results of previous steps or activities. Individual work steps and / or activities are therefore interdependent. A workflow contains multiple interrelated work steps. A workflow has a defined beginning, a structured flow, and a defined end. Workflows are characterized by their collaborative nature. They should be distinguished from collaborative systems, where synchronous and strictly separated execution of steps and / or activities is important.

[0037] A work step comprises an activity or series of activities performed towards the achievement of a given research and / or development goal. The activities and their execution of a work step may be meaningfully separable from other activities. However, the activities and their execution of the same work step may not be meaningfully separable or may only be meaningfully separable with difficulty due to their internal structure and / or interdependencies. In particular, a work step may be an experiment, test, measurement, observation, and / or mechanical, physical, and / or chemical modification of a tangible object and / or substance or material.

[0038] Energetic materials include in particular materials essential for technologies for scalable energy conversion in or from electrical energy and / or energy storage, such as for fuel cells, electrolysis of water or CO2, photovoltaics and / or rechargeable or primary batteries.

[0039] 1 shows an overview of an implementation of a research and development system 100 according to the present invention. The system comprises a database 110 with an interface 120 for inputting and outputting data, a data processing unit 130, and an execution unit 140. The database comprises a graph database 110 configured to store data according to a data model representing a well-defined ontology 16. An ontology refers to a fixed set of classes, rules, and constraints for the formal description of knowledge.

[0040] Data processing unit refers to a unit for electronic evaluation and processing of electronically stored data, in particular for recognizing connections, similarities, patterns, dependencies and / or redundancies for classification, assignment and / or derivation of models, predictions, concepts and plans.

[0041] An execution unit refers to a unit for selecting and executing workflows, work steps and / or activities to achieve a goal, and for selecting and / or initiating terminals 170, in particular the time, place and / or organizational unit for executing the workflows, work steps and / or activities.

[0042] Furthermore, the research and development system 100 according to the present invention may comprise a data acquisition unit 150 and / or an interactive, preferably graphical, human-machine interface 160. The data acquisition unit 150 may be configured to localize, capture, and / or store in the graph database 110 information related to the research goal and / or research area, such as expert articles, publications, test series, lectures, comments, and / or documents. Such information may be stored electronically in a distributed manner, for example, on a public server 220, in a research database 230, or on a website 240. This information can be localized, for example, using a crawler, and, if necessary, evaluated and captured using text mining methods. Similarly, information may not be available in electronic form. It may be possible to capture it electronically, for example, using a scanner 250.

[0043] An interactive human-machine interface refers to an input / output device that allows information, data, and / or commands to be exchanged between a human user and a data processing system. An interactive graphical human-machine interface 160 refers to the output of information and / or options for inputting information, data, and / or commands that are adapted to human perception and / or that a human user can understand and / or learn particularly quickly and easily. Such an interface can comprise a dashboard, i.e., a graphical user interface used to visualize data and / or operating elements.

[0044] A data acquisition unit refers to a unit for the manual and / or electronic automated identification and / or capture of analog and / or electronic, structured and / or unstructured data. Manual capture can include input by a human user via keyboard, voice, camera or scanner, for example. Automated capture in this context refers to capture, especially by machine-to-machine communication, for example by using crawlers or text mining units.

[0045] The (software) programs, also called bots or spiders, are called crawlers. They search communication networks, especially the Internet, in an automated way. To do this, crawlers perform a predetermined task, e.g., sequentially visiting a large number of addresses on the net. The content stored at the addresses is searched and checked, e.g., for the presence of predetermined related content, and / or copied for storage in a database. Similarly, crawlers can follow links found at one address to other addresses in order to continue or extend the search for related content.

[0046] Text mining refers to algorithm-based analytical methods for discovering semantic structures from unstructured and / or weakly structured text data. Text mining typically proceeds in several steps. First, appropriate data material is collected, for example, with the help of crawlers specialized in relevant topics. In a second step, this data is prepared, for example, including automatic and / or optical text and / or character recognition, so that it can be analyzed using text mining methods. Text mining methods involve statistical and linguistic tools that allow structure and data to be extracted from text, ideally in an automated manner, that can be captured and stored in a database. However, at a minimum, text mining methods must enable a human user to quickly identify important information in the processed text. Ideally, text mining methods provide information that the human user does not know in advance whether or not the processed text contained such information. When used in a targeted manner, text mining techniques can also generate hypotheses, test them, and refine them over time.

[0047] The research and development system 100 may comprise a terminal 170 for carrying out research activities. The terminal 170 may be communicatively connected to the database 110 and / or the interface 120, the execution unit 140 and / or the data acquisition unit 150. The communication connection 200 may be established, for example, via the Internet and / or a dedicated private communication network for voice and / or data.

[0048] Terminal refers to a system, instrument, computer, or other device, and methods implemented therein, for performing a work step or activity. Terminal 170 may be, for example, a high-performance computer, a self-driving lab or laboratory (SDL), a high-throughput screening (HTS), a potentiometer, a porosimeter, a viscometer, an imaging method, a mathematical, numerical, or theoretical analytical model, or a computational and atomistic mesoscale simulation method. Terminals generate vast amounts of data, such as simulation and calculation data, imaging data from transmission electron microscopy (TEM), imaging data from scanning electron microscopy (SEM), electroanalytical measurement data, such as impedance, power curves, and any other form of current-voltage data or material characterization data.

[0049] In one implementation, the data acquisition unit 150 is configured to capture results of the performance of work steps by the terminal 170, label the results with respect to their origin and origin, and / or store them in the graph database 110. To reduce the data to be communicated, the research and development system 100 may comprise an edge computing unit 180 and / or a fog computing unit 190.

[0050] Edge or fog systems refer to intermediate layers between a core data center, particularly a server or cloud computing system, and the terminals 170 connected by a network infrastructure. These intermediate layers comprise analysis units, so-called edge computing units 180 and / or fog computing units 190, which are located at or near the respective terminals 170. Typically, the fog computing units 190 are located between the edge computing units 180 and the central unit. These edge / fog computing units 180, 190 analyze large amounts of raw data from the terminals 170 and simply forward the results and / or findings obtained therefrom to the core data center, e.g., a server or cloud. The original raw data is discarded. Edge / fog systems thus shift data processing to the "edge" of the network, i.e., the edge and / or the "fog between the edge and the cloud," thereby helping to minimize latency and prevent bottlenecks in data transmission on the network.

[0051] Furthermore, the data acquisition unit 150 can be configured to localize, retrieve, and / or store in the graph database 110 information about the type, extent, or availability and / or non-availability of the terminal 170 or any other related resource. In addition, the data acquisition unit 150 can be configured to locate, retrieve, and / or store in the graph database 110 information about the technical, operational, legal, contractual, and / or other availability conditions of the terminal 170 or other resource and / or about the fulfillment of the availability conditions. Such information can in particular be stored in a decentralized research facility, for example in a local file system or database 210.

[0052] The data processing unit 130 can be configured to supplement, unify, and / or enrich the data stored in the graph database 110. Furthermore, the data processing unit 130 can be configured to identify statistical or causal relationships between the data in the database 110 and model these relationships using statistical and physico-chemical models. Additionally, the data processing unit 130 can be configured to identify research and development goals within the stored data and to generate appropriate workflows and / or work steps for achieving these goals based on the identified goals and relationships between the data, and / or to adapt existing workflows and / or work steps based on the additional data.

[0053] The execution unit 140 can be configured to select a workflow and / or a next work step in the workflow for achieving a development goal, in particular the selection can be made depending on the development goal or the previous work step and / or its outcome.

[0054] Furthermore, the execution unit 140 can be configured to select a terminal 170 for performing a work step or activity, the selection being made depending on the type, range and / or time of the work step / activity to be performed and the fulfillment of the type, range, time or availability conditions of the terminal 170.

[0055] The interactive human-machine interface 160 may be configured to graphically display a work step or work steps of a workflow and / or their results. Similarly, the interface 160 may be configured to display the availability of terminals 170 and / or any other relevant resources. The interactive human-machine interface 160 may also be configured to receive input from a human user to select a research goal, a workflow or work step to be performed, or a terminal 170 for performing a work step or activity.

[0056] In one implementation, the units, devices and systems included in the research and development system 100 according to the present invention may be IoT-enabled, and may be configured to provide and / or communicate results in real time and to perform methods of artificial intelligence (AI), in particular machine learning (ML) and / or deep learning (DL).

[0057] IoT stands for "Internet of Things." The "Internet of Things" refers to the connection of uniquely identifiable physical objects (or "things") with electronic interfaces and virtual representations in a (global) internet-like infrastructure. It includes communication protocols optimized for machine-to-machine communication, which allows communication not only between humans, but also between humans and things, and between things. Objects in the Internet of Things are thus given the opportunity to organize themselves, exchange information, and interact with each other. Human intervention remains possible in principle, but is no longer absolutely necessary. Artificial intelligence (AI) refers to the ability of technological systems to exhibit human-like intellectual functions such as reasoning, learning, planning, creativity, vision, hearing and / or comprehension.

[0058] Machine learning (ML) refers to the ability of technical systems to generate knowledge from experience. Deep learning (DL) refers to a machine learning method that develops an extensive internal structure by using artificial neural networks (ANNs) with many intermediate layers ("hidden layers") between the input and output layers. Such artificial systems are able to learn from examples and generalize after a learning period. Various methods can be used for this, such as supervised learning, unsupervised learning, reinforcement learning, and deep / multilayer learning.

[0059] In this context, real time refers to an operation in which the processing results are available within a predetermined, particularly guaranteed, period of time, and in particular the processing and / or communication of data takes place almost simultaneously, preferably simultaneously, with the actual corresponding processing steps.

[0060] Furthermore, the research and development system 100 may include a server and / or a cloud system. In particular, the database 110, the data processing unit 130, the execution unit 140, and the data acquisition unit 150 may be implemented and operated on a (central) server and / or a cloud system.

[0061] A server refers to a computing unit that performs certain tasks for other systems connected in a network and on which those systems may rely in whole or in part. Servers help to improve the integration, management, and control of multiple devices, particularly various devices in a research and development system and / or devices in various geographic locations.

[0062] A cloud computing system is a system built according to the cloud computing model. Cloud computing describes a model in which shared computing resources, such as servers, data storage, and applications ("apps"), are provided as an on-demand service, particularly over the Internet, quickly, easily, and device-independently, and charged for according to use. The provision and use of these computer resources is defined, typically through an application program interface (API) and / or for users through a website or app. Characteristics of a cloud computing system include, for example, on-demand self-service, standardized broad network access for various devices, resource bundling, rapid elasticity to meet demand, and continuous performance measurement to optimize and control the cloud system.

[0063] In particular, the research and development system 100 according to the present invention allows for networking, coordination and interaction of various terminals 170 and participants, which in particular provides for fast retrieval, fast availability and traceability of result data, thus improving the R&D management of distributed and heterogeneous R&D units.

[0064] 2 and 3 illustrate an exemplary sequence of steps of a method according to the present invention for designing, initializing, and continuously using a research and development system 100 according to the present invention to support a research and development project. The illustrated steps and sequence represent merely one example implementation of a method according to the present invention and should not be construed as limiting.

[0065] Steps S1 to S9 in FIG. 2 have the following meanings: S1: Define the ontology and data model; S2: Steps to design a graph database; S3: localizing and capturing information and data related to the research area and / or research objectives; S4: Importing / storing the captured information into a graph database according to the interface structure and / or data model; S5: Create a training dataset; complete and unify the data; train a model for enrichment; S6: Complement, unify, and enrich the data in the graph database using (AI) models; S7: Identifying relationships; creating statistical data models; and assigning physicochemical models; S8: Identifying research goals within the data; generating work steps and workflows to achieve the research goals; S9: localizing and capturing information about the availability of terminals, devices, systems and / or other resources; Here, PI stands for Phase I and PII stands for Phase II. Steps S10 to S21 in FIG. 3 have the following meanings: S10: Select a research goal; Select a workflow to achieve the research goal; S11: A step of determining the next work step from the workflow; S11a, S11b: Was step S11 successful? - If yes, continue with step S11a; if no, continue with step S11b; S12: Select / start the device to perform the work step / activity; label for tracking if necessary; S13: Performing work steps / activities via the terminal; sending the results to the edge unit; S14: Evaluate the raw data from the terminal; and send the result to the fog unit; S15: Evaluate the results from the edge unit; and send the results to the intake unit; S16: Capture and store the results of the work steps / activities; if necessary, label them for tracking purposes; S17: Steps to complete, unify and enrich the resulting data, if necessary (see S6); S18: Characterize the resulting data (see S7): Create / assign statistical data models; assign physicochemical models; S19: Generate / adapt new work steps and / or workflows to achieve the research goal (see S8); S20: Manual intervention step to generate / select the next work step; S20a / S20b: Was step S20 successful? - If yes, continue with step S20a; if no, continue with step S20b; S21: Have the study goals been achieved? Steps to terminate and / or discontinue; Here, PIII is Phase III. <Capture unit>

[0066] Using current R&D methods, three phases can usually be distinguished: in the first phase, relevant data sources for each topic or objective are identified, articles are researched, and data are collected and synthesized (see Figure 2, Phase I). This first phase alone can take several days or weeks. In the second phase, Phase II, after data acquisition and preparation, the data are analyzed in consultation with the respective R&D team and / or experts, a competitive comparison overview is created, and a decision is made about the next action steps. Practical R&D work only begins in the subsequent third phase, Phase III. The objectives of this practical R&D work may include, among other things, the manufacture (production) of new substances, materials, and prototypes. To achieve this, it is usually necessary to carry out experiments, measurements, and simulations, as well as to analyze, model, and verify the observed results and relationships.

[0067] The practical R&D work described above in the third phase, Phase III, requires the use of highly specialized equipment and devices, which can also be referred to here as terminals. These are often located in various facilities, both inside and outside the research organization. Contracts must be established, and workflows and results for samples must be coordinated. This is typically done manually and in one-on-one discussions with individual institutions. This process is inefficient and subject to many uncertainties, for example, due to differences in test protocols, equipment integrity, and / or human skills and methods. This alone increases variability.

[0068] Especially when cross-functional, cross-organizational, and / or cross-location R&D facilities are required, the data generated is often highly heterogeneous. For example, a wide variety of data types are generated by different application systems and / or users, such as data from experiments, simulations, or scientific literature such as journals, conferences, blogs, and online databases. Furthermore, different domains and subdomains use their own specialized vocabulary, which may differ at least slightly or partially from the vocabulary and / or semantics of other domains or subdomains. Similarly, a wide variety of units of measurement and / or reference points are used depending on the domain, subdomain, and / or data source.

[0069] Furthermore, data and its data types, classifications, and ranges can refer to different levels of analysis or abstraction, for example, up to the macro, meso, and micro levels. The macro, meso, and micro levels refer to different levels of analysis and / or abstraction. At the macro level, large collections and / or systems are examined. At the meso level, the focus is on the parts and components of these collections or systems. At the micro level, individual elements and / or interactions between individual elements are considered.

[0070] In addition, data management, i.e., the processing and management of research data and results from various sources, is also carried out in a non-uniform and non-standardized manner. The quality of data management rarely meets the requirements of professional organizations with well-defined, standardized and comparable data structures and formats. This makes it more difficult, and in practice often impossible, to use the results of other units in ongoing research projects, and to reuse results from other areas of previous research by other disciplines, either in related fields or in the context of subsequent research projects.

[0071] In order to enable efficient and effective use of the very large and very heterogeneous data masses described above during the preliminary phases I and II of the research project, and during the implementation of the practical research work, Phase III, the data is stored in a common database 110 with an appropriate centralized data structure and an appropriate common data model.

[0072] To enable this, as shown in Figure 2, a well-defined ontology 16 is selected and / or defined in a first step S1, on the basis of which the data model of the database 110 according to the invention is defined. In one configuration, for example, the European Materials Modelling Ontology EMMO of the European Materials Modelling Council EMMC, shown in Figure 4, can be used for this purpose.

[0073] The ontology 16 EMMO1 shown in Figure 4 contains classes with attributes and relationships that may have directions. Examples of relationship types are "general-specific (isA)"2, "set-element (hasMember)"3, "whole-part (hasPart)"4, and "whole-temporal part (hasTemporalPart)5. Further rules and constraints can be introduced for classes and relationships. An example of a relationship is that a <CollectionClass> is "related to" an <ItemClass>. An example of a rule and / or constraint is that each instance of a <CollectionClass> must have at least two "set-element" relationships to different instances of a <ItemClass>.

[0074] Instances represent concrete objects in the ontology. They are created using previously defined classes, for example "Berlin", "London", "Paris", and "Rome" are different instances of the "City Type" of the "Topological Location" class.

[0075] Figure 5 shows an implementation of an ontology 16 according to the present invention. The classes and properties of the constituent ontology 16 shown in Figure 5 are based on the basic methodology and framework concepts of the European Materials Modeling Ontology EMMO1. Such an ontology 16 according to the present invention can be extended to different and / or related R&D domains, and in particular enables interoperability with other EMMO-based platforms. In the implementation shown in Figure 5, the ontology according to the present invention includes the objects "Object / Material / Component" 6, "Process" 7, "Measurement" 8, "Property" 9, and "Metadata" 10, and the properties "Process" 11, "Production" 12, "Involvement" 13, "Whole-Part" 4, "Measurement" 14, and "Gain" 15.

[0076] In this case, classes and class properties are branched into class hierarchies to describe specific materials, components, properties, and processes in an R&D domain. These specialized hierarchies are further refined with rules and constraints that are also domain-specific. This provides a flexible and powerful framework for formalizing, structuring, representing, and storing knowledge.

[0077] Such an ontology 16 can be used to design the data model of the database 110 according to the invention, in which case the ontology 16 makes it possible to define the properties of the data model and also the structure of the interface 120. The ontology 16 thus helps to design the database 110 according to the invention, which provides a clearly defined input / output interface 120 at the application level, thereby enabling efficient data storage and utilization of otherwise heterogeneous application data.

[0078] The most significant advantage of using and / or providing data in a common ontology 16 according to the present invention is illustrated in FIG. 6. In particular, the use of a common, well-defined ontology 16 allows data to be found and utilized from various data sources 170, 220, 230, 240, 250. Similarly, data exchange and further use between different R&D units 18 is possible. Similarly, previously obtained results and previously generated data can be reused for further R&D projects or later steps. On the other hand, in the absence of such a unified ontology 16, results and data generally cannot be objectively interpreted and understood. In this case, results and data ultimately end up being evaluated as subjective data that are of little use in achieving R&D goals.

[0079] A data model suitable for the ontology 16 and the graph database 110 according to the present invention can be built, for example, using the Resource Description Framework (RDF) for standardization of data models. Data models for graph databases built using this framework concept are particularly suited to making information exchangeable between different applications and machine-readable.

[0080] As shown in Figure 7, a graph database designed according to the Resource Description Framework RDF is constructed using so-called RDF representations. An RDF representation is a triple consisting of a subject 22, a predicate 23, and an object 24. The subject 22 is the resource being described, e.g., catalyst ink. The predicate 23 is a property of the described resource, e.g., processed. The object 24 is a concrete value of this property, e.g., a uniquely specified processing step. Each triple 21 thus represents a logical statement about the relationship between the subject 22 and the object 24. Several of these RDF representations 21 form a coherent RDF graph, which can be viewed as a semantic network.

[0081] Subjects 22 and objects 24 are represented in the graph database 110 as nodes 25, and predicates 23 are represented as edges. An edge 26 connects each two nodes 25, thus representing a relationship. Edges 26 can have properties and direction. Edges 26 must have a type. Nodes 25 are instances of related classes and can have any number of properties. In addition, nodes 25 can have any number of "names" 27. Names 27 group nodes into sets such as materials 6 and processes 7. Particularly for large graph databases 110, the edges 26 of a node 25 can be stored in an adjacency list. The adjacency list stores all edges 26 leaving a node 25. Thus, unlike, for example, a matrix structure, it is not necessary to query entire rows and / or columns to identify all of a node 26's neighbors.

[0082] As shown in FIG. 2, in a second step S2, a graph database 110 according to the present invention is designed based on the ontology 16 and the data model. A graph database is a database based on graph theory. It consists of a set of objects, which can be nodes or edges. Nodes represent tangible, intangible, concrete, and / or abstract objects. Edges connect nodes to other nodes and represent their relationships. Graph databases can be constructed, for example, according to the concepts of the so-called Labeled Property Graph (LPG) or the so-called Resource Description Framework (RDF). Accessing nodes and edges in a (native) graph database according to the present invention is an efficient operation with constant execution time, allowing for rapid and exhaustive traversal of millions or even a very large number of edges per second. Regardless of the total size of the dataset, graph databases are particularly well-suited for handling highly interconnected data and complex queries.

[0083] Considering the requirements of the system 100 of the present invention, the use of a graph database is advantageous as opposed to the alternative of a relational database. The system 100 of the present invention handles highly heterogeneous data. The data structure of a relational database is rigid, while the data structure of a graph database is flexible. For the system 100 of the present invention, the recognition of interrelationships and / or direct and indirect relationships is important. While it is difficult to represent indirect relationships in relational databases, the representation of relationships and chains of relationships is a key feature of graph databases. The system 100 of the present invention is able to identify / predict interrelationships, direct and indirect relationships, and similarities. This is only possible with relational databases using specialized AI-based techniques, while graph databases can do so using classical and / or graph-based techniques. The ability to visualize data is essential for the system of the present invention. While relational databases require the use of separate visualization tools for this purpose, the data structure of a graph database already provides visualization of the data and its relationships.

[0084] As shown in FIG. 8, the specific implementation of the second step S2 of the graph database 110 according to the present invention can be performed with the help of programming tools such as Neo4j, Neomodel 20a, and Cypher 20b. Neo4j is an open-source graph database implemented in Java, and its version 1.0 was released in February 2010. Neomodel 20a is the object graph mapper OGM for the Neo4j graph database 110. The object graph mapper OGM maps graph nodes and relationships to objects and references in a specific data model. Object instances are mapped to nodes, and object references are mapped to properties using relationships and / or sequences. Cypher 20b is an open-source graph query language for graph databases based on Neo4j. The Cypher open-source project provides all the specifications necessary to create efficient queries for creating, reading, updating, or deleting graphs without specialized knowledge of specific storage formats.

[0085] In a third step S3, information and data related to a research area or a specific research goal are localized and captured by the data acquisition unit 150. An example of a research area or research question is, for example, the effect of solvents on the production of catalyst layers for PE fuel cells. The information and data can be localized in an automated or semi-automated manner using a suitably configured data acquisition unit 150, for example with the help of a crawler. In doing so, potentially relevant (previously used) information is localized, in particular expert articles, publications, test series, lectures, comments, and / or other relevant experimental records and / or documents. In addition, the relevance of the content is checked, and if necessary, the content is extracted, for example, using text mining methods in the case of less structured text data. In this way, all external / public and internal / private experimental, modeling, and raw data related to the research area and / or question can be collected and classified, for example, based on the production steps.

[0086] In a fourth step S4, the data and information captured by the data acquisition unit 150 can be imported and stored in the graph database 110 according to the present invention. This data can include, for example, measurement data and simulation data resulting from experiments and manufacturing processes. FIG. 9 shows an example of importing raw data into the graph database 110 in tabular form. In the example of FIG. 9, one row of the table represents a fuel cell manufacturing process, characterized, for example, by a manufacturing identification number 19. The columns represent fuel cell manufacturing parameters 19a and materials 6. Importing raw data requires a proper understanding of the manufacturing process so that the table content can be recognized and assigned to the data model of the interface 120 or graph database 110. Alternatively, a properly structured and easy-to-understand interface 120, such as an electronic lab notebook (ELN) interface or a suitably specified .out file 120a, can be used. Similarly, for example, an application program interface (API) 120b, as shown in FIG. 8, can be used to provide the user with a suitable and easy-to-understand structure for inputting data. By capturing and storing such data in a graph database 110 according to the present invention, the data and information can be quickly found, traced, and used uniformly and efficiently by a network of interconnected machines and computer data centers and various operators and participants.

[0087] FIG. 10 shows an example of a visualization of fuel cell manufacturing based on data stored in a graph database 110. Graph storage allows mapping and displaying all parameters 19a and all relationships between the materials 6, manufacturing steps 6a, and parameters 19a, all the way to the materials 6, manufacturing steps 6a, and the final product 6b of the manufacturing process. Additionally, the data model of the graph database 110 based on the ontology 16 according to the present invention allows displaying the manufacturing process, including the characterization of properties 9 and / or measurements 8. This illustrates the flexibility of the resulting model, since the number of processing steps and parameters is not specified. The rules and constraints of the ontology 16 and the resulting data model ensure that only meaningful relationships are introduced between nodes. The ontology 16 underlying the data model facilitates extending the data model as needed, appropriately adapting additional data models, and / or unifying the model.

[0088] Additionally, data processing according to the present invention allows for simple and meaningful visualizations. Figure 11 shows an example of a visualization of a fuel cell manufacturing process 29, including the process from starting materials to the finished fuel cell and measurements on the fuel cell. Meanwhile, a visualization of a simulation 28 is also shown.

[0089] In the fifth and sixth steps S5, S6 (see FIG. 2), completion, unification and enrichment of the data stored in the graph database 110 can be performed, for example, based on suitable regression and / or pattern matching methods.

[0090] In a fifth step S5, training datasets 30 (see FIG. 12 ) are preferably compiled and / or generated based on the datasets stored in the graph database 110. These training datasets can be used to train appropriate algorithms that can be executed, for example, by the data processing unit 130, in particular artificial intelligence (AI) models 31, such as machine learning models ML and deep learning models DL. Such AI 31 can be trained, for example, using the generated training datasets 30 and unsupervised learning. In this process, the model reflects comprehensive knowledge. Further refinement and improvement for more specific tasks and / or partial datasets can be performed, in particular by adjusting the weights of the trained AI model. Subsequently, in a sixth step S6, the trained AI can be applied to other datasets in the database 110 to complement, unify, and enrich the data content, format, attributes, and identifiers.

[0091] The collection of a large amount of data on the physicochemical properties of catalytic materials in a database 110 can serve as an example of the fifth and sixth steps S5, S6 (see FIG. 2). The collected dataset can include, for example, conductivity, electrical properties, current, and voltage. For some materials, the database may lack entries for faradaic efficiency, and / or this data may not have been captured or measured. In such cases, an AI / ML algorithm can correlate all entries (in real time) to derive a relationship between voltage and faradaic efficiency. This auto-correlation function can then be used to predict and impute faradaic efficiency for datasets and / or materials lacking it. This makes datasets / data more comparable and easier to analyze. This helps perform analysis across various data sources 170, 220, 230, 240, 250 (see FIG. 12) and identify and model comprehensive correlations, relationships, and structures within otherwise incomplete and / or heterogeneous data.

[0092] In the third phase, Phase III, researchers, especially in the fields of fuel cells, electrolyzers, and batteries, deal with highly complex systems, devices, and devices in their practical work. This complexity results in a high-dimensional parameter space. Therefore, data-driven models are a promising approach for determining and planning research work in the second phase, Phase II, and for facilitating and optimizing the workflow in carrying out practical research and development work in the third phase, Phase III. For example, data-driven models can be implemented in a self-regulating laboratory 170 to create a feedback loop that can iteratively optimize manufacturing processes and / or discover new materials.

[0093] To generate a data-driven model for optimally planning and performing research work, in a seventh step S7 (see FIG. 2), a data processing unit 130 can be used to identify statistical and causal relationships between the data in the database 110 and generate or assign known statistical models 33 and / or physico-chemical models 34 for modeling these relationships (see FIG. 13). In one implementation, this can be done using artificial intelligence AI methods, such as machine learning ML or deep learning DL.

[0094] As shown in FIG. 2, in the eighth step S8, a research goal, such as the production of a material and / or its integration into a device, can be identified using the data processing unit 130 and based on the data completed, unified, and enriched in the preceding steps S5 and S6 and / or the statistical, causal, and physicochemical relationships identified and modeled in the preceding step S7. Additionally, work steps and activities relevant to achieving the goal, along with their interdependencies, can be identified within the data. In particular, work steps and activities that are not required to achieve the goal can also be identified using the identified statistical and causal relationships. Based on this, an optimized workflow can be identified and / or generated. This can be done, for example, using artificial intelligence (AI) methods, particularly machine learning (ML) or deep learning (DL).

[0095] As mentioned above, practical work in the third phase, Phase III, typically requires the use of highly specialized equipment and devices, which may also be external to the organization conducting the research. In a ninth step S9 (see FIG. 2 ), information about the availability of terminals, equipment, systems, and / or other resources 170 can be localized, captured, and stored in the graph database 110, e.g., by the data acquisition unit 150. Such information and data can be stored, e.g., in the networked terminals 170 and / or in the management unit 210 assigned to the terminals, e.g., in a database, PC, server, etc. For example, physicochemical data on all catalytic materials used in hydrogen production can be stored in a so-called data lake. In this way, it is possible to restrict the use of all or part of the materials stored in the data lake to internal and / or external personnel, and / or the terms of use can be defined, communicated, and managed. Localization, capture, and storage in the graph database 110 facilitates planning in the second phase, Phase II, and efficient, preferably automated, subsequent use in the third phase, Phase III. The stored information can be used both in ongoing research projects and in further, subsequent research projects.

[0096] In a tenth step S10 (see FIG. 3 ), an appropriate workflow can be selected to achieve a research goal corresponding to the selected research goal. This step S10 can preferably be performed in an automated manner by the execution unit 140. In a next step S11, the next work step to be performed is selected. If the next work step is successfully determined, the execution unit 140 can select, and in a preferred implementation start, a terminal 170 in a next step S12 to perform the work step or activity of the work step. The selection of the terminal 170 can be made depending on the type, range, and / or time of the work step to be performed, and on the fulfillment of the type, range, time, and / or availability conditions of the terminal 170.

[0097] Additionally, when selecting the terminal 170 in S12, a label can be created for the terminal 170 and the work step to be performed. Such a label can also be captured and stored by the data acquisition unit 150 in a later step (see S16) when capturing the result data. This allows for a clear identification of the origin and type of data formation. Such a label can be, for example, a unique coding in the form of an alphanumeric code, a barcode, or a QR code. Similarly, such a label can include, for example, an embedded memory chip and a simple implementation function for ensuring data quality and data curation, for example in the case of physical material to be shipped. The implementation function can be, for example, based on a simple AI algorithm embedded in the memory chip.

[0098] In the next step S13, the selected terminal 170 can perform the selected work step or activity. As explained above, the terminal 170 can generate a very large amount of raw data as a result. This raw data can then be analyzed and processed in the assigned edge / fog units 180, 190 in the subsequent steps S14, S15. The raw data / input data is discarded after the analysis is completed, and only the result data is transferred. This helps to efficiently use the capacity of the communication network 200 and avoid bottlenecks and delays. Furthermore, it also helps to remove noise from the raw data in an improved manner before it is sent to the data acquisition unit 150 and structurally captured and stored by the data acquisition unit 150 in the next step S16.

[0099] In a subsequent step S17, similar to the sixth step S6 described above, the data can be supplemented, unified and enriched by the data processing unit 130. Similarly, data for evaluating the results of a work step, e.g. experimental, manufacturing, simulation or metrology results, can be verified, validated and / or validated and characterized by the data processing unit 130. To do this, and similar to the seventh step S7 described above, relationships can be identified and statistically modeled 33 or assigned to a statistical model 33 and / or a causal model 34 (see FIG. 13 ).

[0100] The selected workflow may not necessarily predict and / or encompass all possible outcomes and conclusions and the work steps that result therefrom. This may result in increased effort, increased time, suboptimal decisions, and / or suboptimal research and development results. Therefore, in a subsequent step S19, similar to the eighth step S8 described above, each executed work step and workflow is analyzed, and the results of this analysis may be used, for example, in a pre-trained AI algorithm 31, to suggest and / or generate new work steps and / or workflows for the next execution. In this way, new work steps and / or workflows may be proposed with each iteration.

[0101] Step S11 can then be performed again to determine the next work step. The selection of the next work step can then vary depending on the previous work step and / or its results. For example, the process of manufacturing a catalyst layer involves selecting precursor materials, e.g., solvent, catalyst, ionomer medium, followed by specific mixing conditions, e.g., pH and temperature, and finally specific characterization before the coating process. Therefore, for a given purpose, such as manufacturing a catalyst layer, a specific optimized workflow can be selected, based on which the work steps of the entire manufacturing process can be automatically selected and executed. In this case, the relevant data pipeline, i.e., the relevant measurement data and the communication channel 200 for transmitting results from the execution terminal 170, can be automatically retrieved for each step of the manufacturing process and included in the selected workflow.

[0102] If the next work step can be determined in an automated manner by the execution unit 140 (S11a), a new cycle S12-S19 of selecting a terminal 170, performing the work step and evaluating the results is started. If the work step cannot be determined in an automated manner by the execution unit 140 (S11b), the generation and / or determination of the next work step can be performed manually by the research personnel in a further step S20 (S20a), and the cycle S12-S19 can be performed again. For such manual intervention, in this step S20, the content and relationships of the data stored in the graph database 110 can be visualized by the interactive human-machine interface 160 and selections and inputs can be captured. If the next step cannot be generated and selected manually (S20b), this may mean the achievement of the research goal and / or the (early) termination of the research project (S21).

[0103] The described visualizations, as exemplified in Figures 10 and 11, and the possibility of manual intervention may also be provided in the other steps described above. According to the invention, different data centers or instrument measurements 170 in a decentralized measurement center can be linked and mapped by different operators or personnel 18. This allows a timely and efficient (preliminary) evaluation of the data and results by expert researchers. The possibility of visualization and manual intervention is compatible with machine-to-machine communication (based on IoT) between the units of the research and development system according to the invention. FIG. 14 illustrates the process and architecture of an example embodiment involving semantic search using large language models (LLMs).

[0104] The data models and ontologies 16 used to label datasets are crucial components of a semantic search pipeline that uses large-scale language models 350. LLMs 350 are used to generate descriptions and alternative representations for ontology classes. These names and descriptions are used to generate embeddings 320, i.e., vector representations of human language. The generated embeddings 320 are linked to ontology classes in database 110.

[0105] To query the database 110, for example, for a manufacturing process, a user 300 must describe the structure of the desired process as a search query 310, such as materials, intermediates, products, parameters, properties, and manufacturing steps. For each portion of the search query 310, an embedding 320 is generated that is sent to the database 110 to find the closest embedding 320 in a set of ontology embeddings. The resulting matches are checked to see if there are any patterns 330 within them that match the described process or sub-process. The matching patterns and / or node patterns 330 are retrieved, and the data stored in the nodes 25 is parsed and converted into a predetermined output structure 340, such as a tabular or JSON format.

[0106] JSON (JavaScript Object Notation) refers to a standardized text-based format for representing structured data based on JavaScript object syntax.

[0107] In another exemplary embodiment, the data transfer is performed in an automated way, where technical tabular data is read (semi-)automatically by a pipeline, regardless of the table structure, terminology or size. To start the conversion process for the automatic data transfer, the tables are parsed and a dictionary with headings, some row examples and strings providing additional context is generated and transferred to the pipeline. The pipeline consists of a series of LLM instances linked together to solve the following tasks: Task 1, mapping table columns to node labels in database 110; Task 2, identifying the column headers and node attributes found in the cells; Task 3, combining columns that need to be mapped to the same node, for example, column 1 with the header MaterialA_name and column 2 with the header MaterialA_ID, with two attributes of the same node; and Task 4: Deriving relationships, i.e., semantic connections between nodes extracted from the sequence.

[0108] (Semi-)automated data transfer is crucial, as materials science data assets are often small, individual tables, and automating their compilation is a necessary step to create powerful training datasets. In addition, pipelines enrich and maintain data by mapping terms in tables to ontologies16, which increases interoperability. Furthermore, pipelines facilitate data interconnection, which is even more valuable as it allows for the generation of highly specific training datasets.

Claims

1. A research and development system (100) suitable for research and / or development of products and methods for manufacturing products, particularly energy materials, comprising a database (110) and an interface (120) for inputting and outputting data, a data processing unit (130), and an execution unit (140); The database (110) is a graph database configured to store data according to a data model that maps to a well-defined ontology (16); The data processing unit (130) unifying, supplementing, and / or enriching data stored in the graph database (110); Identifying statistical and / or causal relationships between the data in the database and modeling these relationships with statistical models (33) and / or physico-chemical models (34); and / or Identifying research and development goals within the stored data and / or generating and / or adapting appropriate workflows and / or work steps to achieve the research and development goals It is configured as follows: The execution unit (140) Selecting a workflow to achieve the development goals according to the development goals, and / or Selecting a next work step from a series of work steps of a selected workflow, in particular depending on the preceding work step and / or the result of the preceding work step A research and development system that is structured as follows:

2. The research and development system (100) of claim 1, wherein the execution unit (140) is configured to select a terminal (170) for performing a work step, particularly depending on the type, scope and / or time of the work step to be performed and / or the fulfillment of the type, scope, time and / or availability conditions of the terminal (170).

3. The research and development system (100) localizing, capturing, and / or storing in a graph database (110) information related to the selected research goal, in particular expert articles, publications, test series, lectures, comments, and / or other related records and / or documents; and / or Localizing the results of the performance of the work steps by the terminal (170), capturing the results, labeling the results with respect to origin and origin, and / or storing the results in the graph database (110).

3. The research and development system (100) of claim 2, comprising a data acquisition unit (150) configured to:

4. The research and development system (100) Information about the type, range, and / or availability and / or non-availability of the terminal (170); and / or Information about the technical, operational, legal, contractual and / or other availability conditions of the terminal (170) and the status of fulfillment of the availability conditions and / or Storing the information in a graph database (110) The research and development system (100) according to any one of claims 1 to 3, comprising a data acquisition unit (150) configured to:

5. The research and development system (100) graphically displaying the results of the work steps and / or work steps of the workflow; Indicating workflows, work steps, and availability of devices for performing work steps; and / or receiving input from a human user to select a workflow to be performed, a work step to be performed, and / or a terminal (170) for performing the work step; The research and development system (100) according to any one of claims 1 to 4, comprising an interactive human-machine interface (160) configured to:

6. The research and development system (100) according to any one of claims 1 to 5, which in a preferred embodiment comprises an IoT-enabled server, a cloud system, a terminal (170), an edge computing unit (180) and / or a fog computing unit (190).

7. 7. The research and development system (100) according to any one of claims 1 to 6, wherein the graph database (110), the data processing unit (130), the execution unit (140), the data acquisition unit (150), the terminal (170), the edge computing unit (180), the fog computing unit (190) and / or the interactive human-machine interface (160) are configured to apply methods of artificial intelligence (AI), in particular machine learning (ML) and / or deep learning (DL), and / or to provide results in real time.

8. A method for operating a research and development system (100) according to any one of claims 1 to 7, comprising steps a) to k): a) storing data in a graph database (110) according to a data model representing a well-defined ontology (16); b) unifying, supplementing, and / or enriching the data stored in the graph database (110); c) identifying statistical and / or causal relationships between the data stored in the graph database (110) and / or modeling these relationships using statistical models (33) and / or physico-chemical models (34); d) identifying research and development objectives contained in the stored data and / or generating workflows and / or work steps suitable for achieving the research and development objectives; e) selecting a workflow for achieving the development goal as a function of the development goal and / or selecting a work step from the sequence of work steps of the selected workflow, in particular as a function of the preceding work step and / or the result of the preceding work step; f) selecting a terminal (170) for performing a work step, in particular depending on the type, extent and / or time of the work step to be performed and / or the fulfillment of the type, extent, time and / or availability conditions of the terminal (170); g) analyzing the results of the work steps by the edge computing unit (180) and / or the fog computing unit (190) and / or forwarding the analysis results as results of the work steps; h) capturing, labeling with respect to origin and provenance, and / or storing the results of the work steps in a graph database (110); i) localizing, capturing and / or storing in a graph database (110) information related to the research goal, in particular expert articles, publications, test series, lectures, comments and / or other related records and / or documents; j) localizing, capturing, and / or storing in the graph database (110) information about the type, range, availability and / or unavailability of the terminals (170), about technical, operational, legal, contractual, and / or other conditions related to the availability of the terminals (170), and / or about the fulfillment of the availability conditions; k) graphically displaying the results of the work steps, the availability of certain work steps of the workflow, the workflow, the work steps and / or the terminals for performing the work steps; and / or receiving from a human user a selection of a workflow or work step to be performed and / or a terminal for performing the work step; and performing said steps preferably using artificial intelligence methods and / or in real time.

Images