Automated-reconciliation method for the life cycle inventory data of an industrial product or process
The automated reconciliation method addresses inefficiencies in LCI data by using a knowledge base, reconciliation tree, and semantic algorithm to match primary and secondary data, enhancing LCA study efficiency and consistency.
Patent Information
- Application Number
- PCT/EP2026/062983
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2025-05-06
- Filing Date
- 2026-05-05
- Publication Date
- 2026-11-12
AI Technical Summary
Current methods for reconciling life cycle inventory (LCI) data in life cycle assessment (LCA) are inefficient, prone to human bias, and lack consistency, hindering large-scale adoption and integration of LCA practices across industries due to manual interventions and insufficient automated solutions.
An automated reconciliation method integrating a comprehensive knowledge base, structured reconciliation tree, and semantic algorithm to match primary and secondary data, with confidence scoring and iterative learning to ensure accuracy and consistency.
This approach enhances the efficiency, consistency, and scalability of LCA studies by reducing processing time, minimizing human error, and improving data interoperability, while ensuring transparent and reproducible results.
Smart Images

Figure EP2026062983_12112026_PF_FP_ABST
Abstract
Description
[0001] AUTOMATED RECONCILIATION METHOD FOR INVENTORY DATA FROM THE LIFECYCLE OF A PRODUCT OR INDUSTRIAL PROCESS
[0002] Scope of the invention
[0003] The present invention relates to automated methods and systems for reconciling life cycle inventory data in life cycle analysis studies of an industrial product or process.
[0004] Previous technique
[0005] Life cycle assessment (LCA) has become an increasingly important tool for evaluating the environmental impacts of products and processes throughout their lifecycle. However, the LCA process remains complex and time-consuming, particularly in the life cycle inventory (LCI) phase, where primary product data—defined as information specific to the product or process under study—must be reconciled, or harmonized and matched, with secondary data. Secondary data can be defined as generic information on the environmental impacts of the materials, processes, and activities used in an LCA. This information is not specific to the product or system being studied but provides context and baseline values for the LCA.Current methods for reconciling LCA data often rely heavily on manual interventions by LCA experts, leading to inefficiencies, potential inconsistencies, and limitations in scaling up LCA studies.
[0006] Primary data, for example, corresponds to information collected from stakeholders directly involved in the life cycle of the product or industrial process under study. This data generally comes from measurements, production reports, or internal company databases and relates to parameters specific to the analysis system of that product or industrial process. This data may include information on the quantities of raw materials used, energy consumption, direct emissions, transport distances, or recycling rates specific to the product or industrial process being studied. In some cases, primary data may also come from interviews with experts in the field or from questionnaires completed by suppliers. The collection and use of primary data can provide a more accurate and specific representation of the industrial system or process being studied in the life cycle assessment.
[0007] Secondary data can be defined as generic information from existing LCA databases or scientific literature. It relates to LCIs that are not directly under the control of the actor performing the LCA but may be necessary for modeling. This secondary data can provide information on the environmental impacts of materials, industrial processes, and activities commonly used in various industrial sectors. In some cases, it can serve as baseline or substitute values when specific primary data are unavailable. The use of secondary data can complement the life cycle assessment by providing broader context and filling gaps in the collected primary data.
[0008] Semantic analysis techniques have been proposed to link primary and secondary data, but these methods may be insufficient when dealing with incomplete databases or selecting appropriate surrogate data when exact matches are unavailable. Prior art has relied on manual methods for reconciling lifecycle inventory data. However, these methods have several limitations. They can be inefficient for handling large volumes of complex data, but more importantly, they introduce inaccuracies due to human bias and a lack of consistency in LCAs, since each expert, in addition to their biases, possesses their own experience and perspective.In some cases, predefined proximity criteria can be used to identify substitute data, but these criteria may be limited and may not take into account the full complexity of the relationships between materials and industrial processes.
[0009] The lack of standardized approaches to address data gaps, such as missing or incomplete data on materials or industrial processes, can pose a significant challenge in the field of automated LCI modeling. Selecting appropriate substitutes when exact matches are unavailable and ensuring consistency in the application of modeling methods are also aspects to consider.
[0010] This gap hinders the ability to conduct large-scale LCA studies effectively and consistently, limiting the potential for widespread adoption of LCA practices across industries and businesses. Furthermore, the lack of robust knowledge management frameworks integrated into LCA data reconciliation processes restricts the ability to leverage insights and learnings across multiple products and studies.
[0011] There are state-of-the-art solutions that attempt to address similar problems. For example, EP 4 332 856 A1 describes a system and a method for recommending actions to promote a circular economy framework. This system uses an ontology-based knowledge graph to store semantic information about the properties and behavior of a product's components at each phase of its life cycle. From this graph, which is enriched with data on physical and chemical properties, as well as regulatory compliance information, such a system determines performance indicators. Based on these indicators, it recommends actions, such as choosing a supplier or a material, to improve circularity.This system can also infer new information from existing data, for example, by deducing that a supplier has a higher carbon footprint due to its geographical distance. CN 116 894 108 A describes a method for determining a life cycle inventory (LCI). This method receives a request containing information about a product, extracts "level data" (such as the grade or standard of a material), and uses this data to query a pre-existing knowledge graph. Such a knowledge graph, according to CN 116 894 108 A, links this level data with LCI data, thus enabling the retrieval of the life cycle inventory corresponding to the specified material. However, these approaches have significant limitations.They do not offer an architecturally structured solution that distinctly combines a rich knowledge base and a procedural decision tree for the systematic selection of substitutes. Furthermore, they do not incorporate an iterative learning mechanism to refine the relevance of reconciliations over time based on expert feedback, nor a system to quantify the reliability of each match with a confidence score, thus leaving a need for a more robust, transparent, and self-improving reconciliation process.
[0012] It has been recognized that an automated reconciliation method for lifecycle inventory data is needed to overcome one or more of these problems.
[0013] The invention thus aims to address at least partially the technical problems presented above by providing an automated reconciliation method for lifecycle inventory data that integrates a comprehensive knowledge base, a structured reconciliation tree and a semantic algorithm to streamline the LCI process and enable more efficient, consistent and scalable lifecycle analyses.
[0014] The invention thus aims to address at least partially the technical problems presented above.
[0015] Description of the invention
[0016] The invention relates in particular to an automated reconciliation method for Life Cycle Inventory (LCI) data of a product or industrial process within the framework of a Life Cycle Assessment (LCA). The computer-implemented method comprises the following steps: accessing a knowledge base stored in the computer's memory, where the knowledge base contains mixed data on the environmental impacts of industrial materials and processes. The method also includes a step of constructing a reconciliation tree designed to automate decisions based on pre-established criteria. The method further includes a step of implementing a semantic algorithm that interacts with the knowledge base and the reconciliation tree. This interaction enables the matching of primary data with appropriate secondary data, both derived from the mixed data.Primary data are specific to the product or industrial process under study, while secondary data are generic data concerning the environmental impacts of industrial materials and processes. Although not specific to the product or industrial process in question, these secondary data provide essential context and reference values for the LCA. The process also includes a step to select alternative data based on semantic algorithm matches when exact matches between primary and secondary data are not available.
[0017] The process further includes a step of providing an ICV deliverable, which includes reconciled data, associated confidence scores, and detailed documentation of the substitutions made. This ICV deliverable, generated by the computer running the process, is stored in non-transient memory and includes specific data structures enabling complete traceability of reconciliation decisions. The deliverable facilitates the identification of potential areas for improvement in primary data collection via electronic markers and ensures the reproducibility of environmental analyses through a standardized data format.The standardized deliverable structure, implemented as an XML or JSON schema, for example, also improves interoperability with different LCA software, significantly reduces the processing time required for comparative analyses between similar products, and enables better communication of results to stakeholders through a clear presentation of data sources and methodologies used. Furthermore, automating the reconciliation process eliminates the human biases inherent in manual methods, thus ensuring greater objectivity and consistency in LCA results, regardless of the expert performing the analysis.
[0018] Mixed data can include technical data sheets for materials and industrial processes containing detailed information such as a name, synonyms, a brief description, and a set of physical, thermal, and mechanical properties. These data sheets may also include information on the applications, criticality, and recycling of the material or industrial process. One aspect of mixed data is mapping these data sheets to corresponding environmental data, enabling the linking of technical characteristics to environmental impacts. This mixed data structure facilitates comprehensive and contextual analysis within life cycle assessment studies, providing a rich foundation for the automated reconciliation of inventory data.
[0019] A knowledge base is a structured collection of information that combines various types of data to support decision-making and analysis. Material and industrial process datasheets are documents containing detailed information about specific materials or industrial manufacturing processes. Technical characteristic data refers to quantitative and qualitative attributes of materials or industrial processes.
[0020] An ontology is a formal representation of a set of concepts within a domain and the relationships between them. An ontology can include definitions and relationships between different materials, industrial manufacturing processes, environmental impacts, and other concepts relevant to LCA. For example, ontologies can define material hierarchies (such as metals, alloys, and their subcategories), relationships between industrial manufacturing processes and their impacts, or links between different types of environmental data. These structured ontologies can help organize and connect information within the knowledge base, thereby facilitating the automated reconciliation of lifecycle inventory data.
[0021] A reconciliation tree is a hierarchical structure used to organize and automate decision-making. A semantic algorithm is a computational technique that analyzes and interprets the meaning of data, using, for example, natural language processing techniques.
[0022] This method according to the invention offers several advantages, including increased efficiency in performing life cycle assessments by automating the time-consuming data reconciliation process, improved consistency in data selection across multiple studies, and the ability to handle large volumes of data for complex products, systems, or industrial processes. The combination of a comprehensive knowledge base with an intelligent reconciliation system enables more accurate and reliable matching of primary and secondary data, even when exact matches are unavailable. Indeed, the method described in the invention is more efficient than using a general artificial intelligence or model prediction approach for LCI data reconciliation due to its specialized design.It incorporates a custom knowledge base and a structured reconciliation tree specifically designed for lifecycle inventory data, enabling more accurate and contextual matching. This domain-specific approach leverages pre-established semantic relationships and decision-making criteria relevant to materials, industrial processes, and environmental impacts. Unlike general AI, which would need to learn these complex relationships from scratch, this process can immediately apply relevant industry knowledge and established LCA practices. Furthermore, the structured nature of the process ensures consistency, transparency, and easier integration with existing LCA software—something that might not be as readily achievable with a more opaque AI system.Targeted design also allows for targeted improvements based on expert feedback, potentially leading to faster refinement of the reconciliation process compared to training a general AI for that specific task.
[0023] According to one embodiment, the mixed data may include at least technical data sheets for materials and industrial processes including technical specifications, physical properties and manufacturing parameters; environmental data corresponding to the materials and industrial processes; and ontologies defining relationships and hierarchies between materials, industrial processes and environmental impacts.
[0024] This comprehensive knowledge base structure enables a more nuanced and precise reconciliation process by providing detailed information on materials and industrial processes, their environmental impacts, and the relationships between the various elements. The inclusion of technical specifications and manufacturing parameters allows for a more accurate matching of primary data with relevant secondary data.
[0025] According to one embodiment, the construction of the reconciliation tree includes defining a hierarchical structure to organize industrial materials and processes; establishing semantic relationships between data elements; and determining proximity criteria to select substitute data when exact matches are not available.
[0026] The hierarchical structure and semantic relationships within the reconciliation tree provide a framework for systematic decision-making in the data reconciliation process. Proximity criteria enable intelligent selection of substitute data when exact matches are not found, ensuring that the closest appropriate match is chosen based on predefined rules and relationships.
[0027] According to one embodiment, the implementation of the semantic algorithm includes the analysis of textual descriptions and metadata of primary product data; the comparison of the analyzed data with the knowledge base entries using natural language processing techniques; and the ranking of potential matches based on semantic similarity and defined proximity criteria.
[0028] This semantic approach to data reconciliation enables smarter, more context-sensitive matching of primary and secondary data. By analyzing textual descriptions and metadata, the algorithm can understand the meaning and context of the data, rather than relying solely on exact keyword matches. Ranking potential matches based on semantic similarity and proximity criteria ensures that the most appropriate secondary data is selected, even when exact matches are unavailable.
[0029] According to one embodiment, the process may further include a refinement step comprising the following substeps: comparing past selections of the process with expert selections and modifying the reconciliation tree and the semantic algorithm based on the comparison; and updating the knowledge base.
[0030] The validation and iterative refinement steps allow the process to continuously improve its accuracy and reliability over time. By incorporating expert feedback and updating the knowledge base and algorithms, the process can adapt to new data, evolving industry standards, and changing methodologies for assessing environmental impacts.
[0031] According to one embodiment, the process further includes a step of generating a confidence score for each data reconciliation or substitution, and flagging each reconciliation or substitution having a low confidence match.
[0032] Confidence scoring provides a measure of reliability for each automated match or substitution, allowing users to quickly identify areas that may require further review. Flagging low-confidence matches for review ensures that human expertise is applied where it is most needed, balancing automation with quality control.
[0033] According to one embodiment, the generation of the confidence score is based on the semantic similarity between the primary and selected secondary data; the number of matching criteria in the reconciliation tree; and the availability and quality of surrogate data when exact matches are not found.
[0034] This multidimensional approach to confidence scoring provides a comprehensive assessment of the quality and reliability of each automated match or substitution. By considering semantic similarity, matching criteria, and the quality of the substitution data, the process can provide users with a nuanced understanding of the confidence level for each reconciliation decision.
[0035] According to one embodiment, the process is implemented within an LCA software platform to automate the process of reconciling LCI data for the performance of LCA studies.
[0036] Implementing the process within an LCA software platform allows for seamless integration of the automated reconciliation process into existing workflows. This integration can significantly reduce the time and effort required to conduct LCA studies, particularly for complex products or systems with large amounts of data.
[0037] In one embodiment, the LCA software platform is configured to import primary product data from various sources and formats; access multiple secondary LCI databases; and export reconciled data. This configuration enhances the flexibility and usability of the automated reconciliation process by enabling it to work with diverse data sources and formats. The ability to access multiple secondary LCI databases increases the likelihood of finding suitable matches or substitutes for the primary data. Exporting reconciled data in a format suitable for subsequent analysis and impact assessment streamlines the overall LCA process.
[0038] The invention also relates to an automated reconciliation device for Life Cycle Inventory (LCI) data of a product or industrial process within the framework of a Life Cycle Assessment (LCA). The device comprises a knowledge base stored in memory, said knowledge base comprising mixed data, as previously described; a reconciliation tree designed to automate decisions based on pre-established criteria;and a processor configured to implement a semantic algorithm interacting with the knowledge base and reconciliation tree to match primary data from the mixed data with appropriate secondary data from the mixed data, such that the primary data is data specific to the product or industrial process under consideration and the secondary data is generic data relating to the environmental impacts of materials and industrial processes which are not specific to the product or industrial process under consideration but provide context and reference values for the LCA, and to select substitute data based on the matches of the semantic algorithm when exact matches between primary and secondary data are not available.
[0039] A processor is a component of a computer system that executes the instructions of a computer program by performing basic arithmetic, logical, control, and input / output operations specified by the instructions. In this context, the processor is specifically configured to implement the semantic algorithm and manage the interactions between the knowledge base and the reconciliation tree. This device provides a complete and integrated solution for automating the reconciliation process of LCI data. By combining a rich knowledge base, a structured decision-making framework (the reconciliation tree), and an intelligent semantic algorithm, the device can efficiently and accurately match primary product data with the appropriate LCI data.This automation reduces the time and effort required for manual data reconciliation, while improving consistency and reducing the potential for human error.
[0040] According to one embodiment, the mixed data may include at least technical data sheets for materials and industrial processes including technical specifications, physical properties and manufacturing parameters; environmental data corresponding to the materials and industrial processes; and ontologies defining the relationships and hierarchies between materials, industrial processes and environmental impacts.
[0041] This comprehensive knowledge base structure provides a rich foundation for the automated reconciliation process. The inclusion of detailed technical specifications, physical properties, and manufacturing parameters enables a more accurate matching of primary data with relevant secondary data. ICV data and ontologies allow the system to consider environmental impacts and the relationships between different elements when making reconciliation decisions.
[0042] According to one embodiment, the reconciliation tree includes a hierarchical structure for organizing materials and industrial processes; semantic relationships between data elements; and proximity criteria for selecting substitute data when exact matches are not available.
[0043] This structured approach to decision-making within the reconciliation tree enables systematic and consistent data reconciliation. The hierarchical organization and semantic relationships provide a framework for navigating complex data relationships, while proximity criteria can enable intelligent selection of substitute data when exact matches are not found.
[0044] According to one embodiment, the semantic algorithm is configured to analyze the textual descriptions and metadata of the primary product data; compare the analyzed data to the knowledge base entries using natural language processing techniques; and rank the potential matches based on semantic similarity and defined proximity criteria.
[0045] This semantic algorithm configuration enables intelligent and contextual data reconciliation. By analyzing textual descriptions and metadata, the algorithm can understand the meaning and context of the primary data, leading to a more accurate match with the secondary data. Ranking potential matches based on semantic similarity and proximity criteria ensures that the most appropriate secondary data is selected, even when exact matches are unavailable.
[0046] According to one embodiment, the device further includes a validation module configured to iteratively validate and refine the automated reconciliation process by comparing automated selections with expert judgments; adjusting the reconciliation tree and the semantic algorithm based on feedback; and updating the knowledge base with new data and relationships.
[0047] The inclusion of a validation module allows the system to improve its performance over time. By comparing automated selections with expert judgments and incorporating feedback, the system can learn from its mistakes and adapt to new data and changing requirements. This iterative refinement improves the accuracy and reliability of the automated reconciliation system.
[0048] According to one embodiment, the validation module is further configured to generate a confidence score for each automated reconciliation or data substitution based on semantic similarity, the number of matching criteria and the availability of substitution data; and to flag low confidence matches.
[0049] This validation module configuration provides an additional layer of quality control to the automated reconciliation process. By generating confidence scores and flagging low-confidence matches, the system can identify areas where human expertise may be needed to ensure reconciliation accuracy. This approach balances the efficiency of automation with the need for human oversight in complex or ambiguous cases.
[0050] The various non-incompatible aspects defined above can be combined.
[0051] Brief description of the drawings
[0052] The embodiments of the invention will be described below with reference to the drawings, briefly described below:
[0053] - Figure 1 is a flowchart illustrating an automated data reconciliation process, according to one embodiment of the present invention; and
[0054] - Figure 2 is a schematic representation of an automated reconciliation device, according to one embodiment of the present invention.
[0055] Detailed description of the invention
[0056] The present invention relates to an automated method for reconciling Life Cycle Inventory (LCI) data for a product or industrial process within the framework of a Life Cycle Assessment (LCA). This method aims to improve the efficiency and consistency of the LCI data reconciliation process, which can be complex and time-consuming when performed manually.
[0057] The proposed method uses a systematic approach that combines several key elements. It relies on a comprehensive knowledge base, stored in computer memory, which includes mixed data on the environmental impacts of industrial materials and processes. A reconciliation tree is constructed to automate decisions based on pre-established criteria. A semantic algorithm interacts with the knowledge base and the reconciliation tree to match primary data with appropriate secondary data.
[0058] This process effectively handles cases where exact matches between primary and secondary data are unavailable by selecting relevant substitute data. It also incorporates validation and continuous improvement mechanisms, as well as a confidence scoring system for each reconciliation or substitution.
[0059] This process streamlines the LCA process, enabling more efficient, consistent, unbiased, and scalable life cycle assessments. This automated approach can significantly reduce the time and resources required to conduct LCA studies, while simultaneously improving the quality and reliability of the results.
[0060] As illustrated in Figure 1, the automated reconciliation process for lifecycle inventory data leverages a knowledge base that plays a central role in the process. Figure 1 illustrates step 102 of accessing the knowledge base, which initiates interaction with this structured database.
[0061] The knowledge base combines diverse data from multiple sources. It includes detailed technical data sheets on materials and processes, providing precise information on their properties and characteristics. For example, for a material like aluminum, the data sheet includes data on its density, thermal conductivity, mechanical strength, and other relevant physicochemical properties. Similarly, for a manufacturing process such as injection molding, the data sheet describes the industrial process parameters, compatible materials, and the characteristics of the parts produced. In addition to these data sheets, the knowledge base integrates environmental data associated with each material and process.This environmental data comes from recognized life cycle assessment databases and provides information on potential environmental impacts throughout the life cycle of materials and processes. For example, for aluminum, the environmental data includes greenhouse gas emissions related to its extraction, production, and recycling.
[0062] The knowledge base also incorporates data relating to the technical characteristics of materials and processes. This information allows for a thorough understanding of the properties and performance of the various elements, thus facilitating the comparison and selection of the most appropriate alternatives when reconciling data.
[0063] One advantage of a knowledge base is its ability to efficiently manage both primary and secondary data. Primary data, specific to the product or industrial process under study, is collected directly from stakeholders involved in the product's lifecycle. This data can include precise information on the quantities of raw materials used, specific energy consumption, or measured direct emissions. Secondary data, on the other hand, is generic information from standardized environmental databases that provide broader context and reference values for the LCA. The knowledge base organizes these two types of data and facilitates their matching during the reconciliation process.
[0064] Another advantage of the knowledge base lies in the integration of ontologies that structure information hierarchically and relationally. These ontologies define the relationships between different concepts, materials, and processes, creating a semantic network that facilitates navigation and querying of the database. For example, an ontology establishes links between different classes of materials, their subcategories, and their specific properties, enabling efficient searching and comparison of potential alternatives.
[0065] The knowledge base is organized hierarchically, reflecting the classification of materials and industrial processes. This structure allows for intuitive navigation and efficient searching of relevant information. The data is interconnected, linking materials, industrial processes, and their respective environmental impacts. This interconnectedness facilitates the rapid identification of potential alternatives when reconciling inventory data.
[0066] The purpose of this knowledge base is to provide a centralized and reliable source of information for the automated reconciliation process. It serves as a reference for comparing and selecting appropriate data, thereby reducing the need for manual intervention and improving the consistency of lifecycle analyses. The richness and structure of the data contained in the knowledge base enable informed and transparent decision-making during inventory data reconciliation, thus contributing to the efficiency and reliability of the automated lifecycle analysis.
[0067] Figure 1 illustrates an automated data reconciliation process, which includes a step of building a reconciliation tree 104. The reconciliation tree plays a central role in automating lifecycle inventory data reconciliation decisions.
[0068] The reconciliation tree presents a hierarchical structure that logically organizes materials and processes. For example, materials are classified into categories such as metals, polymers, ceramics, etc. Each category is then divided into subcategories and specific materials. Similarly, processes are organized into families such as forming, heat treatments, or surface treatments.
[0069] This hierarchical structure facilitates navigation and matching between primary and secondary data. The reconciliation tree also defines semantic relationships between data elements. These relationships include links such as "is a type of", "belongs to", or "similar to", allowing for logical connections between different elements.
[0070] The reconciliation tree incorporates proximity criteria to select substitute data when an exact match is not available. These criteria take into account technical characteristics, physical properties, or functional similarities between materials or processes. For example, if a specific aluminum alloy is not present in the database, the reconciliation tree uses these criteria to identify the closest alloy in terms of composition and properties.
[0071] In one embodiment, the proximity criteria for selecting surrogate data may include both elimination and comparative criteria. The elimination criteria may include:
[0072] - risk indicators of regulated substances, such as those defined by the SI N (Substitute It Now) lists or the REACH regulation;
[0073] - an application or end use of the material or industrial process;
[0074] - a family of materials to which the element in question belongs; and
[0075] - an overall environmental impact.
[0076] These elimination criteria allow for the rapid filtering of irrelevant options or options that do not comply with regulatory and functional requirements.
[0077] Comparative criteria, meanwhile, may include:
[0078] - the density of the material;
[0079] - a Young's modulus, which characterizes the elastic properties; - a maximum operating temperature;
[0080] - compatibility with existing manufacturing processes; and - resource availability.
[0081] These comparative criteria allow for a more precise selection among the remaining options after the elimination criteria have been applied. They provide a quantitative basis for comparison to identify the most suitable substitutes in terms of physical properties, performance, and industrial feasibility.
[0082] The combined use of these elimination and comparison criteria in the reconciliation tree can enable a more precise and relevant selection of surrogate data. This structured approach can improve the quality and reliability of automated reconciliations, particularly in cases where exact matches are not available in the database.
[0083] By automating decisions based on these pre-established criteria, the reconciliation tree reduces the need for manual intervention and ensures consistency in the data reconciliation process. The tree guides the system in selecting the most appropriate matches, taking into account the hierarchical structure, semantic relationships, and defined proximity criteria.
[0084] The semantic algorithm plays a key role in the automatic reconciliation process of lifecycle inventory data. This algorithm interacts dynamically with the knowledge base and reconciliation tree to establish relevant matches between primary and secondary product data.
[0085] The process begins with a data matching step (106), where the semantic algorithm analyzes the information provided in the product's primary data. The algorithm then compares this data to the elements stored in the knowledge base, using the semantic relationships and hierarchical structure defined in the reconciliation tree. This approach allows for the rapid identification of potential matches with minimal computational load, even when the terms used are not exactly identical.
[0086] For example, if the primary data mentions an "aluminum alloy 6061-T6", the algorithm searches the knowledge base for related terms such as "aluminum", "alloy", "6000 series", etc. The algorithm then uses the reconciliation tree to navigate the material hierarchy and identify the most relevant matches in the lifecycle inventory database.
[0087] In cases where an exact match is not found, the algorithm proceeds to a substitution data selection step. This step relies on the proximity criteria defined in the reconciliation tree to identify the most suitable alternatives. The algorithm evaluates technical properties, chemical composition, and other relevant characteristics to select a substitute that most closely resembles the original material or process.
[0088] For example, if the specific material "aluminum alloy 6061-T6" is not present in the life cycle inventory database, the semantic algorithm searches for alternatives in the same family of aluminum alloys, prioritizing those with similar mechanical properties and chemical composition.
[0089] The semantic algorithm also ensures the consistency of selections by verifying that the units, scales, and application contexts are compatible between the selected foreground and background data. This verification reduces the risk of errors in lifecycle inventory modeling.
[0090] By continuously interacting with the knowledge base and the reconciliation tree, the semantic algorithm refines its selections over time. The results of each reconciliation are used to update and enrich the knowledge base, thereby improving the accuracy and efficiency of future matches.
[0091] The validation and refinement process for the automatic reconciliation of primary and secondary data relies on an iterative approach. A comparison of the selections is performed between the automated choices of the system and the judgments of life cycle assessment experts. This step allows for the evaluation of the relevance and accuracy of the matches established by the semantic algorithm. Experts interact with the process by examining a representative sample of reconciliations, providing feedback on the quality of the selections, and contributing to the improvement of the decision criteria used in the reconciliation tree. Their evaluations are then integrated into the system's learning process to refine future reconciliations. Based on these comparisons, a modification of the tree and the algorithm is made.The reconciliation tree is adjusted to refine the selection criteria and semantic relationships between the data. The semantic algorithm is also optimized to improve its ability to identify the most relevant matches.
[0092] These changes result in update 114 of the knowledge base. The new information and relationships established during the validation process are integrated into the knowledge base, thus enriching the body of data available for future reconciliations.
[0093] This iterative cycle of comparison, modification, and update allows the system to continuously improve. The system's performance is refined over time, progressively narrowing the gap between automated selections and expert judgments. This approach ensures that the automated reconciliation system remains accurate and reliable, while adapting to evolving knowledge and practices in lifecycle analysis.
[0094] The automated data reconciliation process incorporates a confidence scoring step to assess the reliability of the established matches between primary and secondary data. This scoring step includes generating a confidence score for each reconciliation,116 of which helps identify potentially problematic matches. Confidence scores are calculated based on various factors, such as the semantic similarity between terms, the matching of technical attributes, and consistency with the ontological relationships defined in the knowledge base.
[0095] Once the confidence scores are generated, a step 118 flags low-confidence matches. This step automatically marks reconciliations whose confidence score falls below a predefined threshold. For example, a match between a specific aluminum alloy and a generic aluminum in the environmental database might receive a high confidence score, while a match between a complex polymer and a standard plastic might be flagged as having low confidence.
[0096] The confidence scoring and signaling system improves the reliability and transparency of the automated reconciliation process in several ways. First, it enables the rapid identification of matches requiring verification, thereby reducing the risk of errors in the life cycle assessment. Second, it provides a quantitative measure of match quality, which helps assess the overall reliability of the analysis results. Finally, by documenting confidence scores for each reconciliation, the system offers complete traceability of the data selection process, thereby enhancing the transparency and reproducibility of life cycle assessment studies.
[0097] Figure 1 illustrates the automated reconciliation process for ICV data. The process begins with access to the knowledge base, which contains the primary and secondary data necessary for the analysis. This knowledge base provides the informational foundation for the subsequent steps.
[0098] A reconciliation tree is then constructed. This tree structures the criteria and decision rules used for data reconciliation. The reconciliation tree guides the data matching and selection process throughout the analysis.
[0099] The process then includes a data matching step 106. This step compares the primary data specific to the product under study with the secondary data available in the knowledge base. The objective is to identify exact matches or the best possible matches.
[0100] When an exact match is not found, a selection of 108 substitute data points is performed. This step uses the reconciliation tree to choose the most appropriate data based on predefined criteria. A comparison of 110 of the selections is then carried out. This step evaluates the relevance and quality of the matches and substitutions made. The results of this comparison inform the subsequent steps of the process.
[0101] Based on these comparisons, a modification of the tree and the algorithm is made if necessary. This step allows for refining the decision criteria and improving the accuracy of the reconciliation process.
[0102] A knowledge base update 114 follows, incorporating new information and adjustments from previous steps. This update continuously enriches the database, thereby improving future reconciliations.
[0103] The process also includes a step 116 that generates a confidence score for each reconciliation. This score quantifies the reliability of the matches and substitutions performed, providing a measure of the quality of the reconciled data.
[0104] Finally, a 118-point flagging of low-confidence matches is performed. This step identifies potentially problematic reconciliations, allowing for revision if necessary.
[0105] The knowledge base stored in the computer's memory comprises mixed data on the environmental impacts of industrial materials and processes. This mixed data includes datasheets for materials and industrial processes containing detailed information such as a name, synonyms, a brief description, and a set of physical, thermal, and mechanical properties. These datasheets may also include information on the applications, criticality, and recycling of the material or industrial process. An important aspect of the mixed data is the inclusion of standardized environmental data from recognized LCA databases, which provide generic information on the environmental impacts of commonly used industrial materials and processes.This standardized data is mapped with corresponding technical data sheets, enabling the establishment of links between technical characteristics and environmental impacts. This mixed data structure, combining specific technical information and standardized environmental data, facilitates comprehensive and contextual analysis within the framework of life cycle assessment studies, providing a rich basis for the automated reconciliation of inventory data.
[0106] All of these steps together form a 100% iterative and self-improving process. Data flows from the knowledge base to the reconciliation tree, then through the matching, selection, and comparison stages. The results of these operations are then used to refine the reconciliation tree and update the knowledge base. This continuous flow of information and decisions enables constant improvement in the accuracy and efficiency of the automated lifecycle inventory data reconciliation process.
[0107] Figure 2 illustrates the automated reconciliation system 200 for Life Cycle Inventory (LCI) data of a product or industrial process. This system comprises several interconnected components that work together to implement the reconciliation process.
[0108] The automated reconciliation device 200 includes a memory 202 which stores the knowledge base 203. This knowledge base 203 can contain mixed data, including technical data sheets for materials and industrial processes, environmental data in the form of a standardized environmental database, and ontologies defining the relationships between the different elements.
[0109] The device also includes a processor 204 which is configured to perform reconciliation operations. This processor implements a semantic algorithm 205 which interacts with the knowledge base 203 and the reconciliation tree 208 to perform the mapping between primary and secondary data.
[0110] The reconciliation tree 208 is a data structure that organizes the decision criteria and rules used in the reconciliation process. It can be stored in memory 202 and used by the semantic algorithm to guide the data selection and substitution process. A validation module 206 is also included in the device. This validation module 206 is responsible for generating confidence scores for each reconciliation and flagging low-confidence matches. It can also play a role in the iterative improvement process of the automated reconciliation device 200 by comparing automated selections with expert judgments.
[0111] The components of the automated reconciliation device 200 are interconnected to allow for an efficient flow of data and information. For example, the semantic algorithm 205 can access the knowledge base 203 stored in memory 202, use the reconciliation tree 208 to guide its decisions, and transmit its results to the validation module 206 for evaluation.
[0112] This automated reconciliation device can offer an integrated solution to improve the efficiency and accuracy of the ICV data reconciliation process, reducing the need for manual interventions while maintaining a high level of quality and reliability in the results produced.
[0113] Although exemplary embodiments of the invention have been described, it will be understood by those competent in the field that various modifications, omissions, and / or additions may be made, and that equivalents may be substituted for elements thereof without departing from the spirit and scope of the invention. Furthermore, numerous modifications may be made to adapt a particular situation or material to the teachings of the invention without departing from its scope. Therefore, it is anticipated that the invention will not be limited to the particular embodiments disclosed for the implementation of this invention, but that the invention will include all embodiments within the scope of the appended claims. Moreover, unless otherwise indicated, any use of the terms first, second, etc., does not denote any order or importance, but the terms first, second, etc.are used to distinguish one element from another.
Claims
DEMANDS 1. A method (100) for the automated reconciliation of Life Cycle Inventory (LCI) data of a product or industrial process within the framework of a Life Cycle Assessment (LCA), said method being implemented by a computer (200) and comprising the following steps: - accession (102) to a knowledge base 203 stored in a memory (202) of the computer (200), such that said knowledge base (203) includes mixed data relating to the environmental impacts of industrial materials and processes; - construction (104) of a reconciliation tree 208 designed to automate decisions based on pre-established criteria; - matching (106) of data implementing a semantic algorithm (205) interacting with the knowledge base (203) and the reconciliation tree (208) in order to match primary data from the mixed data with appropriate secondary data from the mixed data, such that the primary data are data specific to the product or industrial process under consideration and the secondary data are generic data relating to the environmental impacts of materials and industrial processes which are not specific to the product or industrial process under consideration but provide context and reference values for the LCA;and - selection (108) of substitution data based on the matches of the semantic algorithm (205) when exact matches between primary and secondary data are not available.
2. Method (100) according to claim 1, wherein the mixed data comprise at least:; - technical data sheets for materials and industrial processes including technical specifications, physical properties and manufacturing parameters; - environmental data corresponding to industrial materials and processes; and - ontologies defining the relationships and hierarchies between materials, industrial processes and environmental impacts.
3. A method (100) according to any one of claims 1 or 2, wherein the construction of the reconciliation tree (208) comprises: - the definition of a hierarchical structure to organize materials by classifying them into categories and / or subcategories and industrial processes by classifying them into families; - re-establishing semantic relationships between elements, these relationships including links such as "belongs to", "similar to", or "is a type of"; and - the determination of proximity criteria to select surrogate data when exact matches are not available, these criteria being based on physical properties, technical characteristics or functional similarities.
4. A method (100) according to any one of claims 1 to 3, wherein the implementation of the semantic algorithm (205) comprises: - Analysis of textual descriptions and metadata of primary product data; - Comparison of the analyzed data with knowledge base entries using natural language processing techniques; and - the ranking of potential matches based on semantic similarity and defined proximity criteria.
5. A process (100) according to any one of claims 1 to 4, further comprising a refining step (120), the refining step (120) comprising the following substeps: - comparison (110) of past selections of the process with selections from trusted third parties; - modification (112) of the reconciliation tree (208) and of the semantic algorithm (205) based on comparison; and - update (114) of the knowledge base (203).
6. Method (100) according to any one of claims 1 to 5, further comprising a step (116) of generating a confidence score for each reconciliation or substitution of data, and for each reconciliation or substitution of the signaling (118) of low confidence matches.
7. A method according to any one of claims 1 to 6, wherein the generation of the confidence score is based on: - the semantic similarity between the primary data and the selected secondary data; - the number of corresponding proximity criteria in the reconciliation tree; and - the availability and quality of substitute data when exact matches are not found.
8. Method (100) according to any one of claims 1 to 7, wherein said method (100) is implemented within an LCA software platform to automate the process of reconciling LCA data for carrying out LCA studies.
9. Method (100) according to claim 8, wherein the LCA software platform is configured to: - import primary product data from various sources and formats; - access multiple secondary ICV databases; and - export reconciled data.
10. Automated reconciliation device (200) for Life Cycle Inventory (LCI) data of a product or industrial process within the framework of a Life Cycle Assessment (LCA), said device comprising: - a knowledge base (203) stored in a memory (202), said knowledge base comprising mixed data relating to the environmental impacts of industrial materials and processes; - a reconciliation tree (208) designed to automate decisions based on pre-established criteria; and - a processor (204) configured to implement a semantic algorithm (205) interacting with the knowledge base (203) and the reconciliation tree (208) in order to match primary data from the mixed data with appropriate secondary data from the mixed data, such that the primary data are data specific to the product or industrial process under consideration and the secondary data are generic data relating to the environmental impacts of materials and industrial processes which are not specific to the product or industrial process under consideration but provide context and reference values for the LCA, and to select substitute data based on the matches of the semantic algorithm when exact matches between primary and secondary data are not available.
11. Device (200) according to claim 10, wherein the mixed data comprises at least: - technical data sheets for materials and industrial processes including technical specifications, physical properties and manufacturing parameters; - environmental data corresponding to materials and industrial processes; and - ontologies defining the relationships and hierarchies between materials, industrial processes and environmental impacts.
12. Device (200) according to any one of claims 10 or 11, wherein the reconciliation tree (208) comprises: - the definition of a hierarchical structure to organize materials by classifying them into categories and / or subcategories and industrial processes by classifying them into families; - re-establishing semantic relationships between elements, these relationships including links such as "belongs to", "similar to", or "is a type of"; and - the determination of proximity criteria to select surrogate data when exact matches are not available, these criteria being based on technical characteristics, physical properties or functional similarities.
13. Device (200) according to any one of claims 10 to 12, wherein the semantic algorithm (205) is configured to: - analyze the textual descriptions and metadata of primary product data; - compare the analyzed data to the knowledge base entries (203) using natural language processing techniques; and - classify potential matches based on semantic similarity and defined proximity criteria.
14. Device according to any one of claims 10 to 13, further comprising a validation module (206) configured to iteratively validate and refine the automated reconciliation process (100) according to any one of claims 1 to 9 by: - comparing past selections of the process with selections from trusted third parties; - modifying the reconciliation tree (208) and the comparison-based semantic algorithm (205); and - updating the knowledge base (203).
15. Device (200) according to claim 14, wherein the validation module is further configured to generate a confidence score for each reconciliation or substitution of data, and to flag each reconciliation or substitution having a low confidence match.
