Multi-source heterogeneous supply chain data extraction method and system

Abstract

The invention is suitable for the technical field of data extraction, and particularly relates to a multi-source heterogeneous supply chain data extraction method and system, and the method comprises the steps: S100: obtaining a source of supply chain data, classifying the supply chain data collected by the source into a pre-constructed data pool, carrying out the standardization of the supply chain data, and storing the standardized supply chain data in a database; identifying a key entity, wherein the key entity at least comprises a supplier, a logistics node, a transportation mode and a warehouse; s200, collecting attributes of the key entity, wherein the attributes at least comprise; geographic positions of suppliers, processing capability of logistics nodes and warehouse capacity. According to the method, the supply chain data can be integrated and visually displayed by constructing the multi-modal knowledge graph, the extraction accuracy of the supply chain data is greatly improved, the extraction speed of the supply chain data is greatly improved by defining the extraction target and the template, and a user can make a decision more quickly and accurately.

Description

Technical Field

[0001] The present invention relates to the technical field of data extraction, and particularly to a method and system for extracting multi-source heterogeneous supply chain data. Background Art

[0002] Multi-source heterogeneity refers to data from multiple different sources and in different formats, which differ in structure, type, and semantics. How to extract valuable information from multi-source heterogeneous supply chain data is the key to eliminating the heterogeneity and complexity of supply chain data.

[0003] A knowledge graph can represent data in a structured manner, with rich semantic information, dynamic update, and scalability. It can integrate multi-source heterogeneous supply chain data into a unified graphical model, thereby performing integration processing and visual display on supply chain data. Therefore, "how to use a knowledge graph to extract supply chain data" is the technical problem to be solved by the present invention. Summary of the Invention

[0004] The purpose of the present invention is to provide a method and system for extracting multi-source heterogeneous supply chain data to solve the problem of "how to use a knowledge graph to extract supply chain data" raised in the above background art.

[0005] To achieve the above purpose, the present invention provides the following technical solutions:

[0006] A method for extracting multi-source heterogeneous supply chain data, the method comprising:

[0007] S100: Obtain the source of the supply chain data, classify the supply chain data collected via the source into a pre-constructed data pool, standardize the supply chain data, and identify key entities, where the key entities at least include: suppliers, logistics nodes, transportation modes, and warehouses;

[0008] S200: Collect the attributes of the key entities, where the attributes at least include; the geographical location of the supplier, the processing capacity of the logistics node, and the warehouse capacity, traverse the key entities, attributes, and the corresponding relationships between the key entities and attributes, and configure the connections between the supply chain data;

[0009] S300: Determine the graph structure, where the graph structure includes: graphics, text, and data structure, create a multi-modal knowledge graph, build a connection link between the multi-modal knowledge graph and the data pool, and transfer the supply chain data in the data pool to the multi-modal knowledge graph via the connection link;

[0010] S400: Split the multi-modal knowledge graph into several levels, where the levels include: vocabulary, sentences, and paragraphs, and mark the connections into the levels, define extraction targets and templates, and integrate them into the multi-modal knowledge graph.

[0011] Further, the S100 includes:

[0012] Configure write ports corresponding to the sources one by one. When write data is received, establish a mapping between the write data and the current time;

[0013] Generate description tags for the sources, and insert the description tags into the supply chain data, where the tags at least include: supplier name and data collection device number.

[0014] Further, the S100 also includes:

[0015] Create a name comparison table, where the name comparison table includes: name items and corresponding items, and one name item corresponds to at least one corresponding item;

[0016] Convert the supply chain data into a unified sentence pattern, and use the sentence pattern and the name comparison table to standardize the supply chain data;

[0017] Identify and extract key entities in the standardized supply chain data, sort the supply chain data in the data pool according to the key entities, and create a linked list according to the sorting result, and embed a growth mechanism.

[0018] Further, the S200 includes:

[0019] Insert the attributes into the linked list, and mark the connections;

[0020] Obtain production tasks, and use the connections to evaluate the production tasks to determine whether they are feasible;

[0021] If feasible, activate the production tasks. If not feasible, generate an evaluation report.

[0022] Further, the S300 includes:

[0023] Create graph architectures for several independent multi-modal knowledge graphs, and insert root nodes into the multi-modal knowledge graph, where the root nodes are used to manage all graph architectures and supply chain data;

[0024] Determine the format of the supply chain data, and transfer the linked list into the corresponding graph architecture via the format to generate a multi-modal knowledge graph;

[0025] Synchronize the connection links to the multi-modal knowledge graph according to the connections.

[0026] Further, the S400 includes:

[0027] Using the hierarchy, refine the linked list, and based on natural language processing technology, traverse the semantic associations between different levels of the linked list and add the semantic associations to the connections;

[0028] Use the extraction target to retrieve in the multimodal knowledge graph, fill the retrieval results into the template to obtain the extraction results;

[0029] Output the extraction results through the multimodal knowledge graph.

[0030] Further, the method further includes:

[0031] Determine the initial version of the multimodal knowledge graph, number it, and construct an upgrade mechanism for the initial version and the number;

[0032] Based on the mapping, compare the changed levels in the multimodal knowledge graph, define them as incremental levels, and synchronize the numbers to the incremental levels;

[0033] Store the initial version and the incremental levels to construct a version chain.

[0034] Further, the system includes:

[0035] An identification module, configured to obtain the source of the supply chain data, classify the supply chain data collected via the source into a pre-constructed data pool, standardize the supply chain data, and identify key entities, where the key entities at least include: suppliers, logistics nodes, transportation modes, and warehouses;

[0036] A configuration module, configured to collect the attributes of the key entities, where the attributes at least include; the geographical location of the supplier, the processing capacity of the logistics node, and the warehouse capacity, traverse the key entities, attributes, and the corresponding relationships between the key entities and attributes, and configure the connections between the supply chain data;

[0037] A transfer module, configured to determine the graph structure, where the graph structure includes: graphics, text, and data structures, create a multimodal knowledge graph, build a connection link between the multimodal knowledge graph and the data pool, and transfer the supply chain data in the data pool to the multimodal knowledge graph via the connection link;

[0038] A segmentation module, configured to segment the multimodal knowledge graph into several levels, where the levels include: vocabulary, sentences, and paragraphs, and mark the connections to the levels, define the extraction target and the template, and integrate them into the multimodal knowledge graph.

[0039] Furthermore, the recognition module includes:

[0040] A mapping unit for configuring a write port corresponding to each source one by one. After receiving write data, it establishes a mapping between the write data and the current moment.

[0041] An insertion unit for generating a description tag of the source and inserting the description tag into the supply chain data, where the tag at least includes: supplier name and data collection device number.

[0042] A creation unit for creating a name comparison table, where the name comparison table includes: name items and comparison items, and one name item corresponds to at least one comparison item.

[0043] A conversion unit for converting the supply chain data into a unified sentence pattern, and using the sentence pattern and the name comparison table to standardize the supply chain data.

[0044] An embedding unit for identifying and extracting key entities in the standardized supply chain data, sorting the supply chain data in the data pool according to the key entities, and creating a linked list according to the sorting result and embedding a growth mechanism.

[0045] Furthermore, the configuration module includes:

[0046] A marking unit for inserting the attribute into the linked list and marking the connection.

[0047] A judgment unit for obtaining a production task, evaluating the production task using the connection, judging whether it is feasible, activating the production task if it is feasible, and generating an evaluation report if it is not feasible.

[0048] Compared with the prior art, the beneficial effects of the present invention are:

[0049] 1. By constructing a data pool, it is possible to aggregate supply chain data from different sources, thereby obtaining more comprehensive supply chain data and providing decision support, further optimizing the effect of supply chain management and strategy formulation. By configuring connections, it is possible to improve the visibility of the supply chain, enhance the synergy effect, and improve the accuracy of decision-making. By constructing a multi-modal knowledge graph, it is possible to integrate supply chain data and perform visual display, greatly improving the readability of supply chain data. By defining extraction targets and templates, the extraction efficiency of supply chain data is greatly improved, facilitating users to make decisions more quickly and accurately.

[0050] 2. By performing incremental backup on the multi-modal knowledge graph, it is convenient to centrally manage supply chain data, optimize the supply chain process, and enhance the communication and collaboration efficiency of supply chain data. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings in the following description are only some embodiments of the present invention.

[0052] Figure 1 It is a flowchart of a multi-source heterogeneous supply chain data extraction method provided by an embodiment of the present invention;

[0053] Figure 2 It is a first sub-flowchart of a multi-source heterogeneous supply chain data extraction method provided by an embodiment of the present invention;

[0054] Figure 3 It is a second sub-flowchart of a multi-source heterogeneous supply chain data extraction method provided by an embodiment of the present invention;

[0055] Figure 4 It is a third sub-flowchart of a multi-source heterogeneous supply chain data extraction method provided by an embodiment of the present invention;

[0056] Figure 5 It is a fourth sub-flowchart of a multi-source heterogeneous supply chain data extraction method provided by an embodiment of the present invention;

[0057] Figure 6 It is a block diagram of the composition of a multi-source heterogeneous supply chain data extraction system provided by an embodiment of the present invention;

[0058] Figure 7 It is a block diagram of the composition of the recognition module in a multi-source heterogeneous supply chain data extraction system provided by an embodiment of the present invention;

[0059] Figure 8 It is a block diagram of the composition of the configuration module in a multi-source heterogeneous supply chain data extraction system provided by an embodiment of the present invention;

[0060] Figure 9 It is a block diagram of the composition of the transfer module in a multi-source heterogeneous supply chain data extraction system provided by an embodiment of the present invention;

[0061] Figure 10 It is a block diagram of the composition of the segmentation module in a multi-source heterogeneous supply chain data extraction system provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0062] In order to make the objectives, technical solutions and advantages of the present invention more comprehensible, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0063] In Embodiment 1,Figure 1 The implementation process of the multi-source heterogeneous supply chain data extraction method provided by the embodiments of the present invention is shown, and the details are as follows:

[0064] S100: Obtain the sources of supply chain data, classify the supply chain data collected via the sources into a pre-constructed data pool, standardize the supply chain data, and identify key entities, where the key entities at least include: suppliers, logistics nodes, transportation modes, and warehouses.

[0065] Obtain the sources of supply chain data, where the sources can be supplier management systems, purchase orders, production data, logistics data, etc. Collect supply chain data through the sources and classify the collected supply chain data into the data pool; the data pool is a structure for storing data, which can centrally store supply chain data from different sources and provide a unified storage location for convenient management and access to supply chain data.

[0066] Standardize the supply chain data. Standardization is to convert the supply chain data into a unified sentence pattern for convenient processing of the supply chain data, and identify the key entities in the supply chain data. The key entities are the participating entities in the supply chain data, and the key entities include: suppliers, warehousing manufacturers, or transportation personnel, etc.

[0067] S200: Collect the attributes of the key entities, where the attributes at least include; the geographical location of the supplier, the processing capacity of the logistics node, and the warehouse capacity, traverse the key entities, attributes, and the corresponding relationships between the key entities and attributes, and configure the relationships between the supply chain data.

[0068] Collect the attributes of the key entities, where the attributes are the location of the supplier, warehousing capacity, and remaining warehousing volume, etc. Establish the corresponding relationships within the key entities, within the attributes, and between the key entities and attributes, and map the corresponding relationships to the supply chain data to determine the relationships.

[0069] For example, supplier A transports 10 tons of raw materials to warehousing manufacturer B. Warehousing manufacturer B transports the raw materials to manufacturer C regularly and quantitatively according to the monthly demand of 4 tons. On the 12th of the month, warehousing manufacturer B transports the first batch of 2 tons of raw materials through transport vehicle D; among them, after the transportation is completed, the remaining warehousing volume of warehousing manufacturer B changes from 12 tons to 10 tons, and the remaining warehousing volume of manufacturer C changes from 1 ton to 3 tons.

[0070] In the above example, the key entities are A, B, C, and D, the attributes are the specific values, and correspondingly, the corresponding relationships include B and C, C and D, B and D, 12 and 10, 1 and 3, D and 10, and D and 3; the relationship is the qualitative visualization of the above corresponding relationships.

[0071] S300: Determine the atlas architecture, where the atlas architecture includes: graphics, text, and data structures, create a multi-modal knowledge graph, build a connection link between the multi-modal knowledge graph and the data pool, and transfer the supply chain data in the data pool to the multi-modal knowledge graph via the connection link.

[0072] Determine a number of atlas architectures, where the atlas architecture includes: a graphics atlas architecture, a text atlas architecture, and a data structure atlas architecture. The graphics atlas architecture is mainly used to receive supply chain data with graphics as the main body; similarly, the text atlas architecture is mainly used to receive supply chain data with text as the main body, and the data structure atlas architecture mainly receives supply chain data with data structures as the main body.

[0073] Use each atlas architecture to construct a knowledge graph respectively, integrate all the knowledge graphs, generate a multi-modal knowledge graph, and classify the supply chain data in the data pool into the multi-modal knowledge graph.

[0074] S400: Split the multi-modal knowledge graph into several levels, where the levels include: vocabulary, sentences, and paragraphs, and mark the relationships in the levels, define extraction targets and templates, and integrate them into the multi-modal knowledge graph.

[0075] After the multi-modal knowledge graph is constructed, split the multi-modal knowledge graph into several levels, and each level corresponds to a data type; for example, split the knowledge graph corresponding to text into levels of vocabulary, sentences, and paragraphs, and each level is used to process supply chain data of different lengths, so as to further refine the supply chain data.

[0076] After the refinement is completed, determine the extraction target, retrieve in the multi-modal knowledge graph, find the corresponding data, and integrate the found data into the template to complete the extraction of the supply chain data.

[0077] In Embodiment 2, Figure 2 The implementation process of the multi-source heterogeneous supply chain data extraction method provided by the embodiment of the present invention is shown. The following details S100 as follows:

[0078] S101: Configure a write port corresponding to each source, and when the write data is received, establish a mapping between the write data and the current time.

[0079] Configure a write port for each source, and after the write data is received, record the write port and time corresponding to the write data, etc. The advantage of doing this is to facilitate tracing back to the original supply chain data.

[0080] S102: Generate a description tag for the source and insert the description tag into the supply chain data, where the tag at least includes: the supplier name and the data acquisition device number.

[0081] Obtain a description tag for the source, where the description tag is the description information of the source, including: the number of the purchase order, the specific value of the production data, or the logistics number, etc., and can also be the supplier name and the data acquisition device number, etc.

[0082] In Embodiment 3, Figure 2 The implementation process of the multi-source heterogeneous supply chain data extraction method provided by the embodiment of the present invention is shown. The following details S100 in detail as follows:

[0083] S103: Create a name comparison table, where the name comparison table includes: a name item and a comparison item, and one name item corresponds to at least one comparison item.

[0084] Create a name comparison table. By using the name comparison table, the supply chain data can be better integrated.

[0085] For example, in the supply chain data from different sources, for the key entity of "supplier", there may be multiple aliases, such as: "supplier", "provider", or "supplier code", etc. By querying the name comparison table, the supply chain data corresponding to these aliases can be integrated to the same position in the multi-modal knowledge graph.

[0086] S104: Convert the supply chain data into a unified sentence pattern, and use the sentence pattern and the name comparison table to standardize the supply chain data.

[0087] Convert the supply chain data into a unified sentence pattern, that is, convert the text or chart data in the supplier data into a unified expression; for example, the text data in the supplier data can be converted by using the subject-predicate-object method. By converting into a unified sentence pattern and combining with the comparison table, the supply chain data can be standardized.

[0088] S105: Identify and extract the key entities in the standardized supply chain data, sort the supply chain data in the data pool according to the key entities, and create a linked list according to the sorting result and embed a growth mechanism.

[0089] Extract the key entities from the standardized supply chain data, sort the key entities according to the order of product or raw material transfer, build a linked list according to the sorting result, and at the same time embed a growth mechanism, where the growth mechanism means that when a new transfer of raw materials occurs, new supply chain data will be generated, and the new supply chain data will be inserted into the corresponding position in the linked list.

[0090] In Embodiment 4, Figure 3 The implementation process of the multi-source heterogeneous supply chain data extraction method provided by the embodiment of the present invention is shown. The following details S200 as follows:

[0091] S201: Insert the attribute into the linked list and mark the connection.

[0092] Insert the attribute into the corresponding supplier data in the linked list and mark the connection in the linked list for the supplier data.

[0093] S202: Obtain the production task, and use the connection to evaluate the production task to determine whether it is feasible;

[0094] If it is feasible, activate the production task; if it is not feasible, generate an evaluation report.

[0095] Collect the production task, where the production task can be the transfer of raw materials, the use of raw materials, etc. At the same time, evaluate the production task according to the connection to determine whether the production task is feasible.

[0096] For example, based on production requirements, the production task of Party A is to require 14 tons of raw materials. However, by querying the connection, there is only a remaining space of 10 tons in the warehouse corresponding to Party A, that is, this production task is not feasible. Then an evaluation report can be generated and sent to a preset terminal; the evaluation report should include conclusions and specific solutions.

[0097] In Embodiment 5, Figure 4 The implementation process of the multi-source heterogeneous supply chain data extraction method provided by the embodiment of the present invention is shown. The following details S300 as follows:

[0098] S301: Create several independent graph architectures of multi-modal knowledge graphs, and insert root nodes into the multi-modal knowledge graphs, where the root nodes are used to manage all graph architectures and supply chain data.

[0099] Create several graph architectures, where the graph architectures include: graphics, texts, data structures, etc. By creating these graph architectures, it is possible to collect supply chain data in different formats and structures, and insert root nodes into the multi-modal knowledge graphs. The root nodes are mainly used to manage all graph architectures and supply chain data to avoid data chaos.

[0100] S302: Determine the format of the supply chain data, and transfer the linked list into the corresponding graph architecture via the format to generate a multi-modal knowledge graph.

[0101] After receiving the supply chain data, determine the format of the supply chain data, transfer the supply chain data into the corresponding graph schema, integrate the supply chain data in all graph schemas, and generate a multi-modal knowledge graph.

[0102] S303: Synchronize the connection link to the multi-modal knowledge graph according to the connection.

[0103] In the multi-modal knowledge graph, establish connection links for different supply chain data; when it is necessary to extract the full-process supply chain data of a certain raw material, all can be extracted through the connection link, greatly shortening the extraction process and improving the extraction efficiency.

[0104] In Embodiment 6, Figure 5 shows the implementation process of the multi-source heterogeneous supply chain data extraction method provided by the embodiment of the present invention. The following details S400 as follows:

[0105] S401: Refine the linked list using the hierarchy, traverse the semantic associations between different levels of the linked list based on natural language processing technology, and add the semantic associations to the connection.

[0106] Refine the supply chain data in the linked list to levels, determine the semantic associations between different levels of the supply chain data, establish the associations of the supply chain data in different levels by using natural language processing technology, and add them to the connection; by traversing the semantic associations in different levels, the connection can be expanded and the supply chain data can be analyzed more comprehensively.

[0107] S402: Use the extraction target to retrieve in the multi-modal knowledge graph, fill the retrieval result into the template, and obtain the extraction result.

[0108] Use the extraction target to retrieve in the multi-modal knowledge graph, determine the retrieval result, where the retrieval result is the required supply chain data, fill the retrieval result into the template, and obtain the extraction result.

[0109] S403: Output the extraction result through the multi-modal knowledge graph.

[0110] Input the extraction target and the template into the multi-modal knowledge graph, and output to obtain the extraction result.

[0111] In Embodiment 7, different from Embodiment 1, in the embodiment of the present invention, the method further includes:

[0112] Determine the initial version of the multi-modal knowledge graph, number it, and construct an upgrade mechanism for the initial version and the number;

[0113] Based on the mapping, the changed levels in the multimodal knowledge graph are compared and defined as incremental levels, and the numbers are synchronized to the incremental levels;

[0114] Store the initial version and the incremental levels to build a version chain.

[0115] After the multimodal knowledge graph is constructed, the multimodal knowledge graph is determined as the initial version, and numbers are determined. At the same time, an upgrade mechanism for the initial version and the numbers is determined; the upgrade mechanism is that when new supply chain data is generated, the supply chain data is written into the initial version to obtain the first modified version, and the number of the first modified version is determined, and so on, until the nth modified version; of course, during this process, the numbers of each modified version also need to be determined accordingly.

[0116] After new supply chain data is written into the multimodal knowledge graph, the specific level is determined, and this level is determined as the incremental level. At the same time, the corresponding numbers are written into the incremental level; the initial version and all incremental levels are stored in sequence to generate a version chain, so as to perform incremental storage on all supply chain data; when all supply chain data needs to be called, the current supply chain data can be obtained by integrating the initial version and all incremental levels.

[0117] Figure 6 The block diagram of the composition structure of the multi-source heterogeneous supply chain data extraction system provided by the embodiment of the present invention is shown. The multi-source heterogeneous supply chain data extraction system 1 includes:

[0118] The recognition module 11 is used to obtain the source of the supply chain data, classify the supply chain data collected via the source into the pre-constructed data pool, standardize the supply chain data, and identify key entities, where the key entities at least include: suppliers, logistics nodes, transportation methods, and warehouses;

[0119] The configuration module 12 is used to collect the attributes of the key entities, where the attributes at least include; the geographical location of the supplier, the processing capacity of the logistics node, and the warehouse capacity, traverse the key entities, attributes, and the corresponding relationships between the key entities and attributes, and configure the connections between the supply chain data;

[0120] The transfer module 13 is used to determine the graph structure, where the graph structure includes: graphics, text, and data structures, create a multimodal knowledge graph, build a communication link between the multimodal knowledge graph and the data pool, and transfer the supply chain data in the data pool to the multimodal knowledge graph via the communication link;

[0121] The splitting module 14 is used to split the multi-modal knowledge graph into several levels, where the levels include: vocabulary, sentences, and paragraphs, and mark the relationships into the levels, define extraction targets and templates, and integrate them into the multi-modal knowledge graph.

[0122] Figure 7 The block diagram of the composition structure of the multi-source heterogeneous supply chain data extraction system provided by the embodiment of the present invention is shown. The recognition module 11 includes:

[0123] The mapping unit 111 is used to configure a write port corresponding to each source one by one. When receiving write data, establish the mapping between the write data and the current moment.

[0124] The insertion unit 112 is used to generate a description label for the source and insert the description label into the supply chain data, where the label at least includes: supplier name and data collection device number.

[0125] The creation unit 113 is used to create a name comparison table, where the name comparison table includes: name items and corresponding items, and one name item corresponds to at least one corresponding item.

[0126] The conversion unit 114 is used to convert the supply chain data into a unified sentence pattern, and use the sentence pattern and the name comparison table to standardize the supply chain data.

[0127] The embedding unit 115 is used to identify and extract key entities in the standardized supply chain data, sort the supply chain data in the data pool according to the key entities, and create a linked list according to the sorting result, and embed a growth mechanism.

[0128] Figure 8 The block diagram of the composition structure of the multi-source heterogeneous supply chain data extraction system provided by the embodiment of the present invention is shown. The configuration module 12 includes:

[0129] The marking unit 121 is used to insert the attribute into the linked list and mark the relationship.

[0130] The judgment unit 122 is used to obtain a production task, evaluate the production task using the relationship, judge whether it is feasible, if it is feasible, activate the production task, and if it is not feasible, generate an evaluation report.

[0131] Figure 9 The block diagram of the composition structure of the multi-source heterogeneous supply chain data extraction system provided by the embodiment of the present invention is shown. The transfer module 13 includes:

[0132] The management unit 131 is used to create the graph architectures of several independent multi-modal knowledge graphs and insert root nodes into the multi-modal knowledge graphs, where the root nodes are used to manage all the graph architectures and supply chain data;

[0133] The generation unit 132 is used to determine the format of the supply chain data, and transfer the linked list into the corresponding graph architecture via the format to generate a multi-modal knowledge graph;

[0134] The synchronization unit 133 is used to synchronize the connection link into the multi-modal knowledge graph according to the connection.

[0135] Figure 10 The block diagram of the composition of the multi-source heterogeneous supply chain data extraction system provided by the embodiment of the present invention is shown. The segmentation module 14 includes:

[0136] The adding unit 141 is used to refine the linked list by using the hierarchy, traverse the semantic associations between different levels of the linked list based on natural language processing technology, and add the semantic associations to the connection;

[0137] The filling unit 142 is used to retrieve in the multi-modal knowledge graph by using the extraction target and fill the retrieval result into the template to obtain the extraction result;

[0138] The output unit 143 is used to output the extraction result through the multi-modal knowledge graph.

[0139] Among them, the recognition module 11 is mainly used to complete step S100, the configuration module 12 is mainly used to complete step S200, the transfer module 13 is mainly used to complete step S300, and the segmentation module 14 is mainly used to complete step S400;

[0140] The mapping unit 111 is mainly used to complete step S101, the insertion unit 112 is mainly used to complete step S102, the creation unit 113 is mainly used to complete step S103, the conversion unit 114 is mainly used to complete step S104, and the embedding unit 115 is mainly used to complete step S105;

[0141] The marking unit 121 is mainly used to complete step S201, and the judgment unit 122 is mainly used to complete step S202;

[0142] The management unit 131 is mainly used to complete step S301, the generation unit 132 is mainly used to complete step S302, and the synchronization unit 133 is mainly used to complete step S303;

[0143] The adding unit 141 is mainly used to complete step S401, the filling unit 142 is mainly used to complete step S402, and the output unit 143 is mainly used to complete step S403.

[0144] The technical features of the above-described embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope described in this specification.

[0145] The above-described embodiments only represent several implementation manners of the present invention. The description is relatively specific and detailed, but it should not be construed as a limitation on the scope of the patent for the present invention. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present invention, several modifications and improvements can still be made, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the patent for the present invention shall be subject to the appended claims.

[0146] The foregoing is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A multi-source heterogeneous supply chain data extraction method, characterized in that: The method comprises: S100: Acquire the source of supply chain data, classify the supply chain data collected from the source into a pre-built data pool, standardize the supply chain data, and identify key entities, wherein the key entities include at least: suppliers, logistics nodes, transportation methods, and warehouses; S200: Collecting attributes of the key entity, wherein the attributes at least include: the geographical location of the supplier, the processing capacity of the logistics node and the warehouse capacity, traversing the key entity, the attributes and the corresponding relationship between the key entity and the attributes, and configuring the connection between the supply chain data; S300: Determine a graph architecture, the graph architecture including: graphics, text and data structure, create a multimodal knowledge graph, build a connection link between the multimodal knowledge graph and a data pool, and transfer the supply chain data in the data pool to the multimodal knowledge graph via the connection link; S400: Divide the multimodal knowledge graph into several levels, wherein the levels include: vocabulary, sentences and paragraphs, and mark the connections into the levels, define extraction targets and templates, and integrate them into the multimodal knowledge graph.

2. The method for extracting multi-source heterogeneous supply chain data according to claim 1 is characterized in that: The S100 includes: configuring a write port corresponding to the source one by one, and establishing a mapping between the write data and the current time after receiving the write data; Generate a description tag of the source and insert the description tag into the supply chain data, wherein the tag at least includes: a supplier name and a data collection device number.

3. The multi-source heterogeneous supply chain data extraction method according to claim 2 is characterized in that: The S100 further includes: Creating a name comparison table, wherein the name comparison table includes: name items and comparison items, and one name item corresponds to at least one comparison item; Converting the supply chain data into a unified sentence structure, and standardizing the supply chain data using the sentence structure and a name comparison table; Identify and extract key entities in the standardized supply chain data, sort the supply chain data in the data pool according to the key entities, and create a linked list based on the sorting results and embed the growth mechanism.

4. The method for extracting multi-source heterogeneous supply chain data according to claim 3 is characterized in that: The S200 includes: Inserting the attribute into the linked list and marking the relationship; Obtaining a production task, and using the connection to evaluate the production task to determine whether it is feasible; If feasible, activate the production task; if not feasible, generate an evaluation report.

5. The method for extracting multi-source heterogeneous supply chain data according to claim 4 is characterized in that: The S300 includes: Creating a graph architecture of several independent multimodal knowledge graphs, inserting a root node into the multimodal knowledge graph, wherein the root node is used to manage all graph architectures and supply chain data; Determine the format of the supply chain data, and transfer the linked list into the corresponding graph architecture through the format to generate a multimodal knowledge graph; Based on the connection, the connection link is synchronized into the multimodal knowledge graph.

6. The method for extracting multi-source heterogeneous supply chain data according to claim 5 is characterized in that: The S400 includes: Using the levels, the linked list is refined, and based on natural language processing technology, the semantic associations between different levels of the linked list are traversed, and the semantic associations are added to the connection; Using the extraction target to search in the multimodal knowledge graph, filling the search results into the template to obtain the extraction results; The extraction results are output through the multimodal knowledge graph.

7. The method for extracting multi-source heterogeneous supply chain data according to claim 2 is characterized in that: The method further comprises: Determine the initial version of the multimodal knowledge graph, number it, and build an upgrade mechanism for the initial version and number; Based on the mapping, the changed levels in the multimodal knowledge graph are compared and defined as incremental levels, and the numbers are synchronized to the incremental levels; The initial version and incremental levels are stored to build a version chain.

8. A multi-source heterogeneous supply chain data extraction system, characterized in that: The system comprises: An identification module is used to obtain the source of supply chain data, classify the supply chain data collected from the source into a pre-built data pool, standardize the supply chain data, and identify key entities, wherein the key entities include at least: suppliers, logistics nodes, transportation methods, and warehouses; A configuration module is used to collect the attributes of the key entities, wherein the attributes include at least: the geographical location of the supplier, the processing capacity of the logistics node and the warehouse capacity, traverse the key entities, attributes and the corresponding relationship between the key entities and attributes, and configure the connection between the supply chain data; A transfer module is used to determine a graph architecture, which includes graphics, text, and data structures, create a multimodal knowledge graph, build a connection link between the multimodal knowledge graph and a data pool, and transfer the supply chain data in the data pool to the multimodal knowledge graph via the connection link; A segmentation module is used to segment the multimodal knowledge graph into several levels, wherein the levels include: vocabulary, sentences and paragraphs, and mark the connections into the levels, define extraction targets and templates, and integrate them into the multimodal knowledge graph.

9. According to the multi-source heterogeneous supply chain data extraction system of claim 8, the identification module comprises: A mapping unit, configured to configure a write port corresponding to the source one by one, and to establish a mapping between the write data and the current moment after receiving the write data; An inserting unit, configured to generate a description tag of the source and insert the description tag into the supply chain data, wherein the tag includes at least: a supplier name and a data collection device number; A creating unit, used to create a name comparison table, wherein the name comparison table includes: name items and comparison items, and one name item corresponds to at least one comparison item; A conversion unit, used to convert the supply chain data into a unified sentence pattern, and standardize the supply chain data using the sentence pattern and a name comparison table; The embedding unit is used to identify and extract key entities in the standardized supply chain data, sort the supply chain data in the data pool according to the key entities, and create a linked list based on the sorting results and embed the growth mechanism.

10. The multi-source heterogeneous supply chain data extraction system according to claim 9, wherein the configuration module comprises: A marking unit, used for inserting the attribute into the linked list and marking the relationship; The judgment unit is used to obtain the production task, evaluate the production task using the connection, and judge whether it is feasible. If it is feasible, activate the production task; if it is not feasible, generate an evaluation report.

Images