Accurate policy pushing method, system and equipment driven by enterprise portrait, and medium

Abstract

The invention relates to an enterprise portrait-driven policy accurate pushing method, system and device and a medium. The method comprises the following steps: acquiring policy document information, and performing semantic embedding on the policy document information based on an embedded vector model to obtain a policy rule label vector and a policy content vector; obtaining enterprise portrait data, and performing semantic embedding on the enterprise portrait data based on the embedded vector model to generate an enterprise portrait label vector; performing vector similarity calculation on the policy rule label vector and the enterprise portrait label vector based on a vector database to obtain an enterprise policy matching result corresponding to the enterprise portrait label vector; and according to a policy matching result, in combination with the policy content vector and the policy rule label vector, generating policy push information of the enterprise. By adopting the method, a service system in which the policy and the enterprise are accurately matched is constructed, so that the two-way trouble that the enterprise cannot be found by the policy and the enterprise cannot reach the policy is solved, the enterprise development is assisted, and the regional economic efficiency is stimulated.

Description

Technical Field

[0001] This application belongs to the field of policy delivery, and in particular relates to a method, system, device and medium for precise policy delivery driven by enterprise profiling. Background Technology

[0002] The "Opinions on Further Optimizing the Business Environment and Reducing Institutional Transaction Costs for Market Entities" explicitly proposes to compile applicable preferential policies for market entities in the region and field. It emphasizes strengthening the collection and sharing of enterprise information, classifying and profiling enterprises, and promoting the intelligent matching and rapid implementation of preferential policies. The "List of Policies to Promote Steady and Sound Economic Growth and Quality Improvement in 2025" points out the need to optimize policies supporting and benefiting enterprises, comprehensively review these policies, and incorporate them into a classified platform management system. Against this backdrop, how to resolve the dual dilemma of "policies not being able to find enterprises, and enterprises not being able to accurately understand policies" has become a key research focus.

[0003] Traditional policy delivery technologies often rely on manual screening of policy documents or simple keyword matching. This typically only enables bulk distribution of policy information and fails to provide differentiated matching based on the characteristics of different enterprises. It also struggles to dynamically respond to policy updates and changes in enterprise operations. Enterprises, in turn, must sift through a large volume of policy information to find suitable content, which is not only inefficient but also prone to overlooking key policy resources. Furthermore, existing policy delivery technologies struggle to fully consider the differentiated characteristics of enterprises, such as industry attributes, current development status, and specific needs, making it difficult to meet their diverse requirements. The lack of real-time update mechanisms also hinders the rapid delivery of policies to target enterprises after their release, impacting the actual effectiveness of policy services. Summary of the Invention

[0004] Therefore, it is necessary to provide a service system that can build a precise matching between policies and enterprises, thereby helping enterprises develop and stimulating regional economic efficiency. This system should be driven by enterprise profiles and include a method, system, equipment, and medium for precise policy delivery.

[0005] Firstly, this application provides a method for precise policy delivery driven by enterprise profiling, including:

[0006] Obtain policy document information and perform semantic embedding on the policy document information based on the embedding vector model to obtain policy rule tag vectors and policy content vectors;

[0007] Acquire enterprise profile data and perform semantic embedding on the enterprise profile data based on the embedding vector model to generate enterprise profile tag vectors;

[0008] Based on the vector database, the vector similarity between policy rule label vectors and enterprise profile label vectors is calculated to obtain the policy matching results of the enterprise corresponding to the enterprise profile label vector.

[0009] Based on the policy matching results, and combining the policy content vector and the policy rule tag vector, policy push information for enterprises is generated.

[0010] In one embodiment, vector similarity is calculated between policy rule label vectors and enterprise profile label vectors based on a vector database to obtain the policy matching result of the enterprise corresponding to the enterprise profile label vector, including:

[0011] Based on the policy rule label vectors stored in the policy application rule vector database, obtain the enterprise profile label vectors stored in the enterprise profile label vector database corresponding to the policy rule label vectors;

[0012] Calculate the vector semantic similarity between the policy rule label vector and the corresponding enterprise profile label vector;

[0013] If the semantic similarity of the vectors is greater than the preset semantic compliance threshold, a metadata filtering condition group corresponding to the enterprise profile tag vector of the policy rule tag vector is set based on the policy rule tag vector. The metadata filtering condition group includes one or more metadata filtering conditions.

[0014] Based on the enterprise profile tag vector and combined with metadata filtering condition groups, policy matching results for the enterprise are generated.

[0015] In one embodiment, the policy matching results include those meeting all policy conditions and those meeting only some policy conditions. Based on the enterprise's enterprise profile tag vector and combined with metadata filtering condition groups, the enterprise's policy matching results are generated, including:

[0016] Extract the enterprise profile tag vector components that correspond to the metadata filtering conditions from the enterprise profile tag vector;

[0017] If the metadata filtering condition group includes a metadata filtering condition, when the enterprise profile tag vector component meets the metadata filtering condition, the enterprise's policy matching result will be set to meet the complete policy condition.

[0018] If the metadata filtering condition group includes multiple metadata filtering conditions, when all enterprise profile tag vector components meet the corresponding metadata filtering conditions, the enterprise's policy matching result will be set to meet the complete policy conditions.

[0019] If a metadata filtering condition group includes multiple metadata filtering conditions, when some enterprise profile tag vector components meet the corresponding metadata filtering conditions, and there are enterprise profile tag vector components that exceed the metadata filtering conditions, the enterprise's policy matching result will be set to meet some policy conditions.

[0020] In one embodiment, the policy push information includes policy content push information and work guidance push information. Based on the policy matching results and the policy content vector, the enterprise's policy push information is generated, including:

[0021] If the policy matching result of the enterprise meets the complete policy conditions, obtain the policy content vector corresponding to the policy rule tag vector, and generate the enterprise's policy content push information based on the policy content vector;

[0022] If the policy matching result of an enterprise meets some policy conditions, obtain the metadata filtering conditions corresponding to the enterprise profile tag vector components that exceed the metadata filtering conditions, and generate work guidance push information for the enterprise based on the metadata filtering conditions.

[0023] In one embodiment, the vector database includes a policy application rule vector database, a policy application content vector database, and an enterprise profile tag vector database. The enterprise profile-driven policy precision push method also includes:

[0024] Store policy rule label vectors in the policy declaration rule vector database;

[0025] Store policy content vectors in a policy application content vector database;

[0026] Store the enterprise profile tag vectors in the enterprise profile tag vector database.

[0027] In one embodiment, the enterprise profile-driven policy precision delivery method further includes:

[0028] A vector database is built based on the Milvus vector database.

[0029] In one embodiment, the enterprise profile-driven policy precision delivery method further includes:

[0030] An embedding vector model is constructed based on the embedding vector model of the Beijing Academy of Artificial Intelligence.

[0031] Secondly, this application also provides a policy precision delivery system driven by enterprise profiles, including:

[0032] The policy data management module is used to acquire policy document information and perform semantic embedding on the policy document information based on the embedding vector model to obtain policy rule tag vectors and policy content vectors.

[0033] The enterprise data management module is used to acquire enterprise profile data and perform semantic embedding on the enterprise profile data based on the embedding vector model to generate enterprise profile tag vectors.

[0034] The tag vector matching module is used to calculate the vector similarity between policy rule tag vectors and enterprise profile tag vectors based on the vector database, and obtain the policy matching result of the enterprise corresponding to the enterprise profile tag vector.

[0035] The policy information push module is used to generate policy push information for enterprises based on policy matching results, combined with policy content vectors and policy rule tag vectors.

[0036] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the method as described in any of the first aspects of this application.

[0037] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described in any of the first aspects of this application.

[0038] The aforementioned enterprise profiling-driven policy delivery methods, systems, devices, and media, through the semantic embedding of policy document information and enterprise profiling data using an embedded vector model, can accurately map policy clauses and enterprise characteristics in the vector space. This overcomes the limitations of traditional technologies based on keyword matching or rule templates, improving the accuracy of policy delivery. Furthermore, it supports dynamic updates to the policy database and enterprise data, thus adapting to real-time changes in the policy environment and enterprise characteristics. Communication based on the vector database performs vector similarity calculations on policy rule tag vectors and enterprise profiling tag vectors, capturing deep semantic relationships between policy clauses and enterprise characteristics, thereby achieving more accurate matching results. By combining policy content vectors and policy rule tag vectors to generate policy delivery information, it ensures a high degree of alignment between the delivered content and the actual needs of enterprises, guaranteeing policy implementation. Attached Figure Description

[0039] To more clearly illustrate the technical solutions in the embodiments or related technologies of this application, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0040] Figure 1 A flowchart illustrating a method for precise policy delivery driven by enterprise profiles, provided as an embodiment of this application. Figure 1 ;

[0041] Figure 2 A flowchart illustrating a method for precise policy delivery driven by enterprise profiles, provided as an embodiment of this application. Figure 2 ;

[0042] Figure 3 This is a schematic diagram of a process for generating policy matching results, provided as an embodiment of this application;

[0043] Figure 4 This is a schematic diagram of the structure of a policy precision push system driven by enterprise profile, provided as an embodiment of this application. Detailed Implementation

[0044] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0045] In one exemplary embodiment of this application, such as Figure 1 As shown, a method for precise policy delivery driven by enterprise profiles is provided. This embodiment uses the application of this method to a policy delivery platform including a policy delivery terminal and a policy delivery server, and its implementation through the interaction between the policy delivery terminal and the policy delivery server, as an example for illustration. It can be understood that this method can also be applied to a policy delivery terminal, and it can also be applied to a policy delivery server. In this embodiment, the method includes the following steps:

[0046] Step S101: Obtain policy document information and perform semantic embedding on the policy document information based on the embedding vector model to obtain policy rule tag vector and policy content vector.

[0047] Specifically, the policy push platform can collect policy document information through the policy push terminal and send the obtained policy document information to the policy push server. The policy push server can input the policy document information into the embedding vector model mounted on the policy push server to perform semantic embedding on the policy document information and obtain policy rule tag vectors and policy content vectors.

[0048] Optionally, the policy content vector may include, but is not limited to, a policy content detail vector component and a policy application method vector component.

[0049] Optionally, the policy rule label vector may include, but is not limited to, the application condition vector component and the deadline vector component.

[0050] Furthermore, the application condition vector components may include, but are not limited to, the application condition vector components for registered address, industry type, enterprise size, and enterprise qualification.

[0051] Optionally, the policy delivery platform can store policy rule tag vectors and policy content vectors based on a vector database.

[0052] Step S102: Obtain enterprise profile data and perform semantic embedding on the enterprise profile data based on the embedding vector model to generate enterprise profile tag vectors.

[0053] Specifically, the policy push platform can collect enterprise profile data through the policy push terminal and send the acquired enterprise profile data to the policy push server. The policy push server can input the enterprise profile data into the embedding vector model mounted on the policy push server to perform semantic embedding on the enterprise profile data and generate enterprise profile tag vectors.

[0054] Optionally, the enterprise profile tag vector may include, but is not limited to, the enterprise profile tag vector component of registered address, the enterprise profile tag vector component of industry type, the enterprise profile tag vector component of enterprise size, and the enterprise profile tag vector component of enterprise qualification.

[0055] Optionally, the policy delivery platform can store enterprise profile tag vectors based on a vector database.

[0056] Step S103: Calculate the vector similarity between the policy rule label vector and the enterprise profile label vector based on the vector database to obtain the policy matching result of the enterprise corresponding to the enterprise profile label vector.

[0057] Specifically, the policy delivery platform can calculate vector similarity between policy rule tag vectors extracted from policy document information and enterprise profile tag vectors extracted from enterprise profile data, based on a vector database. If the semantic similarity of the vectors exceeds a preset semantic compliance threshold, the policy delivery platform can set metadata filtering conditions for the corresponding enterprise profile tag vectors based on the policy rule tag vectors. The platform can then generate policy matching results for each enterprise based on its enterprise profile tag vector and the metadata filtering conditions. These policy matching results can be used to indicate whether an enterprise meets the policy application conditions corresponding to the policy rule tag vector.

[0058] Optionally, the policy push platform can calculate the vector similarity between the registered address application condition vector component in the policy rule tag vector extracted from policy document information and the registered address enterprise profile tag vector component in the enterprise profile tag vector extracted from enterprise profile data, based on a vector database. If the semantic similarity between the registered address application condition vector component and the registered address enterprise profile tag vector component is greater than a preset semantic compliance threshold, the policy push platform can set registration address filtering conditions corresponding to the registered address enterprise profile tag vector component in the policy rule tag vector.

[0059] Optionally, the policy push platform can calculate the vector similarity between the industry type application condition vector component in the application condition vector component of the policy rule tag vector extracted from policy document information and the industry type enterprise profile tag vector component in the enterprise profile tag vector extracted from enterprise profile data, based on a vector database. If the vector semantic similarity between the industry type application condition vector component and the industry type enterprise profile tag vector component is greater than a preset semantic compliance threshold, the policy push platform can set industry type filtering conditions corresponding to the industry type enterprise profile tag vector component in the industry type application condition vector component of the policy rule tag vector.

[0060] Optionally, the policy push platform can calculate the vector similarity between the enterprise size application condition vector component in the policy rule tag vector extracted from policy document information and the enterprise size enterprise profile tag vector component in the enterprise profile tag vector extracted from enterprise profile data, based on a vector database. If the vector semantic similarity between the enterprise size application condition vector component and the enterprise size enterprise profile tag vector component is greater than a preset semantic compliance threshold, the policy push platform can set enterprise size filtering conditions corresponding to the enterprise size application condition vector component in the policy rule tag vector.

[0061] Optionally, the policy push platform can calculate the vector similarity between the enterprise qualification application condition vector component in the policy rule tag vector extracted from policy document information and the enterprise qualification enterprise profile tag vector component in the enterprise profile tag vector extracted from enterprise profile data, based on a vector database. If the vector semantic similarity between the enterprise qualification application condition vector component and the enterprise qualification enterprise profile tag vector component is greater than a preset semantic compliance threshold, the policy push platform can set enterprise qualification screening conditions corresponding to the enterprise qualification enterprise profile tag vector component in the policy rule tag vector.

[0062] For example, if a company's enterprise profile tag vector meets all the metadata filtering conditions of the enterprise profile tag vector corresponding to the policy rule tag vector based on the policy rule tag vector settings, the policy push platform can set the policy matching result of the enterprise corresponding to the enterprise profile tag vector to meet the complete policy conditions.

[0063] For example, if a company's enterprise profile tag vector does not meet the metadata filtering conditions corresponding to the policy rule tag vector set based on the policy rule tag vector, and the company's enterprise profile tag vector does meet the metadata filtering conditions corresponding to the policy rule tag vector set based on the policy rule tag vector, the policy push platform can set the policy matching result of the company corresponding to the enterprise profile tag vector to meet some policy conditions.

[0064] Step S104: Based on the policy matching results, and combining the policy content vector and the policy rule tag vector, generate policy push information for the enterprise.

[0065] Specifically, the policy push platform can generate policy push information for enterprises based on the generated policy matching results, combined with the policy content vectors and policy rule tag vectors stored in the vector database.

[0066] For example, if a company's policy matching result is that it meets all policy conditions, the policy push platform can generate policy content push information for the company based on the policy content vectors stored in the vector database.

[0067] For example, if a company's policy matching result is that it meets some policy conditions, the policy push platform can obtain the metadata filtering conditions corresponding to the company profile tag vector components that exceed the metadata filtering conditions. The policy push platform can then generate work guidance push information for the company based on the metadata filtering conditions corresponding to the company profile tag vector components that exceed the metadata filtering conditions.

[0068] The aforementioned enterprise profiling-driven policy precision delivery method employs an embedding vector model to semantically embed policy document information and enterprise profiling data. This allows for the extraction of rule clauses and content information from policy documents into policy rule tag vectors and policy content vectors, respectively. This enables dynamic updates to the policy database and enterprise data, adapting to real-time changes in the policy environment and enterprise characteristics, and improving the accuracy of policy delivery. Calculating the vector similarity between policy rule tag vectors and enterprise profiling tag vectors using a vector database avoids policy omissions and mismatches common in traditional techniques, improving the efficiency of the policy-enterprise matching process and enhancing the accuracy of the matching results. By combining the policy matching results with policy content vectors and policy rule tag vectors to generate policy delivery information for enterprises, a high degree of alignment between the delivered content and the actual needs of enterprises is ensured. This enhances the relevance and practicality of the policy delivery information, enabling enterprises to quickly and accurately obtain matching policy content and ensuring policy implementation.

[0069] In an optional embodiment of this application, please refer to Figure 1 and Figure 2 Step S103 involves calculating the vector similarity between the policy rule label vector and the enterprise profile label vector based on the vector database to obtain the policy matching result of the enterprise corresponding to the enterprise profile label vector. This may include:

[0070] Step S206: Obtain the enterprise profile label vector stored in the enterprise profile label vector database corresponding to the policy rule label vector stored in the policy declaration rule vector database.

[0071] Step S207: Calculate the vector semantic similarity between the policy rule label vector and the corresponding enterprise profile label vector.

[0072] Step S208: If the semantic similarity of the vectors is greater than the preset semantic compliance threshold, set the metadata filtering condition group for the enterprise profile tag vector corresponding to the policy rule tag vector based on the policy rule tag vector.

[0073] Optionally, a metadata filter group may include one metadata filter, or it may include multiple metadata filter conditions.

[0074] Step S209: Based on the enterprise profile tag vector and combined with the metadata filtering condition group, generate the enterprise's policy matching result.

[0075] In the aforementioned enterprise profiling-driven policy precision delivery method, by retrieving policy rule tag vectors from the policy application rule vector database and then using these vectors to target and associate corresponding enterprise profile tag vectors in the enterprise profile tag vector database, it is possible to avoid indiscriminate searching in a massive amount of enterprise profile vectors, reduce interference from irrelevant vectors, improve the matching efficiency between policies and enterprises, and reduce computational resource consumption. Furthermore, by using metadata filtering condition groups, implicit constraints in policy rules can be transformed into explicit filtering criteria, thereby obtaining clear and specific policy adaptation standards and enhancing the practicality of policy matching.

[0076] In an optional embodiment of this application, the policy matching result may include compliance with all policy conditions and compliance with some policy conditions. Please refer to [reference needed]. Figure 2 and Figure 3 Step S209, based on the enterprise's enterprise profile tag vector and combined with metadata filtering condition groups, generates the enterprise's policy matching results, which may include:

[0077] Step S301: Extract the enterprise profile tag vector components that correspond to the metadata filtering conditions from the enterprise profile tag vector.

[0078] Step S302: If the metadata filtering condition group includes a metadata filtering condition, when the enterprise profile tag vector component meets the metadata filtering condition, the enterprise's policy matching result is set to meet the complete policy conditions.

[0079] Step S303: If the metadata filtering condition group includes multiple metadata filtering conditions, when all enterprise profile tag vector components meet the corresponding metadata filtering conditions, the enterprise's policy matching result is set to meet the complete policy conditions.

[0080] Step S304: If the metadata filtering condition group includes multiple metadata filtering conditions, when some enterprise profile tag vector components meet the corresponding metadata filtering conditions, and there are enterprise profile tag vector components that exceed the metadata filtering conditions, the enterprise's policy matching result is set to meet some policy conditions.

[0081] In an optional embodiment of this application, the policy push information may include policy content push information and work guidance push information. Based on the policy matching result and the policy content vector, the enterprise's policy push information is generated, which may include:

[0082] Specifically, if a company's policy matching result is that it meets all policy conditions, the policy push platform can obtain the policy content vector corresponding to the policy rule tag vector, and the policy push platform can generate policy content push information for the company based on the policy content vector.

[0083] Specifically, if a company's policy matching result is that it meets some policy conditions, the policy push platform can obtain the metadata filtering conditions corresponding to the company profile tag vector components that exceed the metadata filtering conditions. The policy push platform can then generate work guidance push information for the company based on the metadata filtering conditions.

[0084] In an optional embodiment of this application, the vector database may include a policy declaration rule vector database, a policy declaration content vector database, and an enterprise profile tag vector database. Please refer to [reference needed]. Figure 2 Enterprise profiling-driven methods for precise policy delivery can also include:

[0085] Step S202: Store the policy rule label vector in the policy declaration rule vector database.

[0086] Step S203: Store the policy content vector in the policy declaration content vector database.

[0087] Step S205: Store the enterprise profile tag vector in the enterprise profile tag vector database.

[0088] In an optional embodiment of this application, the enterprise profile-driven policy precision push method may further include:

[0089] Specifically, the policy delivery platform can build a vector database based on the Milvus vector database.

[0090] Optionally, Milvus Vector Database is a cloud-native system designed for storing, indexing, and retrieving high-dimensional embedded vectors. Its underlying architecture is geared towards high-dimensional vectors after unstructured data transformation. Milvus Vector Database can achieve millisecond-level similarity search, supports similarity search, scalar filtering, distributed scaling, and cloud-native deployment, and can handle tens of billions to trillions of vectors.

[0091] In an optional embodiment of this application, the enterprise profile-driven policy precision push method may further include:

[0092] Specifically, the policy delivery platform can build an embedded vector model based on the embedded vector model of the Beijing Academy of Artificial Intelligence.

[0093] Optionally, the Beijing Academy of Artificial Intelligence (BAAI) embedding vector model is a text embedding model based on the Transformer architecture, trained through instruction fine-tuning and contrastive learning. The BAAI embedding vector model can map sentences or paragraphs into dense vectors of fixed dimensions. The BAAI embedding vector model can characterize semantic similarity using vector space distance. The output of the BAAI embedding vector model can be directly fed into vector databases such as Milvus to complete fast semantic retrieval.

[0094] In one exemplary embodiment of this application, such as Figure 2 As shown, a method for precise policy delivery driven by enterprise profiles is provided, including:

[0095] Step S201: Obtain policy document information and perform semantic embedding on the policy document information based on the embedding vector model to obtain policy rule tag vector and policy content vector.

[0096] Step S202: Store the policy rule label vector in the policy declaration rule vector database.

[0097] Step S203: Store the policy content vector in the policy declaration content vector database.

[0098] Step S204: Obtain enterprise profile data and perform semantic embedding on the enterprise profile data based on the embedding vector model to generate enterprise profile tag vectors.

[0099] Step S205: Store the enterprise profile tag vector in the enterprise profile tag vector database.

[0100] Step S206: Obtain the enterprise profile label vector stored in the enterprise profile label vector database corresponding to the policy rule label vector stored in the policy declaration rule vector database.

[0101] Step S207: Calculate the vector semantic similarity between the policy rule label vector and the corresponding enterprise profile label vector.

[0102] Step S208: If the semantic similarity of the vectors is greater than the preset semantic compliance threshold, set the metadata filtering condition group for the enterprise profile tag vector corresponding to the policy rule tag vector based on the policy rule tag vector.

[0103] Step S209: Based on the enterprise profile tag vector and combined with the metadata filtering condition group, generate the enterprise's policy matching result.

[0104] Step S210: Based on the policy matching results, and combining the policy content vector and the policy rule tag vector, generate policy push information for the enterprise.

[0105] The aforementioned enterprise profiling-driven policy delivery method can achieve deep semantic association between policy rules and enterprise profiles by embedding vector models, accurately identifying the inherent logic between policy application conditions and enterprise characteristics, and effectively solving the dual dilemma of "policies not finding enterprises and enterprises not understanding policies." Based on vector similarity calculation and dynamic metadata filtering mechanisms, it can generate refined matching results that meet complete or partial policy conditions, significantly improving the conversion rate and ease of operation of enterprise policy applications, helping enterprises efficiently obtain appropriate policy dividends and optimize their own development paths. In turn, it can ensure that policy dividends accurately reach the target group, accelerate the implementation of regional policies and industrial synergy, and fully stimulate the innovation vitality of regional economy.

[0106] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0107] Based on the same inventive concept, this application also provides a system for implementing the enterprise profile-driven policy precision push method described above. The solution provided by this system is similar to the implementation described in the above method. Therefore, the specific limitations of one or more enterprise profile-driven policy precision push system embodiments provided below can be found in the limitations of the enterprise profile-driven policy precision push method described above, and will not be repeated here.

[0108] In one exemplary embodiment, such as Figure 4 As shown, a policy precision push system 400 driven by enterprise profiles is provided, which may include:

[0109] The policy data management module 401 can be used to obtain policy document information and perform semantic embedding on the policy document information based on the embedding vector model to obtain policy rule tag vectors and policy content vectors.

[0110] The enterprise data management module 402 can be used to acquire enterprise profile data and perform semantic embedding on the enterprise profile data based on the embedding vector model to generate enterprise profile tag vectors.

[0111] The tag vector matching module 403 can be used to calculate the vector similarity between policy rule tag vectors and enterprise profile tag vectors based on a vector database, and obtain the policy matching result of the enterprise corresponding to the enterprise profile tag vector.

[0112] The policy information push module 404 can be used to generate policy push information for enterprises based on policy matching results, combined with policy content vectors and policy rule tag vectors.

[0113] In an optional embodiment of this application, the label vector matching module 403 can also be used for:

[0114] Based on the policy rule label vectors stored in the policy declaration rule vector database, obtain the enterprise profile label vectors corresponding to the policy rule label vectors stored in the enterprise profile label vector database.

[0115] Calculate the vector semantic similarity between the policy rule label vector and the corresponding enterprise profile label vector.

[0116] If the semantic similarity of the vectors is greater than the preset semantic compliance threshold, a metadata filtering condition group corresponding to the enterprise profile tag vector of the policy rule tag vector is set based on the policy rule tag vector. The metadata filtering condition group includes one or more metadata filtering conditions.

[0117] Based on the enterprise profile tag vector and combined with metadata filtering condition groups, policy matching results for the enterprise are generated.

[0118] In an optional embodiment of this application, the label vector matching module 403 can also be used for:

[0119] Extract the enterprise profile tag vector components that correspond to the metadata filtering conditions from the enterprise profile tag vector.

[0120] If a metadata filtering condition group includes a metadata filtering condition, when the enterprise profile tag vector component meets the metadata filtering condition, the enterprise's policy matching result will be set to meet the complete policy conditions.

[0121] If a metadata filtering condition group includes multiple metadata filtering conditions, when all enterprise profile tag vector components meet the corresponding metadata filtering conditions, the enterprise's policy matching result will be set to meet the complete policy conditions.

[0122] If a metadata filtering condition group includes multiple metadata filtering conditions, when some enterprise profile tag vector components meet the corresponding metadata filtering conditions, and there are enterprise profile tag vector components that exceed the metadata filtering conditions, the enterprise's policy matching result will be set to meet some policy conditions.

[0123] In an optional embodiment of this application, the policy information push module 404 can also be used for:

[0124] If the policy matching result of the enterprise meets the complete policy conditions, obtain the policy content vector corresponding to the policy rule tag vector, and generate the enterprise's policy content push information based on the policy content vector.

[0125] If the policy matching result of an enterprise meets some policy conditions, obtain the metadata filtering conditions corresponding to the enterprise profile tag vector components that exceed the metadata filtering conditions, and generate work guidance push information for the enterprise based on the metadata filtering conditions.

[0126] In an optional embodiment of this application, the enterprise profile-driven policy precision push system 400 can also be used for:

[0127] Store policy rule label vectors in the policy declaration rule vector database.

[0128] The policy content vector is stored in the policy declaration content vector database.

[0129] Store the enterprise profile tag vectors in the enterprise profile tag vector database.

[0130] In an optional embodiment of this application, the enterprise profile-driven policy precision push system 400 can also be used for:

[0131] A vector database is built based on the Milvus vector database.

[0132] In an optional embodiment of this application, the enterprise profile-driven policy precision push system 400 can also be used for:

[0133] An embedding vector model is constructed based on the embedding vector model of the Beijing Academy of Artificial Intelligence.

[0134] In one embodiment, a computer device is provided, including a memory and a processor, the memory storing a computer program, the processor executing the computer program to implement the steps of the enterprise profile-driven policy precision push method as described above.

[0135] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps in the above method embodiments.

[0136] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The device embodiments described above are merely illustrative. The components described as separate parts may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this disclosure according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0137] The above-described embodiments are merely illustrative of several implementation methods of the embodiments of this application, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of the patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the embodiments of this application, and these modifications and improvements all fall within the protection scope of the embodiments of this application.

Claims

1. A method for precise policy delivery driven by enterprise profiling, characterized in that, The method includes: Obtain policy document information and perform semantic embedding on the policy document information based on the embedding vector model to obtain policy rule tag vectors and policy content vectors; Obtain enterprise profile data, and semantically embed the enterprise profile data based on the embedding vector model to generate enterprise profile tag vectors; Based on the vector database, the vector similarity between the policy rule label vector and the enterprise profile label vector is calculated to obtain the policy matching result of the enterprise corresponding to the enterprise profile label vector. Based on the policy matching results, and combining the policy content vector and the policy rule tag vector, the policy push information for the enterprise is generated.

2. The method according to claim 1, characterized in that, The step of calculating the vector similarity between the policy rule label vector and the enterprise profile label vector based on the vector database to obtain the policy matching result of the enterprise corresponding to the enterprise profile label vector includes: Based on the policy rule tag vector stored in the policy application rule vector database, obtain the enterprise profile tag vector stored in the enterprise profile tag vector database corresponding to the policy rule tag vector; Calculate the vector semantic similarity between the policy rule label vector and the corresponding enterprise profile label vector; If the semantic similarity of the vector is greater than the preset semantic compliance threshold, a metadata filtering condition group corresponding to the enterprise profile tag vector is set based on the policy rule tag vector. The metadata filtering condition group includes one or more metadata filtering conditions. Based on the enterprise profile tag vector and the metadata filtering condition group, the policy matching result of the enterprise is generated.

3. The method according to claim 2, characterized in that, The policy matching results include those meeting all policy conditions and those meeting only some policy conditions. The generation of the policy matching results for the enterprise, based on the enterprise profile tag vector and the metadata filtering condition group, includes: Extract the enterprise profile tag vector components corresponding to the metadata filtering conditions from the enterprise profile tag vector; If the metadata filtering condition group includes one of the metadata filtering conditions, when the enterprise profile tag vector component meets the metadata filtering condition, the enterprise's policy matching result is set to meet the complete policy conditions; If the metadata filtering condition group includes multiple metadata filtering conditions, when all of the enterprise profile tag vector components meet the corresponding metadata filtering conditions, the policy matching result of the enterprise is set to meet the complete policy conditions. If the metadata filtering condition group includes multiple metadata filtering conditions, when some of the enterprise profile tag vector components meet the corresponding metadata filtering conditions, and there are enterprise profile tag vector components that exceed the metadata filtering conditions, the policy matching result of the enterprise is set to meet some policy conditions.

4. The method according to claim 3, characterized in that, The policy push information includes policy content push information and work guidance push information. Generating the enterprise's policy push information based on the policy matching result and the policy content vector includes: If the policy matching result of the enterprise is that it meets the complete policy conditions, obtain the policy content vector corresponding to the policy rule tag vector, and generate the policy content push information of the enterprise based on the policy content vector; If the policy matching result of the enterprise is that it meets some of the policy conditions, obtain the metadata filtering conditions corresponding to the enterprise profile tag vector components that exceed the metadata filtering conditions, and generate the enterprise's work guidance push information based on the metadata filtering conditions.

5. The method according to any one of claims 1 to 4, characterized in that, The vector database includes a policy application rule vector database, a policy application content vector database, and an enterprise profile tag vector database; the method further includes: The policy rule tag vector is stored in the policy declaration rule vector database; The policy content vector is stored in the policy declaration content vector database; The enterprise profile tag vector is stored in the enterprise profile tag vector database.

6. The method according to claim 5, characterized in that, The method further includes: The vector database is constructed based on the Milvus vector database.

7. The method according to any one of claims 1 to 4, characterized in that, The method further includes: The embedding vector model is constructed based on the embedding vector model of the Beijing Academy of Artificial Intelligence.

8. A policy precision delivery system driven by enterprise profiling, characterized in that, The system includes: The policy data management module is used to acquire policy document information and perform semantic embedding on the policy document information based on the embedding vector model to obtain policy rule tag vectors and policy content vectors. The enterprise data management module is used to acquire enterprise profile data and perform semantic embedding on the enterprise profile data based on the embedding vector model to generate enterprise profile tag vectors. The tag vector matching module is used to calculate the vector similarity between the policy rule tag vector and the enterprise profile tag vector based on the vector database, and to obtain the policy matching result of the enterprise corresponding to the enterprise profile tag vector. The policy information push module is used to generate policy push information for the enterprise based on the policy matching results, combined with the policy content vector and the policy rule tag vector.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the method of any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 7.

Images