A method and apparatus for generating a lifecycle inventory process dataset

By generating a lifecycle inventory process dataset based on reference flow retrieval and clustering, the problem of low data acquisition efficiency and poor reliability in existing technologies is solved, and a highly reliable and traceable dataset is generated, which is suitable for lifecycle assessment.

CN123115229APending Publication Date: 2026-11-17TSINGHUA UNIVERSITY
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611089086.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-07-21
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing technologies lack automated data collection methods in lifecycle assessment, resulting in low data acquisition efficiency and poor reliability. It is difficult to distinguish between real and valid process data and similar data, and the generated process datasets have low reliability and insufficient traceability and verifiability.

Method used

By retrieving summary information and job type from the reference stream, available evidence records are filtered, clustering is performed to determine the main evidence cluster, unit process chains and input/output item lists are generated, and stream record matching and unit compatibility processing are performed in the background stream directory to generate quality description fields, thus achieving data quality closure.

Benefits of technology

It automatically generates highly reliable, traceable, and verifiable process datasets, avoiding the mixing of data from different process routes and system boundaries, and ensuring the reliability and publishability of the data.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123115229A_ABST
    Figure CN123115229A_ABST
Patent Text Reader

Abstract

This application provides a method and apparatus for generating a lifecycle inventory process dataset, relating to the field of lifecycle assessment technology. The method includes: retrieving data based on summary information and job type of a reference flow; clustering available evidence records to determine a master evidence cluster; determining multiple sequentially arranged unit processes based on this cluster; generating a corresponding list of input / output items for each unit process to determine the process dataset; performing flow record matching for each input / output item in a background flow directory; writing background flow reference information for successfully matched input / output items and writing placeholder flow references for unmatched input / output items; and generating a quality description field based on the unit compatibility processing results of the input / output items and the matched flow records, as well as material balance analysis information for the process dataset. This allows for the generation of a highly reliable, traceable, verifiable process dataset that has undergone a data quality closure loop.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This application relates to the field of life cycle assessment technology, and in particular to a method and apparatus for generating a life cycle inventory process dataset. Background Technology

[0002] To achieve Life Cycle Assessment (LCA), it is typically necessary to collect and analyze data on the technological route, that is, a series of unit processes included in the entire production chain from raw materials to the final product. The unit process is the smallest unit for data collection and analysis. Existing methods for building LCA databases usually start with enterprise production and consumption process data, statistical data, industry average data, or manually compiled unit process data. This data is then formatted, classified, coded, aggregated, or averaged to form a database from which LCA modeling can directly obtain relevant data.

[0003] In situations where complete enterprise-wide process data is lacking or insufficient, relevant data can be supplemented by searching literature. However, currently, accurately obtaining relevant data from literature relies on manual reading and analysis, resulting in low automation, high labor costs, and low efficiency. Furthermore, some technologies that automatically extract text or automatically identify literature data based on large language models struggle to meet the data constraints in LCA scenarios, making it difficult to distinguish truly valid process data from highly similar data. This leads to process datasets with low reliability, poor traceability, and poor verifiability. Summary of the Invention

[0004] In view of this, this application provides a method and apparatus for generating a lifecycle inventory process dataset, which aims to automatically generate a highly reliable, traceable, verifiable process dataset that has undergone a data quality closure loop.

[0005] According to a first aspect of the embodiments of this application, a method for generating a lifecycle inventory process dataset is provided, comprising: A search is performed based on the abstract information of the reference stream and the job type of the reference stream, and a record of available evidence is obtained based on the candidate document records retrieved.

[0006] The available evidence records are clustered according to preset clustering features to obtain at least one evidence cluster, and the main evidence cluster is determined from the at least one evidence cluster.

[0007] Based on the master evidence cluster, a unit process chain is determined, which includes multiple unit processes arranged in sequence; for each unit process, a corresponding list of input and output items is generated to determine the process dataset.

[0008] For each input / output item in the background stream directory, perform stream record matching. Write background stream reference information for successfully matched input / output items and write placeholder stream references for unmatched input / output items.

[0009] Based on the unit compatibility processing results of the input / output items and the matched flow records, and the material balance analysis information for the process dataset, a quality description field for the process dataset is generated.

[0010] In some possible implementations, the preset clustering features include at least one of system boundaries, process path types, or intermediate flows.

[0011] In some possible implementations, before generating the quality description field of the process dataset, the method further includes: reconstructing a secondary retrieval request for input / output items in the process dataset that failed to match, so as to re-match the input / output items with stream records in the background stream directory.

[0012] The secondary retrieval request has more relaxed matching constraints compared to the retrieval operation of the previous stream record.

[0013] In some possible implementations, generating a corresponding list of input / output items for each unit process includes performing at least one of the following operations: In response to the explicit existence of the magnitude information of the input / output item in the document recorded by the available evidence, the magnitude information is written into the input / output item, or the magnitude information after conversion is written into the input / output item.

[0014] In response to the absence of quantity information for the input / output item in the literature, the estimated quantity information and its description are written into the input / output item, or the missing quantity information is written into the input / output item.

[0015] In some possible implementations, generating a corresponding list of input / output items for each unit process further includes: determining coverage information based on the list of input / output items; the coverage information is used to indicate the proportion of input / output items obtained by different quantitative information acquisition methods, as well as the coverage rate of key unit processes; and determining the next operation task based on the coverage information.

[0016] In some possible implementations, the unit compatibility processing result of the input / output items and the matched stream records is determined by performing the following operations: Obtain the reference unit of the stream record and the physical dimension category to which the reference unit belongs.

[0017] In response to the fact that the unit of the input / output item and the reference unit belong to the same physical dimension category, or that the unit of the input / output item and the reference unit support physical dimension category conversion through attribute parameters, the unit of the input / output item can be converted to the reference unit.

[0018] Alternatively, in response to the missing unit of the input / output item, the missing physical dimension category, or the unit of the input / output item not supporting physical dimension category conversion with the reference unit, the verification identifier is associated with the input / output item.

[0019] In some possible implementations, the material balance analysis information is determined by: determining the input and output items for material balance analysis based on the direction, flow type, and physical dimension category of the input and output items; and calculating the material balance closure error and / or energy balance closure error based on the determined input and output items. At least one of the following operations is performed during the calculation process: Verify whether the characteristics of the input and output items meet the exclusion constraints.

[0020] Verify whether the reference stream and input / output direction are mismatched.

[0021] Verify the method of obtaining the value information of the input and output items, and the consistency of the units of the input and output items with the matched stream records.

[0022] Verify whether the material balance closure error and / or energy balance closure error exceed the preset balance threshold.

[0023] In some possible implementations, the quality description field includes: a data cutoff and integrity principle field, used to describe the system boundary, the scope of the unit process covered by the process dataset, the situation of missing process data, and the situation of mismatched flow records; and a data processing and extrapolation principle field, used to describe the data acquisition method and data usage prompts.

[0024] The method further includes: returning the set steps in the process dataset generation method based on the content of the quality description field.

[0025] In some possible implementations, the step of performing a search based on the summary information of the reference stream and the job type of the reference stream, and obtaining available evidence records based on the retrieved candidate document records, includes: Based on the reference stream dataset and the job type of the reference stream, obtain reference stream summary information, which includes the identity information of the reference stream.

[0026] Based on the reference stream summary information and the job type, document retrieval conditions are constructed.

[0027] Based on the literature search criteria, candidate literature records are obtained.

[0028] In response to the candidate documents containing physical information used to generate the process dataset, they are determined to be usable evidence.

[0029] In some possible implementations, obtaining candidate document records according to the document retrieval conditions includes at least one of the following operations: performing deduplication and merging processing on the retrieved candidate documents; obtaining supplementary materials based on the full text of the candidate documents, and merging the supplementary materials with the corresponding candidate documents.

[0030] In some possible implementations, obtaining the available evidence record based on the retrieved candidate document records further includes: performing usage labeling on the available evidence.

[0031] The input / output item list includes evidence citation information for each input / output item, which is determined according to the intended use.

[0032] The method further includes: determining the source dataset based on the evidence citation information.

[0033] According to a second aspect of the embodiments of this application, a lifecycle inventory process dataset generation apparatus is also provided, comprising: The evidence processing module is used to perform retrieval based on the summary information of the reference stream and the job type of the reference stream, and to obtain available evidence records based on the retrieved candidate document records; and to cluster the available evidence records according to preset clustering features to obtain at least one evidence cluster, and to determine the main evidence cluster from the at least one evidence cluster.

[0034] The process generation module is used to determine a unit process chain based on the master evidence cluster, the unit process chain including multiple unit processes arranged in sequence; and to generate a corresponding list of input and output items for each unit process to determine the process dataset.

[0035] The stream matching module is used to perform stream record matching for each input / output item in the background stream directory, write background stream reference information for successfully matched input / output items, and write placeholder stream references for input / output items that fail to match.

[0036] The quality closed-loop module is used to generate a quality description field for the process dataset based on the unit compatibility processing results of the input / output items and the matching flow records, as well as the material balance analysis information for the process dataset.

[0037] In some possible implementations, the apparatus further includes a reference stream parsing module, configured to obtain reference stream summary information based on the reference stream information and the job type of the reference stream, wherein the reference stream summary information includes the identity information of the reference stream.

[0038] The evidence processing module is used for: Based on the reference stream summary information and the job type, document retrieval conditions are constructed.

[0039] Based on the literature search criteria, candidate literature records are obtained.

[0040] In response to the candidate documents containing physical information used to generate the process dataset, they are determined to be usable evidence.

[0041] In some possible implementations, the device further includes: The unit processing module is used to perform unit compatibility processing on the input / output items and the matching stream records, and to determine the unit compatibility processing result.

[0042] The dataset construction module is used to determine the process dataset based on the unit process chain and the list of input and output items for each unit process.

[0043] The quality closed-loop module is also used to generate material balance analysis information for the process dataset, and generate the quality description field based on the unit compatibility processing result and the material balance analysis information.

[0044] According to a third aspect of the embodiments of this application, a computer program product is also provided, including a computer program that, when executed by a processor, implements the method described in the above embodiments.

[0045] According to a fourth aspect of the embodiments of this application, an electronic device is also provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method described in the above embodiments.

[0046] According to a fifth aspect of the embodiments of this application, a computer-readable storage medium is also provided, on which a computer program is stored, which, when executed by a processor, implements the method described in the above embodiments.

[0047] The technical solutions provided by the embodiments of this application may include the following beneficial effects: This application starts with a reference stream, retrieves information based on the stream's summary and job type, and filters the candidate documents for usability to obtain usable evidence records. The usable candidate documents (i.e., usable evidence) are clustered to determine the main evidence cluster. The process of generating the unit process chain and input / output item list is then constrained by the main evidence cluster. This avoids directly mixing documents from different process routes, system boundaries, or raw material systems, reducing inconsistencies in the process dataset caused by mixed evidence. During the generation of the input / output item list, establishing evidence citation relationships between input / output item content and usable evidence ensures data traceability. Based on stream matching operations and unit compatibility processing using the background stream directory, input / output items are converted into a citationable stream dataset with stream identifiers, versions, resource addresses, and short descriptions. This ensures the traceability and publishability of the process dataset, avoiding the risk of unresolved citations during process dataset publication. Material balance analysis is performed on the processed process dataset. Combining the results of stream matching operations and unit compatibility processing, a quality description field is generated, achieving a data quality closed loop and ensuring data reliability, traceability, and verifiability. Therefore, in the absence of complete enterprise-measured process data or insufficient enterprise process data, it can automatically generate highly reliable, traceable, verifiable process datasets that complete the data quality closed loop.

[0048] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description

[0049] Figure 1 This is a flowchart illustrating a method for generating a lifecycle inventory process dataset according to an embodiment of this application; Figure 2 This is a schematic diagram of the generation architecture of a quality description field provided in an embodiment of this application; Figure 3 This is a schematic diagram of a closed-loop processing flow for process dataset quality provided in an embodiment of this application; Figure 4 This is a schematic diagram of the structure of a lifecycle inventory process dataset generation device provided in an embodiment of this application; Figure 5 This is a schematic diagram of another process dataset generation device provided in an embodiment of this application; Figure 6 This is a schematic diagram of the structure of another process dataset generation device provided in the embodiments of this application; Figure 7 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0050] Currently, in situations where complete enterprise-measured process data is lacking or insufficient, it is typically necessary to consult relevant literature to obtain the missing data in order to ensure the smooth conduct of Life Cycle Assessment (LCA). The aforementioned literature-based data generation process usually relies on those skilled in the art to conduct literature searches and analyses, reading each document one by one, identifying the process data for each unit process, and constructing a process dataset that meets standard data specifications. For example, those skilled in the art can determine the target unit process based on reference flow information and action type. Furthermore, they can conduct literature searches based on the above information to identify the technical approach, system boundary, and unit process chain corresponding to each document's content, as well as the input and output list of each unit process in the unit process chain. Then, based on the degree of matching between the document content and the target unit process, the required process dataset can be determined.

[0051] A typical process dataset includes the process name, geographical scope, technical representativeness, quantitative reference, input and output items, stream dataset references, source references, data cutoff principles, integrity descriptions, and data processing and extrapolation principles. For data systems using the Tiangong Life Cycle Assessment Data System (TIDAS), the International Reference Life Cycle Database (ILCD), or other data standards, input and output items not only need to have name and quantity information, but also need to be able to reference specific stream datasets in the background stream catalog, and meet structured requirements such as reference unit, unit group (i.e., the physical dimension category to which the unit belongs), stream type, direction, and source reference.

[0052] Because the literature in the LCA field contains abundant and highly similar data, using conventional text extraction, knowledge discovery, large language models, or other generative models to obtain this process dataset makes it difficult to accurately distinguish between different types and meanings of data, resulting in poor quality and low reliability of the obtained process dataset. For example, process datasets obtained based on existing data automatic generation methods are prone to the following problems: data from literature at different system boundaries are mixed; the generated unit process chain is inconsistent with the target reference flow; input and output item names cannot be resolved in the background flow directory; input and output directions, flow types, or environmental media are unclear; input and output item units are incompatible with the selected background flow reference units; issues such as missing quantities, estimators, placeholder flows, and unit assumptions are untraceable; there is a lack of traceable citations between the source literature and the process dataset; and the data is untraceable and unverifiable.

[0053] In view of this, this application provides a method for generating a lifecycle inventory process dataset, which aims to automatically generate a highly reliable, traceable, verifiable process dataset that has undergone a data quality closure loop when faced with the problem of lacking complete enterprise measured process data or insufficient enterprise process data.

[0054] like Figure 1 As shown, Figure 1 This is a flowchart illustrating a method for generating a lifecycle inventory process dataset according to an embodiment of this application. The method includes the following steps S100 to S500: Step S100: Perform a search based on the abstract information of the reference stream and the job type of the reference stream, and obtain available evidence records based on the candidate document records obtained from the search.

[0055] The reference stream summary information includes the identification information of the reference stream. For example, it may include at least one of the following: reference stream name, such as product name, waste stream name, or emission stream name; classification information, such as product classification, waste stream classification, or environmental compartment; descriptive information, such as stream description, purpose, chemical formula, Chemical Abstracts Service Registry Number (CAS number), composition, or treatment description. Alternatively, it may also include stream identifier and version, as well as reference attributes, units, or unit groups.

[0056] The job type represents the technical action corresponding to the process dataset to be generated, such as producing a reference stream, processing a waste stream, recycling a substance, or disposing of a material. Job types can be used to determine the system boundary and reference stream transitivity of the target unit process chain. For example, if the job type is production, the target unit process chain can progress from upstream raw material pretreatment, reaction, separation, drying, or packaging processes to the reference stream output process. If the job type is processing or recycling, the target unit process chain can start from the stream to be processed entering the system, progressing through sorting, pretreatment, conversion, recycling, or disposal processes to the reference stream or final output stream.

[0057] Based on the abstract information and job type of the reference stream, the document retrieval criteria can be clearly defined, thus enabling the retrieval of candidate documents. The document retrieval service can be based on local document databases, remote knowledge bases, patent / paper databases, or internal enterprise knowledge bases; no specific limitations are specified here.

[0058] The aforementioned reference stream summary information can be determined based on explicit user instructions, or it can be obtained by parsing given reference stream information. For example, if the given reference stream information includes a reference stream dataset, this dataset can be a structured dataset conforming to ILCD or TIDAS requirements, or it can be a stream dataset with a structured representation based on JSON data format or other formats.

[0059] For example, given a reference stream dataset, step S100 above includes the following steps: First, based on the reference stream dataset and the job type of the reference stream, reference stream summary information is obtained. For example, the reference stream dataset is parsed to extract the identity information of the reference stream.

[0060] Next, based on the reference flow summary information and the job type, document retrieval conditions are constructed. For example, the reference flow summary information includes the name of a reference flow. When the reference flow name is a certain material and the job type is production, the document retrieval conditions may include keywords such as the material name, production, process, life cycle list, unit process, input, and output.

[0061] Next, candidate literature records are obtained based on the literature search criteria.

[0062] In some possible implementations, the identified candidate documents are further deduplicated and merged. For example, based on the Digital Object Identifier (DOI), title, document identifier, or document content similarity, the same document retrieved from different databases, with different keywords, or with different title formats is merged and recorded under the same document entity. Information corresponding to this document entity, such as DOI, title, and full text, is uniformly saved to avoid subsequent duplicate processing of the document.

[0063] Alternatively, supplementary materials can be obtained from the full text of candidate documents, and then merged with the corresponding candidate documents. For example, based on the identification of document content, clues to supplementary materials can be determined, and thus the supplementary materials can be obtained. Supplementary materials can be text, images, spreadsheets, electronic documents, or structured text converted by document parsing tools. Merging the supplementary materials under the document entity maintains the data correlation and ensures data traceability during subsequent processing.

[0064] For the analysis of full-text documents and supplementary materials, PDF analysis, table analysis, optical character recognition (OCR), document structure analysis, or manual input can be used, and there are no restrictions here.

[0065] Finally, in response to the candidate document containing physical information used to generate the process dataset, the candidate document is identified as usable evidence.

[0066] In other words, after identifying candidate literature records, a usability screening process is conducted. This screening does not solely rely on whether the literature discusses environmental impact indicators, but rather assesses whether the literature contains physical information that can be used to generate a Life Cycle Inventory (LCI) process dataset. For example, if a literature only provides Life Cycle Impact Assessment (LCIA) indicators such as global warming potential and acidification potential, but does not disclose physical life cycle inventory data such as raw materials, energy, products, waste streams, or emissions streams, then this literature can serve as background information but cannot be used as evidence for generating input / output item lists; that is, it cannot be used to generate process datasets. If a literature provides an inventory table containing raw material consumption or product output for production routes, unit processes, material inputs, energy consumption, emissions, or supplementary materials, then this literature or supplementary material can be marked as usable evidence.

[0067] After the above processing, usable candidate literature, that is, usable evidence, can be screened out.

[0068] Step S200: Cluster the available evidence records according to preset clustering features to obtain at least one evidence cluster, and determine the main evidence cluster from the at least one evidence cluster.

[0069] Evidence clustering methods can employ rule-based clustering or vector clustering, and can also be supplemented with manual verification to validate the results, which is not limited here.

[0070] Preset clustering features can be determined based on the features of the target process dataset. For example, they may include one or more features such as system boundary, main process chain (i.e., process path type), intermediate flow, reference flow relevance, region, technical route, raw material type, product type, or input / output item coverage.

[0071] Preferably, the preset clustering features include at least one of system boundary, process path type, or intermediate stream. System boundary represents the lifecycle stage or process range covered by the literature; process path type represents the sequence of key unit processes from raw materials, waste streams entering the treatment system, or other upstream streams to the reference stream; intermediate stream represents semi-finished products, precursors, waste streams, or byproducts occurring during the process. These three features are highly correlated with the quality of the process dataset to be generated. Selecting evidence clusters with high matching degrees after clustering based on these features as the master evidence clusters for subsequent data generation helps ensure high data reliability.

[0072] For example, in some possible implementations, the preset clustering features include system boundaries, process path types, and intermediate flows. That is, available evidence records are clustered based on a combination of these three dimensions. This avoids mixing documents from different ranges such as cradle-to-gate, gate-to-gate, and cradle-to-grave, while ensuring that the resulting cluster describes the same type of process path and that the flows of semi-finished products, precursors, byproducts, or waste have high consistency. Therefore, through clustering, the risk of data from different technical routes or with different boundaries being pieced together into the same process dataset can be significantly reduced, resulting in a process dataset with high data reliability.

[0073] For multiple evidence clusters obtained from clustering, a weighted scoring method can be used to determine the master evidence cluster. For example, reference flow relevance is used to indicate the degree of relevance between available evidence and the target unit process, and system boundary integrity is used to represent the extent to which available evidence covers the system boundary range of the target unit process chain. Both of these have a high impact on generating a reliable process dataset. Based on this, the weighted score of each evidence cluster can be determined according to the reference flow relevance, system boundary integrity, and the corresponding weight coefficients.

[0074] Understandably, only two influencing factors are listed here under the weighted scoring method. In actual processing, other features can be combined for weighted calculation. Among them, the weight coefficients corresponding to the reference flow relevance and system boundary integrity can be higher than the weight coefficients of other influencing factors.

[0075] After determining the weighted score for each evidence cluster, the evidence clusters whose weighted scores meet the set values ​​can be identified as the master evidence clusters. For example, the evidence cluster with the highest weighted score, or the evidence cluster with a weighted score greater than a preset value, can be identified as the master evidence cluster. Alternatively, after determining the weighted score for each evidence cluster, candidate evidence clusters can be further filtered based on the input / output item evidence coverage. From the evidence clusters that meet the coverage requirements, the one with the highest weighted score can be selected as the master evidence cluster. Here, input / output item evidence coverage refers to the degree to which the input / output items in the available evidence cover the input / output items in the target process dataset. The higher the coverage, the more conducive it is to ensuring the reliability of the subsequently generated list of input / output items.

[0076] In some possible implementations, the master evidence cluster can also be determined from the at least one evidence cluster according to preset screening rules. For example, evidence clusters with a high degree of matching with the reference flow name and job type can be preferentially selected as the master evidence cluster. Alternatively, evidence clusters with complete system boundaries, sufficient process chain coverage, and consistent intermediate flows can be preferentially selected as the master evidence cluster. Or, evidence clusters containing input / output item evidence or supplementary materials can also be preferentially selected as the master evidence cluster.

[0077] This application uses clustering screening to select a cluster of available candidate documents with the same clustering characteristics for subsequent data generation processing. This avoids directly mixing documents from different process routes, different system boundaries, or different raw material systems, which helps reduce inconsistencies in process datasets caused by the mixing of evidence.

[0078] Step S300: Based on the master evidence cluster, determine the unit process chain, which includes multiple unit processes arranged in sequence; for each unit process, generate a corresponding list of input and output items to determine the process dataset.

[0079] For example, at least one technical route can be determined based on the main evidence cluster. The technical route may include route name, raw material type, key process, product or processing object, system boundary, applicable region, evidence citation and supporting document identification.

[0080] When multiple technical approaches exist, users can choose a target technical approach, or an automatic target technical approach can be determined based on information such as reference relevance, evidence coverage, literature availability, or system boundary consistency.

[0081] For example, for multiple technical routes, each route can be comprehensively scored based on preset exclusion rules and weighting coefficients. The route with the highest comprehensive score exceeding a set threshold is identified as the target route. The exclusion rules can be used to set the comprehensive score of a route that triggers an exclusion condition to zero, or to exclude the route altogether. For example, routes that do not match the target environment, or whose geographical location or raw material system does not meet the requirements, can be directly excluded. During the weighted calculation process, multiple scoring factors can be constructed. For example, scoring factors may include: the degree of matching between the reference flow name, classification, and job type; the quantity and quality of available evidence supporting the technical route; system boundary integrity; unit process chain integrity; intermediate flow consistency, etc. The weight of each scoring factor can be preset and is not limited here.

[0082] If multiple technical approaches are similar or lack sufficient evidence, a status pending manual confirmation can be output.

[0083] Based on the determined target technical route, a corresponding unit process chain can be generated, and a unit process chain can include multiple unit processes. A list of input and output items for each unit process is generated, and by integrating multiple input and output item lists along the order of the unit process chain, the process dataset can be constructed.

[0084] In some possible implementations, the process dataset includes at least several of the following: process information, geographic information, technical information, quantitative references, input / output items, source references, modeling and validation information, and management information.

[0085] For each unit process, one or more of the following information may be included: process name components (e.g., including basic name, route or technology limitation, processing object or product limitation), process sequence number, geographic scope code or geographic scope description, system boundary description, summary description of input and output items, whether quantitative references are included, reference flow association information, and supporting evidence citations.

[0086] In this application, input / output items are structured objects within the process dataset. Each input / output item in the list can include a name (i.e., a single stream name retrieved from the background stream directory), input / output direction, stream type, magnitude information, stream dataset reference, and evidence reference. The evidence reference can include a document identifier, full-text location, supplementary material location, and table or paragraph number, indicating the relationship between the input / output item and the source document.

[0087] In some possible implementations, each input / output item may also include the following fields: input / output item status, including extracted, missing, estimated, pending review, etc.; whether it is a reference flow; flow type; material role, such as raw material, auxiliary material, energy, product, by-product, waste flow, emission, infrastructure, or transportation; balance exclusion flag, used to exclude input / output items that do not participate in the calculation of material balance closure error and / or energy balance closure error in material balance analysis; data source; status label, used for subsequent identification of input / output role, unit processing status, or pending review status.

[0088] In this embodiment, the names of input and output items are the names of individual substances, rather than composite names combining multiple substances. For example, emissions streams such as carbon dioxide, methane, and nitrogen oxides should be written separately, instead of the composite name "greenhouse gas emissions." When a document only provides composite indicators and cannot reconstruct the physical flow, the corresponding content is not directly written as the physical flow in the input / output item list.

[0089] In some possible implementations, the above-mentioned technical route, unit process chain, and input / output item list can be generated by a rule engine, template system, machine learning model, language model, or human-computer collaboration method, and there is no limitation here.

[0090] Step S400: Perform stream record matching for each input / output item in the background stream directory, write background stream reference information for successfully matched input / output items, and write placeholder stream references for input / output items that fail to match.

[0091] The background stream directory here can be a stream directory from ILCD, TIDAS, EcoSpold, JSON-LD, or other lifecycle assessment databases, and the specific directory can be determined according to the constraints of the corresponding standard.

[0092] During this flow matching operation, background flow retrieval conditions are first constructed based on the input and output items. These background flow retrieval conditions include at least one feature of the input and output items, such as input / output item name, synonyms, input / output direction, flow type, environmental medium, unit, classification clues, material role, and reference flow relationship.

[0093] Subsequently, in response to a retrieval request to the background stream directory, candidate stream records are obtained. Candidate stream records may include information such as stream name, stream identifier, version, resource address, short description, reference attributes, reference unit, unit group, stream type, category, and geographical extent.

[0094] Next, candidate stream records that do not meet the preset matching requirements can be eliminated. For example, stream type, input / output direction, unit group compatibility, and environmental medium can be used as hard constraints in certain scenarios. If these information fail to match, the corresponding candidate stream record will be eliminated directly.

[0095] Finally, a weighted scoring method can be used to identify the candidate flow records with the highest overall score or those exceeding a threshold as the flow records that match the input / output item. For example, in the weighted scoring process, the name, synonyms, category, description, and material role of the candidate flow record can be used as scoring factors. Each scoring factor has a corresponding weight coefficient. The matching degree score of each scoring factor is multiplied by its corresponding weight coefficient to obtain the score for that scoring factor. The scores of multiple scoring factors are then added together to obtain the overall score for the candidate flow record.

[0096] For a successfully matched input / output item, write background stream reference information, such as the matched stream identifier, version, resource address, and short description, into the stream reference field of that input / output item.

[0097] For input / output items that fail to match or are uncertain about whether they match, a pending review status flag or a placeholder stream reference can be written to the stream reference field of the input / output item.

[0098] In this application, by performing stream matching in the background stream directory and writing the stream matching results to the input and output items, the input and output items can be converted into a referable stream dataset with stream identifier, version, resource address and short description, thereby avoiding the risk of unresolved references when publishing the process dataset.

[0099] It is understood that the process of integrating multiple lists of input and output items to construct the process dataset can be performed after or before stream matching. No constraints are imposed here, as long as the input and output items in the final output process dataset have undergone the stream matching result writing operation.

[0100] Step S500: Based on the unit compatibility processing results of the input / output items and the matched flow records, and the material balance analysis information for the process dataset, generate the quality description field of the process dataset.

[0101] For example, the material balance analysis information here includes analysis information obtained by verifying the material balance closure error and / or energy balance closure error based on the process dataset, which is used to verify the reliability of the data.

[0102] In other words, in this embodiment of the application, after integrating and forming the process dataset but before outputting the process dataset, a data quality closed-loop processing is implemented, writing the data quality description content into the quality description field of the process dataset. Therefore, problems arising during the generation of the process dataset, such as those generated or identified during input / output item lists, flow matching, unit compatibility processing, and material balance analysis, are not only kept in the internal logs but are also written back to the quality description field of the process dataset. This ensures that data quality issues such as missing content, flow matching failures, system boundary inconsistencies, and unit inconsistencies can be traced in the output process dataset.

[0103] In conjunction with the foregoing, this embodiment of the application starts with a reference stream, performs a retrieval based on the reference stream's summary information and job type, and filters the obtained candidate documents for availability to obtain usable evidence records. The usable candidate documents (i.e., usable evidence) are clustered to determine the main evidence cluster. Subsequently, the process chain of the constrained unit and the list of input / output items are generated based on the main evidence cluster. This avoids directly mixing documents from different process routes, different system boundaries, or different raw material systems, which helps reduce inconsistencies in the process dataset caused by mixed evidence. During the generation of the input / output item list, the traceability of the data is ensured by constructing evidence citation relationships between the input / output item content and the usable evidence. Based on the flow matching operation and unit compatibility processing of the background flow directory, the input / output items are converted into a citationable flow dataset with flow identifiers, versions, resource addresses, and short descriptions. This ensures the traceability and publishability of the process dataset, thereby avoiding the risk of unresolved citations when the process dataset is published. Material balance analysis is performed on the processed process dataset. Combined with flow matching and unit compatibility processing results, a quality description field is generated, achieving a data quality closed loop and ensuring data reliability, traceability, and verifiability. Therefore, even in the absence of complete enterprise-wide measured process data or with insufficient enterprise process data, a highly reliable, traceable, verifiable process dataset with a complete data quality closed loop can be automatically generated.

[0104] In some possible implementations, obtaining the available evidence record based on the retrieved candidate document record further includes: performing usage labeling on the available candidate documents.

[0105] For example, before performing evidence clustering, the available evidence (i.e., available candidate documents) is labeled with its intended use. This labeling is also known as document tagging, such as technical route evidence tags, process breakdown evidence tags, input / output item evidence tags, and / or background evidence tags. A piece of available evidence can have multiple use tags. For instance, the main text of a document can be used to determine a technical route, while its supplementary materials can be used to determine inputs and outputs. After merging the document and its supplementary materials into a unified document entity, that document entity can simultaneously have both a technical route evidence tag and an input / output item evidence tag.

[0106] The input / output item list generated based on documents with usage labels can include evidence citation information for each input / output item, which is determined according to the usage label. For example, the evidence citation field of an input / output item records information such as the title, author, publication year, DOI, or resource address of the document that provided the data for that input / output item.

[0107] Therefore, during the generation of the process dataset, the source dataset can be determined based on the aforementioned evidence citation information. By constructing the source dataset and writing source citations (i.e., evidence citation information) into the process dataset, the technical route, unit process, and input / output items can be associated with corresponding literature evidence. When outputting the final process dataset, the corresponding source dataset can be output simultaneously, thereby ensuring the traceability and verifiability of the process dataset.

[0108] In some possible implementations, before generating the quality description field of the process dataset, placeholder stream parsing is performed on the input and output items of the process dataset, and placeholder stream parsing information is recorded. This includes, for example, recording the parsing status, candidate stream list, reasons for selecting background stream references, confidence levels, or reasons for stream matching failures.

[0109] For input / output items that fail to match the stream, in some possible implementations, a secondary search request will be reconstructed to re-match the input / output items against the stream records in the background stream directory.

[0110] The secondary retrieval request has more relaxed matching constraints compared to the retrieval operation of the previous stream record.

[0111] For example, after the initial flow matching is completed and the process dataset is constructed, the flow reference field of each input / output item in the process dataset is traversed. If background flow reference information exists, such as flow identifier, version, resource address, and short description, it can be confirmed that the input / output item can match the background flow directory. If a placeholder flow reference is identified, a secondary resolution operation can be performed. That is, based on the name, description, CAS number, chemical formula, classification clues, flow type, input / output direction, unit, environmental medium, and other information of the input / output item, as well as the previous flow matching retrieval clues, a secondary retrieval request is constructed to reduce the placeholder flows left after the initial background flow matching failure and improve the resolvability of flow references in the final process dataset. For example, the first match failure may be due to an overly broad name, insufficient synonyms, insufficient classification clues, undefined environmental medium, or a more accurate name in the supplementary materials. Based on the first retrieval clues, a secondary retrieval request is reconstructed by using broader synonyms, finer classifications, adding CAS numbers, supplementing material names, or environmental media, and the flow matching operation is re-executed based on this secondary retrieval request.

[0112] If the secondary parsing is successful, meaning the corresponding stream record is successfully matched based on the secondary search request, the placeholder stream reference will be replaced with the matched background stream reference information. If the secondary parsing fails, meaning the corresponding stream record is still not matched based on the secondary search request, the original placeholder stream reference can be retained, and the placeholder status of the stream reference for that input / output item will be recorded in the subsequently generated quality description field.

[0113] Furthermore, based on the secondary parsing process, the placeholder stream parsing information can be updated, and the final stream reference state, parsing state, candidate stream list, reasons for selecting background stream reference information, confidence level, or reasons for failure can be identified.

[0114] In some possible implementations, generating a corresponding list of input / output items for each unit process includes performing at least one of the following operations: In response to the explicit existence of the magnitude information of the input / output item in the document recorded by the available evidence, the magnitude information is written into the input / output item, or the magnitude information after conversion is written into the input / output item.

[0115] In response to the absence of quantity information for the input / output item in the literature, the estimated quantity information and its description are written into the input / output item, or the missing quantity information is written into the input / output item.

[0116] For example, when input / output items lack quantitative information, explicit quantitative information is extracted first from the full text of available literature or supplementary materials. This quantitative information includes at least one of a numerical value, unit, and measurement standard, and the position of this explicit quantitative information within the available evidence is recorded. For instance, if a table in the supplementary materials provides the electricity consumption per kilogram of reference flow, the numerical value, unit, measurement standard, and table position are written into the corresponding input / output item.

[0117] If the measurement benchmark of the explicit numerical values ​​differs from the reference benchmark of the process dataset, but there is an explicit conversion relationship between the two, the input and output items can be converted to the reference benchmark of the process dataset, while retaining the original values, original units, and conversion instructions. Unit conversion can use the built-in unit library, external unit services, or unit group definitions in the database; no restrictions are imposed here.

[0118] If there are no explicit numerical values ​​in the available evidence, the input and output items can be left as missing. Alternatively, if there is an interpretable source, an estimated value can be written, along with the source type, evidence description, and assumptions of that estimated value. A "to be reviewed" mark can also be written simultaneously for manual review reminders.

[0119] In some possible implementations, generating a corresponding list of input / output items for each unit process further includes: determining coverage information based on the list of input / output items; the coverage information is used to indicate the proportion of input / output items obtained by different quantitative information acquisition methods, as well as the coverage rate of key unit processes; and determining the next operation task based on the coverage information.

[0120] Coverage information can indicate the degree of data integrity. Based on this, it can be used to indicate whether manual review is required and whether to output the process dataset generated in this round.

[0121] For example, in conjunction with the input / output item generation process described above, the coverage information here can include the proportion of input / output items determined based on the explicit quantitative information in available evidence (i.e., explicit input / output item coverage), the proportion of input / output items written into the estimated values ​​based on interpretable sources (i.e., estimated input / output item coverage), a list of missing input / output items, and a list of input / output items requiring manual review. Along the unit process chain, the coverage of key processes within the unit process chain can also be clearly defined. Coverage information can be used to determine whether to continue automatically generating process datasets, whether to stop waiting for manual supplementation, or whether to output only draft-level process datasets, to ensure the reliable implementation of the data quality closure loop.

[0122] In some possible implementations, after performing a stream record matching operation on the input / output items, a compatibility processing step is performed between the input / output items and the matched stream record execution unit, including: Obtain the reference unit and unit group (i.e., the physical dimension category to which the reference unit belongs) of the stream record.

[0123] In response to the fact that the unit of the input / output item and the reference unit belong to the same physical dimension category, or that the unit of the input / output item and the reference unit support physical dimension category conversion through attribute parameters, the unit of the input / output item can be converted to the reference unit.

[0124] Alternatively, in response to the missing unit of the input / output item, the missing physical dimension category, or the unit of the input / output item not supporting physical dimension category conversion with the reference unit, the verification identifier is associated with the input / output item.

[0125] For example, the unit group can represent mass, volume, energy, quantity, area, transport work, time, or other dimensions. After obtaining the reference unit for the matching flow record, it can be determined whether the units of the input and output items belong to the same unit group as the reference unit.

[0126] If they belong to the same unit group, the input and output items can be converted to the reference unit according to the unit conversion relationship. For example, grams and kilograms belong to the same mass unit group and can be converted to each other. When the unit of the input / output item is grams and the reference unit is kilograms, the unit of the input / output item will be converted to kilograms based on the conversion relationship.

[0127] In some cases, even if units do not belong to the same unit group, alignment can be achieved through density conversion. For example, for product or waste streams, when the input / output units and the reference units of the matching stream record are in mass and volume units respectively, density conversion can still align the units of the input / output items to the reference units. The density conversion can be based on literature, supplementary materials, property databases, or estimates with source information. When performing unit alignment based on density conversion, the density value, density unit, density source, applicable temperature or conditions, and conversion instructions must also be written into the corresponding fields of the input / output items.

[0128] If the input / output item is missing units, missing unit groups, or the unit groups are incompatible (cannot be aligned or converted), write a pending review status to that input / output item, retaining the original units, target reference units, and the reason for the incompatibility. Alternatively, if density conversion is possible, but the source of density values ​​is insufficient, the pending review status can still be retained.

[0129] The results of the above-mentioned unit compatibility processing operation can be reflected in the processed process data to perform material balance analysis and complete the data quality closure loop. On the other hand, it can also be directly used to generate the quality description field.

[0130] In some possible implementations, the material balance analysis information is determined by: determining the input and output items for material balance analysis based on the direction, flow type, and physical dimension category of the input and output items; and calculating the material balance closure error and / or energy balance closure error based on the determined input and output items.

[0131] During the calculation process, at least one of the following operations is performed: verifying whether the characteristics of the input and output items meet the exclusion constraint requirements, verifying whether the reference flow and the input and output directions are mismatched, verifying the acquisition method of the quantity information of the input and output items and the consistency of the units of the input and output items with the matched flow records, and verifying whether the material balance closure error and / or energy balance closure error exceed the preset balance threshold.

[0132] For example, this material balance analysis process first filters input / output items for participating in the balance closure error calculation based on input / output direction, flow type, material role, balance exclusion flags, and unit information. For instance, input / output items with the direction of input / output, flow type of product flow / waste flow / material flow / emission flow, and unit belonging to the mass unit group that are not marked as excluded are preferentially selected. Infrastructure, services, transportation, economic flows, or other input / output items marked as not participating in the balance can be excluded in this process. For example, raw material inputs, product outputs, solid waste outputs, and emissions in kilograms can participate in the material balance analysis, while electricity, heat, transportation services, infrastructure, economic flows, or input / output items whose units are not mass units and cannot be converted to mass can be marked as not participating in the balance. Alternatively, if an energy balance analysis is performed, input / output items in energy units such as joules or kilowatt-hours can be selected separately.

[0133] The process can calculate at least one of quality input, quality output, energy input, or energy output, and identify whether the input / output items used for calculation have the following problems: reference stream mismatch, such as using a reference product stream that should be an output item as an input item; inconsistent units or missing units; over-density conversion; data that is an estimate or missing value; the presence of placeholder stream references; or being marked as excluded.

[0134] Based on the above processing steps, material balance analysis information is determined to identify potential issues in the process dataset, such as missing input / output items, incorrect directions, mismatched reference flows, incompatible units, incorrect density conversion assumptions, or abnormal input / output discrepancies. Before outputting the process dataset, a closed-loop data quality check is performed to provide a basis for generating subsequent quality description fields.

[0135] In some possible implementations, the quality description field includes: a data cutoff and integrity principle field, used to describe the system boundary, the scope of the unit process covered by the process dataset, the situation of missing process data, and the situation of mismatched flow records; and a data processing and extrapolation principle field, used to describe the data acquisition method and data usage prompts.

[0136] For example, in conjunction with what has been described above, Figure 2 This is a schematic diagram of the generation architecture of a quality description field provided in an embodiment of this application.

[0137] By combining the above-described input / output item list generation process with the material balance analysis process, missing input / output item information can be identified. Based on the above-described placeholder flow parsing and secondary parsing process, placeholder flow parsing information such as the final flow reference status, parsing status, candidate flow list, selection rationale for background flow reference information, confidence level, or failure reason can be identified for input / output items. Based on the above-described unit compatibility processing process, unit consistency and / or density conversion information can be identified. Based on the above-described material balance analysis process, material balance analysis information can be obtained.

[0138] Based on the above information, a quality description field is generated. Information related to system boundaries, data coverage, missing items, and placeholder items is written into the data cutoff and integrity principle field. Information related to unit conversion, density conversion, estimated values, extrapolation assumptions, and material balance analysis processing methods is written into the data processing and extrapolation principle field. This allows subsequent users to know how the data is processed and which content needs to be used with caution.

[0139] For example, in conjunction with the above, the data cutoff and integrity principle field can be written as follows: This dataset is mainly sourced from literature A and supplementary material table S1, covering raw material pretreatment, reaction and separation processes; catalyst recovery data is missing, and unmatched organic waste liquid streams are retained as placeholder streams; plant infrastructure is not included.

[0140] The data processing and extrapolation principle fields can be written as follows: power consumption is converted from kWh to MJ; solvent volumetric flow rate is converted from density given in reference B to kg; missing cooling water consumption is estimated from the interval of references along the same route and marked as to be verified; material balance closure error is an item to be verified.

[0141] In some possible implementations, the method further includes: returning a setting step in the process dataset generation method based on the content of the quality description field.

[0142] In other words, based on the identified issues, corresponding operational steps can be returned, forming a closed-loop process feedback loop to verify or correct the process dataset. For example, if inconsistencies in units are identified, unit compatibility processing steps can be returned. If placeholder stream citations are identified, a secondary parsing process can be triggered. If key input / output items are missing, literature retrieval or input / output item list generation steps can be returned. If the technical approach corresponding to the process dataset is inconsistent with the technical approach of the cited literature, steps for clustering candidate literature or determining the unit process chain based on the main evidence cluster can be returned, and so on.

[0143] In some possible implementations, material balance analysis information can also be output along with the final process dataset as an additional analysis record or validation report, without being limited to being written directly as a separate field in the ILCD or TIDAS standard process dataset.

[0144] In summary, Figure 3 This is a schematic diagram of a closed-loop processing flow for process dataset quality provided in an embodiment of this application. The flow includes the following steps S301 to S309: Step S301: Obtain the reference stream dataset and job type.

[0145] Step S302: Parse the reference stream dataset and job type to determine the reference stream summary information.

[0146] Step S303: Based on the reference stream summary information and job type, construct literature search conditions, obtain candidate literature records, perform availability screening on candidate literature and corresponding supplementary materials, and determine usable evidence records.

[0147] Step S304: Cluster the available evidence and determine the main evidence cluster. This can reduce the problem of low data reliability caused by the mixed use of literature from different system boundaries or different technical approaches.

[0148] Step S305: Determine the target technical route, unit process chain, and input / output item list based on the main evidence cluster. Each input / output item is not only a natural language description, but also structured data including information such as direction, flow type, unit, material role, balance exclusion mark, evidence reference, and machine-resolvable tags.

[0149] Step S306: Perform background stream matching and unit compatibility processing (e.g., unit alignment and / or density conversion) for each input / output item. That is, convert the input / output items into a referential stream dataset with a stream identifier, version, resource address, and short description to reduce the risk of unresolved references when publishing the process dataset. For input / output items whose reference units are incompatible with the matched stream records (cannot be aligned or converted) or are missing, an explicit pending review status is written to avoid implicit unit errors.

[0150] Step S307: Based on the unit process chain and the input / output list for overcurrent matching and unit compatibility processing, construct a process dataset and a source dataset based on the evidence citation information of the data to ensure data traceability.

[0151] Step S308: Perform placeholder flow analysis and material balance analysis on the process dataset to identify potential data quality issues during the generation of the process dataset and generate a quality description field for the process dataset, forming a data quality closed loop. Combined with the source dataset, this ensures that the output process dataset is traceable and verifiable.

[0152] Step S309: Output the process dataset for closed-loop data quality processing.

[0153] In other words, based on the above processing flow, a highly reliable, traceable, verifiable process dataset that has passed through a data quality closed loop can be output.

[0154] Figure 4 This is a schematic diagram of the structure of a lifecycle inventory process dataset generation device provided in an embodiment of this application. Figure 4 As shown, the device includes: The evidence processing module 41 is used to perform retrieval based on the summary information of the reference stream and the job type of the reference stream, and obtain available evidence records based on the retrieved candidate document records; and to cluster the available evidence records according to preset clustering features to obtain at least one evidence cluster, and determine the main evidence cluster from the at least one evidence cluster.

[0155] The process generation module 42 is used to determine a unit process chain based on the main evidence cluster, the unit process chain including multiple unit processes arranged in sequence; and to generate a corresponding list of input and output items for each unit process to determine the process dataset.

[0156] The stream matching module 43 is used to perform stream record matching for each input / output item in the background stream directory, write background stream reference information for successfully matched input / output items, and write placeholder stream references for input / output items that fail to match.

[0157] The quality closed-loop module 44 is used to generate a quality description field for the process dataset based on the unit compatibility processing results of the input / output items and the matching flow records, as well as the material balance analysis information for the process dataset.

[0158] In this embodiment, based on the aforementioned apparatus, a retrieval is performed starting from a reference stream, according to the summary information and job type of the reference stream. The obtained candidate documents are then screened for availability to obtain usable evidence records. The usable candidate documents (i.e., usable evidence) are clustered to determine the main evidence cluster. Subsequently, the generation process of the unit process chain and input / output item list is constrained based on the main evidence cluster. This avoids directly mixing documents from different process routes, system boundaries, or raw material systems, reducing inconsistencies in the process dataset caused by mixed evidence. During the generation of the input / output item list, the traceability of the data is ensured by constructing evidence citation relationships between the input / output item content and the usable evidence. Based on the flow matching operation and unit compatibility processing of the background flow directory, the input / output items are converted into a citationable flow dataset with flow identifiers, versions, resource addresses, and short descriptions. This ensures the traceability and publishability of the process dataset, avoiding the risk of unresolved citations when the process dataset is published. Material balance analysis is performed on the processed process dataset. Combined with flow matching and unit compatibility processing results, a quality description field is generated, achieving a data quality closed loop and ensuring data reliability, traceability, and verifiability. Therefore, even in the absence of complete enterprise-wide measured process data or with insufficient enterprise process data, a highly reliable, traceable, verifiable process dataset with a complete data quality closed loop can be automatically generated.

[0159] Figure 5 This is a schematic diagram of another process dataset generation device provided in an embodiment of this application.

[0160] For example, the device further includes a reference stream parsing module 45, used to obtain reference stream summary information based on the reference stream information and the job type of the reference stream, wherein the reference stream summary information includes the identity information of the reference stream.

[0161] The evidence processing module 41 is used for: Based on the reference stream summary information and the job type, document retrieval conditions are constructed.

[0162] Based on the literature search criteria, candidate literature records are obtained.

[0163] In response to the candidate documents containing physical information used to generate the process dataset, they are determined to be usable evidence.

[0164] Figure 6This is a schematic diagram of another process dataset generation device provided in the embodiments of this application.

[0165] For example, the device further includes: Unit processing module 46 is used to perform unit compatibility processing on the input / output items and the matching stream records, and to determine the unit compatibility processing result.

[0166] The dataset construction module 47 is used to determine the process dataset based on the unit process chain and the list of input and output items for each unit process.

[0167] The quality closed-loop module 44 is further configured to generate material balance analysis information for the process dataset, and generate the quality description field based on the unit compatibility processing result and the material balance analysis information.

[0168] In an exemplary embodiment, this application also provides a computer program product, including a computer program that, when executed by a processor, implements the method as described in any of the above embodiments.

[0169] For the device embodiments and computer program product embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. Furthermore, the device embodiments described above are illustrative; the modules described as separate components may or may not be physically separate, and the explicit components as modules may or may not be physical modules, i.e., they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this application according to actual needs. Those skilled in the art can understand and implement this without any inventive effort.

[0170] This application can be implemented by executing several computer program code flows using an electronic device. The electronic device loads the computer program into non-volatile memory and uses the processor of the electronic device to read these computer program instructions into memory for execution.

[0171] Figure 7 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. The following is in conjunction with... Figure 7 The electronic devices described in some embodiments of this application will be explained.

[0172] Reference Figure 7 Electronic devices 700 can be mobile phones, computers, digital broadcasting terminals, messaging devices, game consoles, tablets, medical devices, fitness equipment, personal digital assistants, etc.

[0173] Device 700 may include one or more of the following components: processing component 701, memory 702, power supply component 703, multimedia component 704, audio component 705, input / output (I / O) interface 706, sensor component 707, and communication component 708.

[0174] Processing component 701 typically controls the overall operation of device 700, such as operations associated with display, telephone calls, data communication, camera operation, and recording. Processing component 701 may include one or more processors 709 to execute instructions to complete all or part of the steps of the methods described above. Furthermore, processing component 701 may include one or more modules to facilitate interaction between processing component 701 and other components. For example, processing component 701 may include a multimedia module to facilitate interaction between multimedia component 704 and processing component 701.

[0175] Memory 702 is configured to store various types of data to support the operation of device 700. Examples of this data include instructions for any application or method operating on device 700, contact data, phonebook data, messages, pictures, videos, etc. Memory 702 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0176] Power supply component 703 provides power to various components of device 700. Power supply component 703 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power to device 700.

[0177] Multimedia component 704 includes a screen that provides an output interface between the device 700 and the user. In some embodiments, the screen may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen may be implemented as a touchscreen to receive input signals from the user. The touch panel includes one or more touch sensors to sense touch, swipe, and gestures on the touch panel. The touch sensors may sense not only the boundaries of the touch or swipe action but also the duration and pressure associated with the touch or swipe operation. In some embodiments, multimedia component 704 includes a front-facing camera and / or a rear-facing camera. When the device 700 is in an operating mode, such as a shooting mode or a video mode, the front-facing camera and / or the rear-facing camera may receive external multimedia data. Each front-facing camera and rear-facing camera may be a fixed optical lens system or have focal length and optical zoom capabilities.

[0178] Audio component 705 is configured to output and / or input audio signals. For example, audio component 705 includes a microphone (MIC) configured to receive external audio signals when device 700 is in an operating mode, such as call mode, recording mode, and voice recognition mode. The received audio signals may be further stored in memory 702 or transmitted via communication component 708. In some embodiments, audio component 705 also includes a speaker for outputting audio signals.

[0179] Input / output (I / O) interface 706 provides an interface between processing component 701 and peripheral interface modules, such as keyboards, click wheels, buttons, etc. These buttons may include, but are not limited to, home buttons, volume buttons, power buttons, and lock buttons.

[0180] Sensor assembly 707 includes one or more sensors for providing state assessments of various aspects of device 700. For example, sensor assembly 707 can detect the on / off state of device 700, the relative positioning of components, such as the display and keypad of device 700, changes in the position of device 700 or a component of device 700, the presence or absence of user contact with device 700, the orientation or acceleration / deceleration of device 700, and temperature changes of device 700. Sensor assembly 707 may also include a proximity sensor configured to detect the presence of nearby objects without any physical contact. Sensor assembly 707 may also include a light sensor, such as a CMOS or CCD image sensor, for use in imaging applications. In some embodiments, sensor assembly 707 may also include an accelerometer, a gyroscope, a magnetometer, a pressure sensor, or a temperature sensor.

[0181] Communication component 708 is configured to facilitate wired or wireless communication between device 700 and other devices. Device 700 can access wireless networks based on communication standards, such as WiFi, 2G or 3G, 4G or 5G, or combinations thereof. In one exemplary embodiment, communication component 708 receives broadcast signals or broadcast-related information from an external broadcast management system via a broadcast channel. In one exemplary embodiment, communication component 708 also includes a near-field communication (NFC) module to facilitate short-range communication. For example, the NFC module may be implemented based on radio frequency identification (RFID) technology, Infrared Data Association (IrDA) technology, ultra-wideband (UWB) technology, Bluetooth (BT) technology, and other technologies.

[0182] In an exemplary embodiment, device 700 may be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components.

[0183] In an exemplary embodiment, this application also provides a non-transitory computer-readable storage medium including instructions, such as a memory 702 including instructions that can be executed by a processor 709 of a device 700. For example, the non-transitory computer-readable storage medium may be a ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, or optical data storage device, etc.

Claims

1. A method for generating a lifecycle inventory process dataset, characterized in that, include: A search is performed based on the abstract information of the reference stream and the job type of the reference stream, and a record of available evidence is obtained based on the candidate document records obtained from the search. The available evidence records are clustered according to preset clustering features to obtain at least one evidence cluster, and the main evidence cluster is determined from the at least one evidence cluster. Based on the main evidence cluster, a unit process chain is determined, the unit process chain comprising multiple unit processes arranged in sequence; For each unit process, a corresponding list of input and output items is generated to determine the process dataset; For each input / output item in the background stream directory, perform stream record matching, write background stream reference information for successfully matched input / output items, and write placeholder stream references for input / output items that fail to match. Based on the unit compatibility processing results of the input / output items and the matched flow records, and the material balance analysis information for the process dataset, a quality description field for the process dataset is generated.

2. The method according to claim 1, characterized in that, The preset clustering features include at least one of system boundary, process path type, or intermediate flow.

3. The method according to claim 1, characterized in that, Before generating the quality description field of the process dataset, the method further includes: For input / output items that fail to match in the process dataset, a secondary retrieval request is reconstructed to re-match the input / output items with stream records in the background stream directory; The secondary retrieval request has more relaxed matching constraints compared to the retrieval operation of the previous stream record.

4. The method according to claim 1, characterized in that, For each unit process, a corresponding list of input and output items is generated, including performing at least one of the following operations: In response to the explicit existence of the magnitude information of the input / output item in the document recorded by the available evidence, the magnitude information is written into the input / output item, or the magnitude information after conversion is written into the input / output item; In response to the absence of quantity information for the input / output item in the literature, the estimated quantity information and its description are written into the input / output item, or the missing quantity information is written into the input / output item.

5. The method according to claim 4, characterized in that, The process of generating a corresponding list of input and output items for each unit process also includes: Based on the input / output item list, coverage information is determined; the coverage information is used to indicate the proportion of input / output items with different quantitative information acquisition methods, as well as the coverage rate of key unit processes; Based on the coverage information, determine the next operation task.

6. The method according to claim 1, characterized in that, The unit compatibility processing result of the input / output items and the matched stream records is determined by performing the following operations: Obtain the reference unit of the stream record and the physical dimension category to which the reference unit belongs; In response to the fact that the unit of the input / output item and the reference unit belong to the same physical dimension category, or that the unit of the input / output item and the reference unit support physical dimension category conversion through attribute parameters, the unit of the input / output item is converted to the reference unit; or, In response to the missing unit of the input / output item, the missing physical dimension category, or the unit of the input / output item and the reference unit not supporting physical dimension category conversion, the verification identifier is associated with the input / output item.

7. The method according to claim 1, characterized in that, The material balance analysis information is determined using the following method: Based on the direction, flow type, and physical dimension category of the input and output items, determine the input and output items used for material balance analysis; Based on the defined input and output terms, calculate the material balance closure error and / or energy balance closure error, and perform at least one of the following operations during the calculation process: Verify whether the characteristics of the input and output items meet the exclusion constraints; Verify whether the reference stream and input / output direction are mismatched; Verify the method of obtaining the value information of the input and output items, and the consistency of the units of the input and output items with the matched stream records; Verify whether the material balance closure error and / or energy balance closure error exceed the preset balance threshold.

8. The method according to claim 1, characterized in that, The quality description field includes: The Data Cutoff and Integrity Principle field is used to describe the system boundaries, the scope of unit processes covered by the process dataset, missing process data, and mismatched flow records. The "Data Processing and Extrapolation Principles" field describes the data acquisition method and provides information on data usage. The method further includes: Based on the content of the quality description field, return the settings steps in the process dataset generation method.

9. The method according to claim 1, characterized in that, The process of searching based on the abstract information of the reference stream and the job type of the reference stream, and obtaining available evidence records based on the retrieved candidate document records, includes: Based on the reference stream dataset and the job type of the reference stream, obtain reference stream summary information, which includes the identity information of the reference stream; Based on the reference stream summary information and the job type, construct the document retrieval conditions; Based on the aforementioned literature search criteria, candidate literature records are obtained; In response to the candidate documents containing physical information used to generate the process dataset, they are determined to be usable evidence.

10. The method according to claim 9, characterized in that, The step of obtaining candidate document records based on the document retrieval criteria includes at least one of the following operations: The retrieved candidate documents are deduplicated and merged. Supplementary materials are obtained from the full text of the candidate documents, and the supplementary materials are merged with the corresponding candidate documents.

11. The method according to claim 1, characterized in that, The process of obtaining available evidence records based on the retrieved candidate document records also includes: performing usage labeling on the available evidence; The input / output item list includes evidence citation information for each input / output item, and the evidence citation information is determined according to the purpose label. The method further includes: determining the source dataset based on the evidence citation information.

12. A lifecycle inventory process dataset generation device, characterized in that, include: The evidence processing module is used to perform retrieval based on the abstract information of the reference stream and the job type of the reference stream, and to obtain available evidence records based on the candidate document records obtained from the retrieval. as well as The available evidence records are clustered according to preset clustering features to obtain at least one evidence cluster, and the main evidence cluster is determined from the at least one evidence cluster. A process generation module is used to determine a unit process chain based on the master evidence cluster, wherein the unit process chain includes multiple unit processes arranged in sequence; For each unit process, a corresponding list of input and output items is generated to determine the process dataset; The stream matching module is used to perform stream record matching for each input / output item in the background stream directory, write background stream reference information for successfully matched input / output items, and write placeholder stream references for input / output items that fail to match. The quality closed-loop module is used to generate a quality description field for the process dataset based on the unit compatibility processing results of the input / output items and the matching flow records, as well as the material balance analysis information for the process dataset.

13. The apparatus according to claim 12, characterized in that, The device further includes a reference stream parsing module, used to obtain reference stream summary information based on the reference stream information and the job type of the reference stream, wherein the reference stream summary information includes the identity information of the reference stream; The evidence processing module is used for: Based on the reference stream summary information and the job type, construct the document retrieval conditions; Based on the aforementioned literature search criteria, candidate literature records are obtained; In response to the candidate documents containing physical information used to generate the process dataset, they are determined to be usable evidence.

14. The apparatus according to claim 12, characterized in that, The device further includes: The unit processing module is used to perform unit compatibility processing on the input / output items and the matching stream records, and to determine the unit compatibility processing result; A dataset construction module is used to determine the process dataset based on the unit process chain and the list of input and output items for each unit process; The quality closed-loop module is also used to generate material balance analysis information for the process dataset, and generate the quality description field based on the unit compatibility processing result and the material balance analysis information.

15. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1-11.

16. An electronic device, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method as described in any one of claims 1-11.

17. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method as described in any one of claims 1-11.