Product adjustment strategy generation method and device and processor

Abstract

The invention discloses a product adjustment strategy generation method and device and a processor. The method comprises the steps that a product document of a product in a running state is obtained, and the product document is at least used for describing function information and operation information of the product; an information extraction model is used for extracting various kinds of entity information from the product document, relation information among the various kinds of entity information is determined, and the information extraction model is obtained by training a large language model through product document samples, entity samples and relation information samples among the entity samples; the product document sample is at least used for describing function information and operation information of the product sample; based on the relation information and the entity information, constructing map information corresponding to the product document; a strategy generation instruction is responded, an adjustment strategy of the product is generated based on the map information, and the strategy generation instruction comprises operation demand information of the product. The technical problem of low generation efficiency of the product adjustment strategy is solved.

Description

Technical Field

[0001] The present invention relates to the field of knowledge graph technology, and in particular to a method, device and processor for generating an adjustment strategy for a product. Background Art

[0002] In the related art, the generation of product adjustment strategies mainly relies on expert experience and manual product document analysis. Since the generation of high-quality product adjustment strategies relies on the expert's knowledge reserve and experience accumulation, the quality and applicability of product adjustment strategies are difficult to guarantee without expert guidance. In addition, it takes a long time to analyze product documents, especially when faced with a large number of product documents and complex products, the time and energy required to generate product adjustment strategies are huge. Therefore, there is still a technical problem of low efficiency in generating product adjustment strategies.

[0003] To address the above-mentioned problems, no effective solution has been proposed yet. Summary of the invention

[0004] The embodiments of the present invention provide a method, device and processor for generating an adjustment strategy for a product, so as to at least solve the technical problem of low generation efficiency of the adjustment strategy for the product.

[0005] According to one aspect of an embodiment of the present invention, a method for generating an adjustment strategy for a product is provided, comprising: obtaining a product document of a product in operation, wherein the product document is used to at least describe functional information and operational information of the product; extracting a variety of entity information from the product document using an information extraction model, and determining relationship information between the various entity information, wherein the information extraction model is obtained by training a large language model using product document samples, entity samples, and relationship information samples between entity samples, and the product document samples are used to at least describe functional information and operational information of the product samples; constructing graph information corresponding to the product document based on the relationship information and the entity information; generating an adjustment strategy for the product based on the graph information in response to a strategy generation instruction, wherein the strategy generation instruction includes operational requirement information of the product, and the adjustment strategy is a rule for adjusting the operational status, and the adjusted operational status satisfies the operational requirement information.

[0006] Optionally, an information extraction model is used to extract a variety of entity information from product documents, including: parsing the product documents to obtain text content of the product documents, and inputting the text content into the information extraction model; using the information extraction model to infer the text content and extract key information from the text content, wherein the importance of the key information is higher than an importance threshold; and determining the key information as entity information.

[0007] Optionally, the method also includes: using an information extraction model to obtain type information of entity information, normalizing entity information of different types of information to obtain processed entity information; determining relationship information between multiple entity information, including: determining relationship information based on processed entity information and text content.

[0008] Optionally, the method also includes: constructing triple data based on entity information, relationship information and text content; processing the triple data to obtain processed triple data, wherein the accuracy of the processed triple data is higher than the accuracy of the triple data before processing.

[0009] Optionally, based on the relationship information and the entity information, constructing graph information corresponding to the product document includes: constructing the graph information based on the processed triple data; the method also includes: storing the graph information in a graph database.

[0010] Optionally, in response to a policy generation instruction, an adjustment strategy for a product is generated based on the graph information, including: in response to obtaining the policy generation instruction, parsing the policy generation instruction, converting the policy generation instruction into query information, wherein the query information is a structured policy generation instruction; matching graph information corresponding to the query information from a graph database storing graph information; and formulating an adjustment strategy based on the graph information and the query information.

[0011] Optionally, after formulating an adjustment strategy based on the graph information and the query information, the method further includes: in response to the number of adjustment strategies that meet the operating requirement information being at least two, evaluating at least two adjustment strategies to obtain an evaluation result; and sorting multiple adjustment strategies using the evaluation result to obtain a sorting result.

[0012] According to another aspect of an embodiment of the present invention, a device for generating an adjustment strategy for a product is also provided, comprising: an acquisition unit, configured to acquire a product document of a product in operation, wherein the product document is at least used to describe functional information and operational information of the product; an extraction unit, configured to extract a variety of entity information from the product document using an information extraction model, and to determine relationship information between the variety of entity information, wherein the information extraction model is obtained by training a large language model using product document samples, entity samples, and relationship information samples between entity samples, and the product document samples are at least used to describe functional information and operational information of the product samples; a construction unit, configured to construct graph information corresponding to the product document based on the relationship information and the entity information; a generation unit, configured to respond to a policy generation instruction and generate an adjustment strategy for the product based on the graph information, wherein the policy generation instruction includes operational requirement information of the product, and the adjustment strategy is a rule for adjusting the operational status, and the adjusted operational status satisfies the operational requirement information.

[0013] According to another aspect of an embodiment of the present invention, a processor is further provided, which can be used to run a program, wherein when the program is run, any one of the above-mentioned methods for generating an adjustment strategy for a product is executed.

[0014] According to another aspect of an embodiment of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores a plurality of instructions, wherein the instructions are suitable for being loaded by a processor and executing any one of the above-mentioned methods for generating an adjustment strategy for a product.

[0015] According to another aspect of an embodiment of the present invention, there is also provided an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute any one of the above-mentioned methods for generating an adjustment strategy for a product.

[0016] According to another aspect of an embodiment of the present invention, a computer program product is provided, including computer instructions, which, when executed by a processor, implement any one of the above-mentioned methods for generating an adjustment strategy for the product.

[0017] In an embodiment of the present invention, if it is necessary to generate an adjustment strategy for a product, the product document of the product in operation can be obtained. An information extraction model can be used to extract a variety of entity information from the product document, and the relationship information between the various entity information can be determined. Based on the relationship information and the entity information, the graph information (for example, a knowledge graph) corresponding to the product document can be constructed. Instructions can be generated according to the strategy, and the adjustment strategy of the product can be generated based on the graph information. In this embodiment, information extraction is automatically performed through an information extraction model (for example, a large language model), which can not only extract surface entity information, but also capture complex relationship information between entity information. According to the strategy generation instructions, and the knowledge graph constructed by the extracted entity information and relationship information, the adjustment strategy of the product can be automatically queried and intelligently generated. Thereby solving the technical problem of low efficiency in generating product adjustment strategies, and achieving the technical effect of improving the efficiency in generating product adjustment strategies. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0019] Figure 1 is a flow chart of a method for generating an adjustment strategy for a product according to an embodiment of the present invention;

[0020] Figure 2 is a flow chart of a product solution information management method according to an embodiment of the present invention;

[0021] Figure 3 is a flow chart of a method for generating product solution information according to an embodiment of the present invention;

[0022] Figure 4 is a schematic diagram of product solution information management and generation based on a large language model and knowledge graph according to an embodiment of the present invention;

[0023] Figure 5 is a structural schematic diagram of an adjustment strategy generating device for a product according to an embodiment of the present invention;

[0024] Figure 6 is a schematic diagram of an electronic device for a method for generating an adjustment strategy for a product according to an embodiment of the present invention. DETAILED DESCRIPTION

[0025] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0026] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0027] Example 1

[0028] According to an embodiment of the present invention, an embodiment of a method for generating an adjustment strategy for a product is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0029] Figure 1is a flow chart of a method for generating an adjustment strategy for a product according to an embodiment of the present invention. Figure 1 As shown, the method comprises the following steps:

[0030] Step S102, obtaining product documents of the product in operation.

[0031] In the technical solution provided in the above step S102 of the embodiment of the present invention, the product document can at least be used to describe the functional information and operation information of the product, and can be a plain text "product description type" text content, such as a product manual, a tender document, and other product documents.

[0032] In this embodiment, when it is identified that the product is in operation, the corresponding product document contains sufficient entity information and relationship information, and the product document of the product in operation can be obtained to provide accurate and comprehensive basic information for subsequent information extraction and adjustment strategy generation.

[0033] Alternatively, product documentation can be retrieved and downloaded from a document management system or knowledge base, or it can be captured from the Internet or external data sources using automated tools. For publicly released products, product documentation can be obtained through official websites, social media, forums, and other platforms. Product documentation can exist in electronic formats, such as portable document format (PDF), text documents (Word), slide presentation documents (PowerPoint, PPT), etc.

[0034] It should be noted that the above-mentioned method and specific content of obtaining product documents are only examples and are not specifically limited here. As long as the method and content can be used to obtain product documents of a product in operation, they are within the protection scope of the embodiments of the present invention.

[0035] Step S104: extracting a variety of entity information from the product document using an information extraction model, and determining relationship information between the various entity information.

[0036] In the technical solution provided in the above step S104 of the embodiment of the present invention, after obtaining the product document of the product in operation, the information extraction model can be used to extract multiple entity information from the product document and determine the relationship information between the multiple entity information. The information extraction model can be obtained by training a large language model using product document samples, entity samples, and relationship information samples between entity samples. The product document samples can at least be used to describe the functional information and operation information of the product sample.

[0037] In this embodiment, the information extraction model may be a large language model (LLM for short). LLM is based on deep learning technology and has powerful semantic understanding and text generation capabilities. In product documents, LLM can identify and annotate different entity information. Entity information may include product name, product introduction, product release time, product function, product highlight, product application scenario, etc. Relationship information may be relationship information between entity information.

[0038] Optionally, determining the relationship information between entity information is the basis for constructing graph information, which may include identifying the association between entity information, such as the relationship between product name and product function, and the relationship between product application scenario and product highlight.

[0039] Step S106, constructing graph information corresponding to the product document based on the relationship information and entity information.

[0040] In the technical solution provided in the above step S106 of the embodiment of the present invention, after extracting various entity information from the product document using the information extraction model and determining the relationship information between the various entity information, graph information corresponding to the product document can be constructed based on the relationship information and the entity information. The graph information can be graph data information or a knowledge graph.

[0041] In this embodiment, in the process of constructing graph information, since the relationship information between entity information can be clearly expressed, the graph information corresponding to the product document is constructed based on the relationship information and entity information, which can enhance the deeper connection of the graph information in the product document and more comprehensively analyze product features, user behavior, market trends, etc. Thus, a dynamic and coherent knowledge graph is constructed based on the entity information and relationship information in the product document, providing strong data support for product management and market analysis.

[0042] Step S108, in response to the strategy generation instruction, generating a product adjustment strategy based on the graph information.

[0043] In the technical solution provided in the above step S108 of the embodiment of the present invention, after constructing the graph information corresponding to the product document based on the relationship information and the entity information, the product adjustment strategy can be generated based on the graph information according to the strategy generation instruction. Among them, the strategy generation instruction may include the operation requirement information of the product, which may be triggered by user demand or query. The adjustment strategy may be called a product solution, which may be a rule for adjusting the operation status, and the adjusted operation status may meet the operation requirement information.

[0044] In an embodiment of the present invention, according to the policy generation instruction triggered by the user, the specific needs of the user can be analyzed, such as product performance improvement, function expansion, application environment adaptability, etc. The graph query technology can be used to retrieve graph information in the knowledge graph. The retrieved graph information can be integrated and analyzed. For example, by comparing different product functions, analyzing application scenario cases, identifying technology trends, etc., to ensure that the adjustment strategy of the generated product is highly matched with user needs.

[0045] Optionally, based on user needs and the rich relationship information and entity information in the knowledge graph, a customized product adjustment strategy can be generated, which can not only improve the efficiency of generating product solutions, but also ensure the pertinence and effectiveness of the adjustment strategy, allowing enterprises to quickly respond to market changes and enhance product competitiveness.

[0046] In the technical solution provided by the above steps S102 to S108 of the embodiment of the present invention, if it is necessary to generate an adjustment strategy for a product, the product document of the product in operation can be obtained. An information extraction model can be used to extract a variety of entity information from the product document, and the relationship information between the various entity information can be determined. Based on the relationship information and the entity information, the graph information (for example, the knowledge graph) corresponding to the product document can be constructed. Instructions can be generated according to the strategy, and the adjustment strategy of the product can be generated based on the graph information. In this embodiment, information extraction is automatically performed through an information extraction model (for example, a large language model), which can not only extract surface entity information, but also capture complex relationship information between entity information. Users can automatically query and intelligently generate product adjustment strategies based on the strategy generation instructions and the knowledge graph constructed by the extracted entity information and relationship information. Thereby solving the technical problem of low generation efficiency of product adjustment strategies and achieving the technical effect of improving the generation efficiency of product adjustment strategies.

[0047] The embodiment of the present invention is described in detail below in combination with the above steps.

[0048] As an optional embodiment, step S104 uses an information extraction model to extract multiple entity information from the product document, including: parsing the product document to obtain the text content of the product document, and inputting the text content into the information extraction model; using the information extraction model to infer the text content and extract key information from the text content, wherein the importance of the key information is higher than an importance threshold; and determining the key information as entity information.

[0049] In this embodiment, if it is necessary to extract multiple entity information from the product document using the information extraction model, the product document can be parsed to obtain the text content of the product document, and the text content can be input into the information extraction model. The information extraction model can be used to infer the text content and extract key information from the text content. The key information can be determined as entity information. Among them, the key information can also be called key information, and the importance of the key information is higher than the importance threshold.

[0050] Optionally, product documents such as product manuals, tenders, etc. in PDF, Word, PPT and other formats can be parsed to obtain the text content of the product description in plain text. The plain text format can facilitate subsequent information extraction and analysis.

[0051] Optionally, the parsed plain text content is input into a pre-trained information extraction model to parse the semantics and contextual relationship of the text content. After parsing the text content, the information extraction model can automatically identify and extract key information related to the preset entity information.

[0052] Optionally, in order to ensure the high quality of the extracted information, an importance threshold can be set. When the importance of the information evaluated by the information extraction model is higher than the importance threshold, it can be determined as entity information. The above entity information is a structured description of the content of the product document, which can be in a structured data format to facilitate subsequent processing and storage of the entity information.

[0053] As an optional embodiment, an information extraction model is used to obtain type information of entity information, and entity information of different types of information is normalized to obtain processed entity information; relationship information between multiple entity information is determined, including: determining relationship information based on processed entity information and text content.

[0054] In this embodiment, the information extraction model can be used to obtain the type information of the entity information, that is, the entity information type, and the entity information of different types of information can be normalized to obtain the processed entity information. If the relationship information between multiple entity information is to be determined, the relationship information can be determined based on the processed entity information and text content.

[0055] Optionally, after the entity information is extracted, it can be normalized to ensure that the same entity information is expressed consistently in different product documents. For example, different ways of writing product names (such as capitalization differences, abbreviations and full names) can be unified into a standard format to facilitate entity information identification and relationship information establishment in subsequent processing.

[0056] As an optional embodiment, triple data is constructed based on entity information, relationship information and text content; the triple data is processed to obtain processed triple data, wherein the accuracy of the processed triple data is higher than the accuracy of the triple data before processing.

[0057] In this embodiment, triple data can be constructed based on entity information, relationship information and text content. The triple data can be processed to obtain processed triple data, and deduplication, update and optimization of the triple data can be achieved. The accuracy of the processed triple data is higher than the accuracy of the triple data before processing.

[0058] Optionally, based on the normalized entity information, relationship information, and text content, the relationship between entities can be constructed to form triple data. A triple data can be composed of a subject, a predicate, and an object to describe the specific relationship between entities. For example, if the information extraction model determines that the "smart camera" (subject) "has" (predicate) "night infrared shooting function" (object), the corresponding triple data will be generated.

[0059] As an optional embodiment, step S106, constructing graph information corresponding to the product document based on relationship information and entity information, includes: constructing graph information based on processed triple data; the method also includes: storing the graph information in a graph database.

[0060] In this embodiment, if it is necessary to construct the graph information corresponding to the product document based on the relationship information and the entity information, the graph information can be constructed based on the processed triple data and stored in the graph database.

[0061] Optionally, in a graph database, nodes (corresponding to entity information) and edges (corresponding to relationship information) can be created based on triple data. Each entity information (such as product name, function description) can be used as a node, and the relationship information between entity information (such as a certain product has a certain function) can be connected to the corresponding nodes through edges. By connecting the nodes and edges created above in the graph database according to the relationship structure in the triple data, the graph information can be gradually constructed. And it can be stored in the graph database in sequence to achieve deduplication, update and optimization of triple data.

[0062] As an optional embodiment, step S108, in response to a policy generation instruction, generates an adjustment strategy for a product based on graph information, including: in response to obtaining the policy generation instruction, parsing the policy generation instruction, converting the policy generation instruction into query information, wherein the query information is a structured policy generation instruction; matching graph information corresponding to the query information from a graph database storing graph information; and formulating an adjustment strategy based on the graph information and the query information.

[0063] In this embodiment, when the acquired policy generation instruction is detected, the policy generation instruction can be parsed and converted into query information. The graph information corresponding to the query information can be matched from the graph database storing the graph information. An adjustment strategy can be formulated based on the graph information and the query information. The query information can be a structured policy generation instruction.

[0064] Optionally, LLM can parse the user's specific needs and convert them into structured query information. Based on the structured query information, accurate matching can be performed in the graph database that stores the graph information. After matching the relevant graph information, a customized solution text (adjustment strategy) can be generated based on the matching results and the user's needs, including product configuration, implementation steps, expected results, and case descriptions.

[0065] As an optional embodiment, after formulating an adjustment strategy based on graph information and query information, the method also includes: in response to the number of adjustment strategies that meet the operating requirement information being at least two, evaluating at least two adjustment strategies to obtain an evaluation result; and using the evaluation result to sort multiple adjustment strategies to obtain a sorting result.

[0066] In this embodiment, after formulating the adjustment strategy based on the graph information and the query information, when the number of adjustment strategies that meet the operation requirement information is at least two, the at least two adjustment strategies can be evaluated in combination with historical cases and expert experience to obtain an evaluation result. The evaluation result can be used to sort the multiple adjustment strategies to obtain a sorting result.

[0067] Optionally, based on the evaluation results, the generated multiple adjustment strategies can be prioritized to identify adjustment strategies with potential or that meet current needs. Based on the evaluation results and the ranking information, a list containing adjustment strategies and priorities can be generated. The order of the adjustment strategies in the list represents the priority of implementing the adjustment strategies, and the adjustment strategies with high priorities should be considered for implementation first.

[0068] In an embodiment of the present invention, if it is necessary to generate an adjustment strategy for a product, the product document of the product in operation can be obtained. An information extraction model can be used to extract a variety of entity information from the product document, and the relationship information between the various entity information can be determined. Based on the relationship information and the entity information, the graph information (for example, a knowledge graph) corresponding to the product document can be constructed. Instructions can be generated according to the strategy, and the adjustment strategy for the product can be generated based on the graph information. In this embodiment, information extraction is automatically performed through an information extraction model (for example, a large language model), which can not only extract surface entity information, but also capture complex relationship information between entity information. Users can automatically query and intelligently generate product adjustment strategies based on the strategy generation instructions and the knowledge graph constructed by the extracted entity information and relationship information. Thereby solving the technical problem of low efficiency in generating product adjustment strategies, and achieving the technical effect of improving the efficiency in generating product adjustment strategies.

[0069] Example 2

[0070] Another optional specific implementation is described in detail below.

[0071] Information management and solution generation methods mainly rely on manual information collation. Solution experts can collate and extract information based on documents such as product manuals, product instructions, product maintenance manuals, and project information such as project bids and project diaries. Solution generation can be based on the needs of specific scenarios, and relevant information can be searched in product manuals and historical project documents through keyword retrieval and other methods. Finally, experienced solution experts will manually sort out the information and summarize it into a solution. However, manual information collation is time-consuming and labor-intensive, and prone to errors. The relevant keyword retrieval method relies on keyword matching, lacks the ability to deeply understand information and mine relevance, and is difficult to meet the needs of complex scenarios. Generating a product solution is highly dependent on the integrity of existing product solution information and the professionalism of solution experts. In complex product systems and scattered demand scenarios, it will lead to problems such as incomplete information service coverage and low efficiency in generating product solutions. Therefore, there is still a technical problem of low efficiency in generating product adjustment strategies.

[0072] The embodiment of the present invention proposes a product solution information management and generation system based on a large language model and a knowledge graph. Through LLM and knowledge graph technology, information in product documents is automatically extracted and organized, manual intervention is reduced, and information management efficiency is improved. Through structured storage and relational query of knowledge graphs, the accuracy and recall rate of graph information retrieval are improved. Based on LLM and knowledge graph query technology, customized product solutions are automatically generated, reducing dependence on solution experts and improving generation efficiency. Thereby solving the technical problem of low generation efficiency of product adjustment strategies and achieving the technical effect of improving the generation efficiency of product adjustment strategies.

[0073] The method is further described below.

[0074] In this embodiment, Figure 2 is a flow chart of a product solution information management method according to an embodiment of the present invention. Figure 2 As shown, the method may include the following steps:

[0075] Step S201, perform entity extraction.

[0076] In this embodiment, the product documents, product manuals, maintenance manuals, project documents, tender documents, and other PDF, Word, PPT and other documents can be parsed to obtain the plain text "product description" text content. Based on the general semantic understanding ability of LLM, the product text content can be understood and key information can be extracted. This includes extracting entity (node) information such as product name, product function, product release date, product advantages, etc. from various product document texts, and marking the type of the entity information.

[0077] Optionally, the entity information type can be confirmed by determining the entity information list based on the characteristics of product information and expert experience, for example, including product name, product introduction, product release time, product function, product highlight, product application scenario, etc.

[0078] Optionally, entity information extraction can be centered on the product name, and the product name, product release time, product introduction, product highlights, product function list, product cases, etc. can be extracted in sequence. Product function description, product function advantages, product function trial scenarios, scenario status of product function trial scenarios, and scenario difficulties can be extracted from the product function list. User needs can be centered to extract the scenario classification, requirement introduction, detailed requirement list, requirement proposal time, and requirement personnel of user needs. Corresponding product information and product function information can be extracted based on the requirements details.

[0079] Step S202, performing entity unification.

[0080] In this embodiment, entity information unification can be performed based on the extracted entity information. The entity information extracted by LLM is normalized and sorted to facilitate subsequent storage in the graph data. Entity information unification can include normalizing entity information such as product names and product functions. At the same time, the same type of entity information extracted multiple times is supplemented and optimized to achieve accurate extraction of entity information.

[0081] Step S203, establishing a relationship.

[0082] In this embodiment, based on the extracted entity information and text information, the relationship information between entities can be constructed to form triple data.

[0083] In an embodiment of the present invention, after step S201 to step S203, graph data information storage and update can be performed and input into a graph database. For the same product (product A), there are multiple product documents (including instructions, maintenance manuals, project bids, etc.), including function 1, function 2, function 3, function 4, product description, product release date, application scenario a, and customer cases. Function 1 corresponds to function description 5, and function 2 corresponds to function description 6. Function description 6 can describe the application to application scenario b. Graph information update can be achieved based on the knowledge graph architecture and the triple information extraction capability based on LLM. There is repeated and complementary information in the triples extracted based on LLM. A triple data set is constructed based on this data, and it is sequentially stored in the graph database to achieve deduplication, update, and optimization of triple data. Based on the graph search capability, query the same data of entities / relationships, and achieve deduplication and supplementary optimization of entity relationship data based on LLM capabilities. And complete the final data storage, and finally achieve the update and improvement of graph data knowledge.

[0084] Figure 3 is a flow chart of a method for generating product solution information according to an embodiment of the present invention. Figure 3 As shown, the method may include the following steps:

[0085] Step S301, analyzing user needs and obtaining analysis results.

[0086] In this embodiment, LLM is used to understand user needs and convert them into structured query information. Precise matching is performed in the knowledge graph to find solution information entities and attribute values ​​related to the needs. In the LLM part, based on the knowledge graph information and user query, the supervised fine-tuning (SFT) technology of the large language model is used to optimize the accuracy of LLM in converting user demand information and improve the accuracy of LLM in solution information retrieval scenarios. In the knowledge graph information retrieval part, the graph retrieval effect is optimized to improve the accuracy and recall of solution graph information retrieval.

[0087] Step S302: Generate a solution based on the analysis result.

[0088] In this embodiment, after understanding user needs through LLM, solutions can be generated.

[0089] Step S303: Generate a customized solution based on the solution.

[0090] In this embodiment, based on the semantic understanding ability of LLM, a customized solution text can be generated according to the matching results and user needs, including product configuration, implementation steps, expected results and case descriptions. LLM can be customized and optimized by reinforcement learning to improve the feasibility and satisfaction of the final generated solution text. Through the reinforcement learning method, the model is trained to improve the model effect in the field of solution information text generation. Based on the base LLM model capabilities and the high-quality LLM model capabilities, multiple solution texts can be generated using the base model or the high-quality LLM model based on the same user needs and matching results. Combined with historical cases and expert experience, multiple solutions can be prioritized to form a reward modeling (RM) data set.

[0091] Step S304: Output the overall structure of the demand information.

[0092] In this embodiment, after understanding the user's needs through LLM, the demand information can be output in a structured form and input into the graph database.

[0093] Step S305, perform graph information retrieval.

[0094] In this embodiment, graph information can be retrieved from a graph database to provide information for solution generation, thereby generating a customized solution.

[0095] Figure 4is a schematic diagram of product solution information management and generation based on a large language model and knowledge graph according to an embodiment of the present invention, such as Figure 4 As shown, the product solution information management method is connected with the product solution information generation process through a graph database.

[0096] In the product solution information management method, the extraction and management of product solution information can be completed based on LLM technology and knowledge graph technology. Through LLM capabilities, entity extraction, entity unification and relationship establishment can be achieved in product documents based on unstructured information. Complete the construction of triple information, and build a graph database and data storage based on knowledge graph technology to achieve structured storage and management technology of solution information text. In the process of generating product solution information, LLM technology and knowledge graph query technology can be used. Based on the customized needs of the user, structured information can be sorted through LLM, and complete relevant content can be retrieved in the graph database based on knowledge graph query technology. And based on LLM generation technology, refer to the graph information retrieval content to complete the generation of customized solution text information. Thereby solving the technical problem of low generation efficiency of product adjustment strategy and achieving the technical effect of improving the generation efficiency of product adjustment strategy.

[0097] Example 3

[0098] The embodiment of the present invention provides a device for generating an adjustment strategy for a product. It should be noted that the device for generating an adjustment strategy for a product in the embodiment of the present invention can be used to execute Figure 1 The following is an introduction to the method for generating an adjustment strategy for a product provided by an embodiment of the present invention.

[0099] Figure 5 is a schematic diagram of the structure of an adjustment strategy generating device for a product according to an embodiment of the present invention. Figure 5 As shown, the product adjustment strategy generation device 500 may include: an acquisition unit 502 , an extraction unit 504 , a construction unit 506 and a generation unit 508 .

[0100] The acquisition unit 502 is used to acquire the product document of the product in operation, wherein the product document is used to describe at least the functional information and operation information of the product.

[0101] The extraction unit 504 is used to extract multiple entity information from the product document using an information extraction model, and determine the relationship information between the multiple entity information, wherein the information extraction model is obtained by training a large language model using product document samples, entity samples, and relationship information samples between entity samples, and the product document samples are at least used to describe the functional information and operation information of the product samples.

[0102] The construction unit 506 is used to construct graph information corresponding to the product document based on the relationship information and the entity information.

[0103] The generation unit 508 is used to respond to the strategy generation instruction and generate an adjustment strategy for the product based on the graph information, wherein the strategy generation instruction includes the operation requirement information of the product, and the adjustment strategy is a rule for adjusting the operation status, and the adjusted operation status meets the operation requirement information.

[0104] The device for generating an adjustment strategy for a product provided by an embodiment of the present invention obtains a product document of a product in operation by an acquisition unit 502, wherein the product document is at least used to describe the functional information and operation information of the product; extracts a variety of entity information from the product document by using an information extraction model by an extraction unit 504, and determines the relationship information between the various entity information, wherein the information extraction model is obtained by training a large language model using product document samples, entity samples, and relationship information samples between entity samples, and the product document samples are at least used to describe the functional information and operation information of the product samples; constructs graph information corresponding to the product document based on the relationship information and entity information by a construction unit 506; generates an adjustment strategy for the product based on the graph information in response to a strategy generation instruction by a generation unit 508, wherein the strategy generation instruction includes the operational requirement information of the product, and the adjustment strategy is a rule for adjusting the operational status, and the adjusted operational status satisfies the operational requirement information. Thus, the technical problem of low generation efficiency of the product adjustment strategy is solved, and the technical effect of improving the generation efficiency of the product adjustment strategy is achieved.

[0105] The adjustment strategy generating device of the above-mentioned product may further include a processor and a memory, wherein the above-mentioned units are all stored in the memory as program units, and the processor executes the above-mentioned program units stored in the memory to realize corresponding functions.

[0106] The processor includes a kernel, and the kernel retrieves the corresponding program unit from the memory. One or more kernels can be set to control the same device type of devices to be shut down to perform graceful shutdown by adjusting kernel parameters.

[0107] The above-mentioned memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one storage chip.

[0108] The processor contains a kernel, which retrieves the corresponding program unit from the memory. One or more kernels can be set, and the work efficiency of traders can be improved by adjusting the kernel parameters.

[0109] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0110] Example 4

[0111] According to an embodiment of the present invention, there is further provided a computer-readable storage medium on which a program is stored. When the program is executed by a processor, the method for generating an adjustment strategy for a product is implemented.

[0112] Example 5

[0113] According to an embodiment of the present invention, a processor is further provided. The processor is used to run a program, wherein the method for generating an adjustment strategy for a product is executed when the program is run.

[0114] Example 6

[0115] Figure 6 is a schematic diagram of an electronic device for a method for generating an adjustment strategy for a product according to an embodiment of the present invention, such as Figure 6 As shown, an embodiment of the present invention further provides an electronic device, which includes a processor, a memory, and a program stored in the memory and executable on the processor, and a method for generating a product adjustment strategy is implemented when the processor executes the program.

[0116] The devices in this article can be servers, PCs, PADs, mobile phones, etc.

[0117] Example 7

[0118] The invention also provides a computer program product, which, when executed on a data processing device, is suitable for executing a program of initializing a method for generating an adjustment strategy for a product.

[0119] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0120] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0121] These computer program instructions may also be stored in a computer readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0122] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0123] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0124] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0125] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transit media), such as modulated data signals and carrier waves.

[0126] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0127] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0128] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.

Claims

1. A method for generating a product adjustment strategy, characterized in that: include: Acquire a product document of a product in operation, wherein the product document is used to describe at least functional information and operational information of the product; Extracting a plurality of entity information from the product document using an information extraction model, and determining relationship information between the plurality of entity information, wherein the information extraction model is obtained by training a large language model using product document samples, entity samples, and relationship information samples between the entity samples, and the product document samples are used to describe at least functional information and operational information of the product samples; Based on the relationship information and the entity information, construct graph information corresponding to the product document; In response to a strategy generation instruction, an adjustment strategy for the product is generated based on the graph information, wherein the strategy generation instruction includes the operating requirement information of the product, and the adjustment strategy is a rule for adjusting the operating state, and the adjusted operating state satisfies the operating requirement information.

2. The method according to claim 1, characterized in that The information extraction model is used to extract various entity information from the product document, including: Parsing the product document to obtain text content of the product document, and inputting the text content into the information extraction model; Using the information extraction model, reasoning about the text content, and extracting key information from the text content, wherein the importance of the key information is higher than an importance threshold; The key information is determined as the entity information.

3. The method according to claim 2, characterized in that The method further comprises: Using the information extraction model, obtaining type information of the entity information, Normalizing the entity information of different types of information to obtain processed entity information; Determine the relationship information between the plurality of entity information, including: The relationship information is determined based on the processed entity information and the text content.

4. The method according to claim 3, characterized in that The method further comprises: Constructing triple data based on the entity information, the relationship information and the text content; The triple data are processed to obtain the processed triple data, wherein the accuracy of the processed triple data is higher than the accuracy of the triple data before the processing.

5. The method according to claim 4, characterized in that Based on the relationship information and the entity information, construct graph information corresponding to the product document, including: Constructing the graph information based on the processed triple data; The method also includes: storing the graph information in a graph database.

6. The method according to claim 1, characterized in that In response to the strategy generation instruction, based on the graph information, an adjustment strategy for the product is generated, including: In response to obtaining the policy generation instruction, parsing the policy generation instruction and converting the policy generation instruction into query information, wherein the query information is the structured policy generation instruction; Matching the graph information corresponding to the query information from a graph database storing the graph information; The adjustment strategy is formulated based on the graph information and the query information.

7. The method according to claim 6, characterized in that After formulating the adjustment strategy based on the graph information and the query information, the method further includes: In response to the number of the adjustment strategies that meet the operation requirement information being at least two, evaluating at least two of the adjustment strategies to obtain an evaluation result; The plurality of adjustment strategies are sorted using the evaluation result to obtain a sorting result.

8. A device for generating an adjustment strategy for a product, characterized in that: include: an acquisition unit, configured to acquire a product document of a product in operation, wherein the product document is used to describe at least functional information and operation information of the product; an extraction unit, configured to extract a plurality of entity information from the product document using an information extraction model, and determine relationship information between the plurality of entity information, wherein the information extraction model is obtained by training a large language model using product document samples, entity samples, and relationship information samples between the entity samples, and the product document samples are used to describe at least functional information and operational information of the product samples; A construction unit, configured to construct graph information corresponding to the product document based on the relationship information and the entity information; A generation unit is used to respond to a strategy generation instruction and generate an adjustment strategy for the product based on the graph information, wherein the strategy generation instruction includes operation requirement information of the product, and the adjustment strategy is a rule for adjusting the operation state, and the adjusted operation state satisfies the operation requirement information.

9. A processor, characterized in that: The processor is used to run a program, wherein the program, when run by the processor, executes the product adjustment strategy generation method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute the method for generating an adjustment strategy for a product according to any one of claims 1 to 7.

Images