Retrieval enhancement generation method, system and device, computer equipment and storage medium

Abstract

The invention relates to a retrieval enhancement generation method, system and device, computer equipment and a storage medium. The method comprises the steps of obtaining user permission information in response to a received retrieval request; selecting multiple pieces of first data from the multiple pieces of candidate data according to the user permission information and the data permission of each piece of candidate data in the multiple pieces of candidate data; according to the similarity between the question information and each piece of first data and the reference information of each piece of first data, determining the score of each piece of first data; determining context information related to the question information according to the score of each piece of first data; and generating answer information according to the context information and the question information. According to the method, the answer information is ensured to be generated on the basis of the multiple pieces of first data with the access authority of the user account, the data security can be improved, and the accuracy of the answer information can be improved by taking the data quality of the first data as reference, so that the answer security, accuracy and reliability are improved.

Description

Technical Field

[0001] This disclosure relates to the field of data retrieval, specifically to a retrieval enhancement generation method, system, apparatus, computer equipment, and storage medium. Background Technology

[0002] Retrieval-Augmented Generation (RAG) technology typically uses vector databases to perform text retrieval and combines the search results with the original question input by the user, generating an answer through a Large Language Model (LLM). However, in these technologies, different users can retrieve the same data, leading to poor data security during retrieval and low quality of the retrieved data, resulting in low accuracy of the search results. Summary of the Invention

[0003] To overcome the problems existing in related technologies, this disclosure provides a search enhancement generation method, system, apparatus, computer equipment, and storage medium.

[0004] According to a first aspect of the present disclosure, a retrieval enhancement generation method is provided, comprising: In response to receiving a search request, user permission information is obtained. The search request carries question information, and the user permission information is used to indicate the user permissions of the user account that triggered the search request. Based on the user permission information and the data permissions of each candidate data in the plurality of candidate data, a plurality of first data are selected from the plurality of candidate data, wherein the user permission is not lower than the data permission of each first data; Based on the similarity between the question information and each of the first data points, and the reference information for each of the first data points, a score for each of the first data points is determined, wherein the reference information is used to indicate the data quality of the first data points. Based on the score of each of the first data points, context information related to the problem information is determined, wherein the context information includes at least one first data point whose score is not lower than the scores of other first data points among the plurality of first data points; Based on the context information and the question information, an answer is generated.

[0005] In some embodiments, the method further includes: Obtain permission attributes based on the user identity information associated with the user account; The user permission information is generated based on the permission attributes. The permission attributes include at least one of the following: Role permissions; Data access level; Departmental affiliation; Business scenario permissions.

[0006] In some embodiments, determining the score for each first data point based on the similarity between the question information and each of the first data points, and the reference information for each of the first data points, includes: For each first data point, the similarity and reference information corresponding to the first data point are weighted according to a preset coefficient to obtain a score for the first data point. The preset coefficient includes the coefficient corresponding to the similarity and the coefficient corresponding to the reference information.

[0007] In some embodiments, the reference information includes at least one of the following: Credibility weight; Timeliness weight; Content weight; Importance weight; Business relevance weight.

[0008] In some embodiments, the reference information includes the credibility weight, timeliness weight, and importance weight; the step of weighting the similarity corresponding to the first data and the reference information according to a preset coefficient to obtain a score for the first data includes... For each first data point, the score for that first data point is determined using the following formula:

[0009] in, This indicates the rating. , , and This represents the preset coefficient. Indicates the similarity, This represents the credibility weight. Indicates the weight of timeliness. This indicates the importance weight.

[0010] In some embodiments, the method further includes: The response information is subjected to permission checks. If sensitive information is found in the response information, a target operation is performed. The target operation includes at least one of the following: Replace the sensitive information; The sensitive information is hidden; Delete the sensitive information; Regenerate the answer information.

[0011] According to a second aspect of the present disclosure, a retrieval enhancement generation system is provided for implementing the method as described in the first aspect, the retrieval enhancement generation system comprising: The permission identification module is used to obtain user permission information in response to receiving a search request. The search request carries question information, and the user permission information is used to indicate the user permissions of the user account that triggered the search request. The permission filtering module is used to select multiple first data from the multiple candidate data based on the user permission information and the data permission of each candidate data in the multiple candidate data, wherein the user permission is not lower than the data permission of each first data; The scoring module is used to determine the score of each first data based on the similarity between the question information and each first data, and the reference information of each first data, wherein the reference information is used to indicate the data quality of the first data; A context determination module is used to determine context information related to the problem information based on the score of each of the first data points, wherein the context information includes at least one first data point whose score is not lower than the scores of other first data points among the plurality of first data points; The generation module is used to generate answer information based on the context information and the question information.

[0012] In some embodiments, the retrieval enhancement generation system further includes: The permission detection module is used to perform permission detection on the answer information. In response to the presence of sensitive information in the answer information, it performs a target operation, which includes at least one of the following: Replace the sensitive information; The sensitive information is hidden; Delete the sensitive information; Regenerate the answer information.

[0013] According to a third aspect of the present disclosure, a search enhancement generation apparatus is provided, the search enhancement generation apparatus comprising: The permission identification unit is configured to obtain user permission information in response to receiving a retrieval request, wherein the retrieval request carries question information and the user permission information is used to indicate the user permissions of the user account that triggered the retrieval request. The permission filtering unit is configured to select a plurality of first data from the plurality of candidate data based on the user permission information and the data permission of each candidate data in the plurality of candidate data, wherein the user permission is not lower than the data permission of each first data. The scoring unit is configured to determine a score for each of the first data points based on the similarity between the question information and each of the first data points, and reference information for each of the first data points, wherein the reference information is used to indicate the data quality of the first data points. The context determination unit is configured to determine context information related to the problem information based on the score of each of the first data, wherein the context information includes at least one first data whose score is not lower than the scores of other first data among the plurality of first data; The generation unit is configured to generate answer information based on the context information and the question information.

[0014] According to a fourth aspect of the present disclosure, 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 method described in the first aspect.

[0015] According to a fifth aspect of the present disclosure, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described in the first aspect.

[0016] The technical solutions provided by the embodiments of this disclosure may include the following beneficial effects: The method provided in this disclosure, upon receiving a search request, determines the user permissions of the user account that triggered the search request. Based on these user permissions, it first filters multiple candidate data by permission, selecting multiple first data that can be accessed by the user account. Then, it determines the score of each first data by referring to the data quality of each first data, and generates answer information based on the score of each first data. This ensures that the answer information is generated based on multiple first data that the user account has access to, thereby improving data security. Furthermore, by using the data quality of the first data as a reference, it can improve the accuracy of the answer information, thus improving the security, accuracy, and reliability of the answer.

[0017] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0018] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0019] Figure 1 This is a flowchart illustrating a retrieval enhancement generation method according to an exemplary embodiment.

[0020] Figure 2This is a flowchart illustrating another retrieval enhancement generation method according to an exemplary embodiment.

[0021] Figure 3 This is a schematic diagram illustrating a retrieval enhancement generation system according to an exemplary embodiment.

[0022] Figure 4 This is a block diagram illustrating a retrieval enhancement generation apparatus according to an exemplary embodiment.

[0023] Figure 5 This is a block diagram of a computer device according to an exemplary embodiment. Detailed Implementation

[0024] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numerals in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Rather, they are merely examples of apparatuses and methods consistent with some aspects of the invention as detailed in the appended claims.

[0025] In related technologies, RAG lacks access control mechanisms. For example, the RAG framework does not implement access control for search content, allowing different users to retrieve the same text vectors, potentially leading to unauthorized access to sensitive internal enterprise documents. Search ranking and answer quality are unstable. For instance, RAG uses vector similarity as the sole indicator, resulting in insufficient retrieval probability for important documents and affecting answer accuracy. RAG cannot adapt to multi-source heterogeneous knowledge bases. In complex enterprise environments, data comes from various systems, such as emails, Enterprise Resource Planning (ERP), Office Automation (OA), knowledge bases, and databases. The importance and reliability of different data vary, making it impossible to measure the data quality of different data sources during retrieval.

[0026] The search enhancement generation method provided in this embodiment is executed by a computer device, which can specifically be a smart device such as a mobile phone, tablet computer, or laptop. In addition, the computer device is also equipped with various hardware resources and an energy storage device that provides power for the operation of these hardware resources.

[0027] Figure 1 This is a flowchart illustrating a retrieval enhancement generation method according to an exemplary embodiment, executed by a computer device. See also... Figure 1 The method includes the following steps: Step S101: In response to receiving a search request, obtain user permission information. The search request carries question information, and the user permission information is used to indicate the user permissions of the user account that triggered the search request.

[0028] In some embodiments, the search request is triggered by a user account and carries question information, which is content entered by the user.

[0029] Optionally, the search request may carry a user account. The computer device stores the correspondence between user accounts and user permission information. After receiving the search request, the computer device obtains the user permission information based on the user account.

[0030] Optionally, the search request can directly carry user permission information. For example, if a user searches on a local device, and the local device stores the user permission information corresponding to that user account, then that user permission information can be carried in the generated search request. Here, the local device refers to the device used by the user to perform the search.

[0031] Optionally, the search request may also carry a device identifier, which is a device of the local device.

[0032] Optionally, the retrieval request carries user identity information, and the computer device generates user permission information based on the user identity information.

[0033] Optionally, the retrieval request may also carry other information, which is not limited in this embodiment.

[0034] In some embodiments, user permission information is used to indicate the user permissions of the user account that triggered the retrieval request, which refers to whether the user account has the right to access data.

[0035] Optionally, user permission information can be represented as a user permission vector, for example, represented as... .

[0036] Optionally, user permission information can be represented as Level 1 permission, Level 2 permission, Level 3 permission, etc. The higher the user's permission, the higher the corresponding permission level number; for example, Level 2 permission is no lower than Level 1 permission. Alternatively, the higher the user's permission, the lower the corresponding permission level number; for example, Level 1 permission is no lower than Level 2 permission.

[0037] Alternatively, user permission information may be represented in other forms, and this disclosure does not limit this.

[0038] In some embodiments, permission attributes are obtained based on user identity information associated with a user account; user permission information is generated based on the permission attributes. The permission attributes include at least one of the following: role permissions; data access level; department attribute; business scenario permissions.

[0039] Optionally, role-based permissions refer to the permissions a user has based on their role within the enterprise. Different roles have different permissions, such as administrator permissions, manager permissions, and visitor permissions. Data access level refers to the confidentiality level of the data itself; for example, data access levels can include public, internal, confidential, and top secret. Departmental affiliation refers to the permissions a user has based on their department; for example, department 1's permissions are no less than department 2's permissions, so if a user belongs to department 1, they can access more data. Business scenario permissions refer to a user's permission to access data related to the business they are associated with. For example, if a user is responsible for business 1 but not business 2, they can access data related to business 1 but not data related to business 2.

[0040] Optionally, the permission attributes may also include other information, which is not limited in this embodiment.

[0041] It's worth noting that the process of generating user permission information can occur either before or after the computer device receives the retrieval request. Optionally, if the information is generated before receiving the retrieval request, the computer device can store the generated user permission information and the corresponding user account, making it convenient to retrieve the relevant permission information after receiving the retrieval request.

[0042] Another point worth noting is that if a user's identity information changes, the user's permission information needs to be updated promptly based on the changed user identity information. For example, a user may change their responsibility from Business 1 to Business 2, or their department may change from Department 1 to Department 2.

[0043] Step S102: Select multiple first data from multiple candidate data based on user permission information and the data permissions of each candidate data in multiple candidate data, wherein the user permission is not lower than the data permission of each first data.

[0044] In some embodiments, each candidate data has a corresponding data permission. For example, a corresponding data permission label can be added to each candidate data. The data permission label is used to indicate the level of data permission of the corresponding candidate data. For example, the data permission label includes one of L1-L5, where L1 represents the highest data permission and L5 represents the lowest data permission, or L1 represents the lowest data permission and L5 represents the highest data permission.

[0045] Optionally, candidate data can be documents, files, text, images, or other forms of data.

[0046] In some embodiments, data permissions and user permissions of candidate data can be directly compared. For example, user permission information is represented as follows: Data permissions are represented as direct comparison and The size, if Greater than If the user's permissions are not lower than the data permissions, then the candidate data will be used as the first data.

[0047] Optionally, for multiple candidate data, based on the user permission vector The process involves filtering out candidate data that the user is not authorized to access, leaving the remaining candidate data as the first set of data. For example, for each candidate data, the following formula can be used to determine whether to filter it out:

[0048] in, Represents candidate data, Indicates user permission information. Indicates data permissions. Indicates the first data, that is, if This indicates that candidate data will be used. As the first data .

[0049] Alternatively, data permissions may also be referred to as permission levels or other names.

[0050] In some embodiments, when it is necessary to search for data from an enterprise's internal database, a device identifier can be used as a device permission. Before filtering data based on user permissions, it can be determined whether the computer device indicated by the device identifier is a trusted device. If it is, the subsequent search process is executed; otherwise, the search is terminated to further improve data security.

[0051] In this step, candidate data that the user does not have permission to access is filtered out directly based on user permission information and data permissions. The remaining multiple first data are used as data that can be further searched, which can ensure that the data that the user has permission to access is processed in the subsequent processing.

[0052] In some embodiments, for candidate data, the computer device may pre-store semantic information, data permissions, and reference information for the candidate data. The semantic information represents the semantics of the candidate data, and the reference information indicates the data quality of the candidate data.

[0053] Optionally, the reference information includes at least one of the following: credibility weight; timeliness weight; content weight; importance weight; and business relevance weight. The credibility weight indicates the credibility of the candidate data's source; the timeliness weight indicates the timeliness of the candidate data, such as the timeliness weight of candidate data with a more recent date being no less than that of candidate data with a more distant date; the content weight indicates the content of the candidate data, such as a larger content area corresponding to a larger content weight, and a smaller content area corresponding to a smaller content weight; the importance weight indicates the degree of importance of the candidate data; and the business relevance weight indicates the degree of relevance between the candidate data and the business.

[0054] Optionally, if the credibility weight, timeliness weight, content weight, importance weight, and business relevance weight of the candidate data are relatively large, it indicates that the candidate data has high quality; conversely, if these weights are relatively small, it indicates that the candidate data has low quality. In one example, the credibility weight, timeliness weight, content weight, importance weight, and business relevance weight can be weighted to obtain a weighted result, which represents the quality of the candidate data.

[0055] Optionally, the computer device stores the semantic information, data permissions, and reference information of the candidate data in the following form: Di={Vi, Pi, Wi}, where Di represents the candidate data, Vi represents the semantic information of the candidate data, Pi represents the data permissions of the candidate data, and Wi represents the reference information.

[0056] Alternatively, semantic information can also be called semantic vector, Pi can also be called weight requirement vector, and Wi can also be called multidimensional weight set.

[0057] In this embodiment, the credibility, timeliness, content, importance, and business relevance of candidate data are quantified into weights, so that the data quality of candidate data can be measured during retrieval by the credibility weight, timeliness weight, content weight, importance weight, and business relevance weight.

[0058] Step S103: Determine the score of each first data point based on the similarity between the problem information and each first data point, and the reference information of each first data point. The reference information is used to indicate the data quality of the first data points.

[0059] In some embodiments, the computer device determines the similarity between the problem information and each piece of first data. For example, the similarity is determined using the following formula:

[0060] in, Indicates similarity. Indicates problem information. This represents the semantic information of the first data.

[0061] In some embodiments, for each first data point, the similarity and reference information corresponding to the first data point are weighted according to preset coefficients to obtain a score for the first data point. The preset coefficients include the coefficients corresponding to the similarity and the coefficients corresponding to the reference information.

[0062] Optionally, the reference information includes credibility weight, timeliness weight, and importance weight. For each first data point, the score of the first data point is determined using the following formula:

[0063] in, Indicates rating, , , and Indicates the preset coefficient. Indicates similarity. Indicates the credibility weight. Indicates the weight of timeliness. This represents the importance weight. Among them, The coefficient corresponding to the similarity. The coefficient corresponding to the credibility weight. This is the coefficient corresponding to the timeliness weight. These are the coefficients corresponding to the importance weights.

[0064] It is worth noting that the reference information may include one or more weights. The above embodiment takes the reference information including credibility weight, timeliness weight and importance weight as an example. In other embodiments, the reference information may also include credibility weight and timeliness weight, or the reference information may also include content weight, importance weight and business relevance weight. This disclosure does not limit which items are included in the reference information.

[0065] Optionally, the computer device can employ a weighted model to determine the score, that is, inputting the similarity and reference information into the weighted model, which then outputs the score. The weights corresponding to each piece of information can be model parameters of the weighted model, i.e., preset coefficients are obtained by training the weighted model. This embodiment of the disclosure does not limit the training process of the weighted model.

[0066] Step S104: Based on the score of each first data, determine the context information related to the problem information, wherein the context information includes at least one first data whose score is not lower than the scores of other first data among the multiple first data.

[0067] Among them, other first data in the plurality of first data refers to the first data other than at least one first data included in the context information.

[0068] In some embodiments, based on the score of each first data point, multiple first data points are sorted in descending order, and the top K first data points are selected from the sorted multiple first data points and used as context information. Here, K is an integer greater than or equal to 1.

[0069] In some embodiments, based on the score of the first data, a first data point with a score greater than a preset score is selected from a plurality of first data points, and the selected first data point is used as context information. The preset score is a pre-set score.

[0070] Optionally, if there is no first data with a score greater than the preset score among the multiple first data, the K first data with the highest scores can be selected.

[0071] Step S105: Generate answer information based on context information and question information.

[0072] In some embodiments, contextual information and question information are input into a large language model, which outputs answer information. The large language model is a pre-trained model, and this disclosure does not limit the training process of the large language model.

[0073] The method provided in this disclosure, upon receiving a search request, determines the user permissions of the user account that triggered the search request. Based on these user permissions, it first filters multiple candidate data by permission, selecting multiple first data that can be accessed by the user account. Then, it determines the score of each first data by referring to the data quality of each first data, and generates answer information based on the score of each first data. This ensures that the answer information is generated based on multiple first data that the user account has access to, thereby improving data security. Furthermore, by using the data quality of the first data as a reference, it can improve the accuracy of the answer information, thus improving the security, accuracy, and reliability of the answer.

[0074] Figure 2 This is a flowchart illustrating another retrieval enhancement generation method according to an exemplary embodiment, executed by a computer device, see [link to flowchart]. Figure 2 The method includes the following steps: Step S201: In response to receiving a retrieval request, obtain user permission information.

[0075] Step S202: Select multiple first data from multiple candidate data based on user permission information and the data permissions of each candidate data among multiple candidate data.

[0076] Step S203: Calculate the similarity between the problem information and each first data point.

[0077] Step S204: Determine the score for each first data point based on the similarity between the question information and each first data point, as well as the reference information for each first data point.

[0078] Step S205: Determine the context information related to the problem information based on the score of each first data point.

[0079] Step S206: Generate answer information based on context information and question information.

[0080] In some embodiments, the implementation of steps S201-S206 is the same as that of steps S101-S105 described above, and will not be repeated here.

[0081] Step S207: Perform permission checks on the response information. If sensitive information is found in the response information, execute the target operation.

[0082] In some embodiments, sensitive information may include personal sensitive information, such as ID card numbers and fingerprint information, enterprise sensitive information, such as core enterprise technologies and financial data, and national security sensitive information, such as key technologies and classified documents.

[0083] Alternatively, to avoid leaking sensitive information, it can be assumed that all sensitive information is beyond the user's permissions, and sensitive information cannot be directly output when outputting the response information.

[0084] In some embodiments, the target operation includes at least one of the following: Replace sensitive information; Hiding sensitive information; Delete sensitive information; Regenerate the answer information.

[0085] Optionally, replacing sensitive information may include replacing the sensitive information with information that is most similar to the sensitive information from a plurality of first data.

[0086] Optionally, hiding sensitive information may include: masking sensitive information, or blurring sensitive information, that is, making it impossible for users to see the specific content of the sensitive information.

[0087] Optionally, regenerating the response information may involve either returning to step S202 and re-executing the complete generation process, or returning to step S206 and regenerating the response information.

[0088] In this embodiment, to avoid missing sensitive information that exceeds the user's permissions during permission filtering, a second check is performed before outputting the answer information.

[0089] In some implementation sets, after performing the target operation, the final answer information is obtained and sent to the device that sent the search request so that the user can obtain the search results.

[0090] In some embodiments, after step S206, the computer device may directly output the answer information and send it to the device that sent the search request so that the user can obtain the search results, i.e., step S207 may not be executed.

[0091] In this embodiment of the disclosure, while ensuring that the answer information is generated based on multiple first data that the user account has access to, the answer information is further detected before being output to ensure that there is no sensitive information in the final output answer information, thereby further improving data security.

[0092] Figure 3 This is a schematic diagram illustrating a retrieval enhancement generation system according to an exemplary embodiment. The retrieval enhancement generation system is used to implement the retrieval enhancement generation method shown in the foregoing embodiments. The retrieval enhancement generation system includes: The permission identification module 301 is used to respond to a received search request, obtain user permission information, and the search request carries question information. The user permission information is used to indicate the user permissions of the user account that triggered the search request. The permission filtering module 302 is used to select multiple first data from multiple candidate data based on user permission information and the data permission of each candidate data in multiple candidate data, wherein the user permission is not lower than the data permission of each first data; The scoring module 303 is used to determine the score of each first data based on the similarity between the question information and each first data, and the reference information of each first data. The reference information is used to indicate the data quality of the first data. The context determination module 304 is used to determine context information related to the problem information based on the score of each first data, wherein the context information includes at least one first data whose score is not lower than the scores of other first data among a plurality of first data; The generation module 305 is used to generate answer information based on context information and question information.

[0093] In some embodiments, the retrieval enhancement generation system further includes: The permission detection module 306 is used to perform permission detection on the answer information. In response to the presence of sensitive information in the answer information, it executes a target operation, which includes at least one of the following: Replace sensitive information; Hiding sensitive information; Delete sensitive information; Regenerate the answer information.

[0094] In some embodiments, the operations performed by each module in the retrieval enhancement generation system can be found in the steps and optional implementations of other embodiments described before or after this embodiment, as well as other related parts of the specification, which will not be repeated here.

[0095] Figure 4 This is a block diagram illustrating a search enhancement generation apparatus according to an exemplary embodiment, the search enhancement generation apparatus being configured in a computer device, such as... Figure 4 As shown, the device includes: The permission identification unit 401 is configured to obtain user permission information in response to receiving a retrieval request. The retrieval request carries question information, and the user permission information is used to indicate the user permissions of the user account that triggered the retrieval request. The permission filtering unit 402 is configured to select multiple first data from multiple candidate data based on user permission information and the data permission of each candidate data in multiple candidate data, wherein the user permission is not lower than the data permission of each first data; Scoring unit 403 is configured to determine a score for each first data point based on the similarity between the question information and each first data point, and reference information for each first data point, wherein the reference information is used to indicate the data quality of the first data point. The context determination unit 404 is configured to determine context information related to the problem information based on the score of each first data, wherein the context information includes at least one first data whose score is not lower than the scores of other first data among a plurality of first data; The generation unit 405 is configured to generate answer information based on context information and question information.

[0096] In some embodiments, the apparatus further includes: The permission generation unit is configured to obtain permission attributes based on user identity information associated with the user account; and generate user permission information based on the permission attributes. The permission attributes include at least one of the following: Role permissions; Data access level; Departmental affiliation; Business scenario permissions.

[0097] In some embodiments, the scoring unit 403 is configured to, for each first data point, perform weighted processing on the similarity and reference information corresponding to the first data point according to preset coefficients to obtain a score for the first data point. The preset coefficients include the coefficients corresponding to the similarity and the coefficients corresponding to the reference information.

[0098] In some embodiments, the reference information includes at least one of the following: Credibility weight; Timeliness weight; Content weight; Importance weight; Business relevance weight.

[0099] In some embodiments, the apparatus further includes: The permission detection unit is configured to perform permission detection on the response information, and in response to the presence of sensitive information in the response information, execute a target operation, which includes at least one of the following: Replace sensitive information; Hiding sensitive information; Delete sensitive information; Regenerate the answer information.

[0100] Each unit in the aforementioned search enhancement generation device can be implemented entirely or partially through software, hardware, or a combination thereof. These units can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the operations corresponding to each unit.

[0101] In one exemplary embodiment, a computer device is provided, including a processor and a memory, the memory storing a computer program, the processor executing the computer program to implement the steps of any of the above-described search enhancement generation methods.

[0102] In one exemplary embodiment, a computer-readable storage medium is provided, on which a computer program is stored, which, when executed by a processor, implements the steps of any of the above-described search enhancement generation methods. The computer-readable storage medium may be a ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, or optical data storage device, etc.

[0103] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps of any of the above-described search enhancement generation methods.

[0104] refer to Figure 5The following description serves as a structural block diagram of the computer device disclosed herein. The computer device includes a computing unit 501, which can perform various appropriate actions and processes based on a computer program stored in a read-only memory (ROM) 502 or a computer program loaded from a storage unit 508 into a random access memory (RAM) 503. The RAM 503 may also store various programs and data required for the operation of the computer device 500. The computing unit 501, ROM 502, and RAM 503 are interconnected via a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.

[0105] Multiple components in computer device 500 are connected to I / O interface 505, including: input unit 506, output unit 507, storage unit 508, and communication unit 509. Input unit 506 can be any type of device capable of inputting information into computer device 500. Input unit 506 can receive input numerical or character information and generate key signal inputs related to user settings and / or function control of computer device 500, and may include, but is not limited to, a mouse, keyboard, touchscreen, trackpad, trackball, joystick, microphone, and / or remote control. Output unit 507 can be any type of device capable of presenting information, and may include, but is not limited to, a monitor, speaker, video / audio output terminal, vibrator, and / or printer. Storage unit 508 may include, but is not limited to, hard disk and optical disk. Communication unit 509 allows computer device 500 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks, and may include, but is not limited to, modems, network cards, infrared communication devices, wireless communication transceivers, and / or chipsets, such as Bluetooth™ devices, WiFi devices, WiMax devices, cellular communication devices, and / or the like.

[0106] The computing unit 501 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 501 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 501 performs the various methods and processes described above, such as the retrieval enhancement generation method. For example, in some embodiments, the retrieval enhancement generation method may be implemented as a computer software program tangibly contained in a machine-readable medium, such as storage unit 508. In some embodiments, part or all of the computer program may be loaded and / or installed on the computer device 500 via ROM 502 and / or communication unit 509. When the computer program is loaded into RAM 503 and executed by the computing unit 501, one or more steps of the retrieval enhancement generation method described above may be performed. Alternatively, in other embodiments, the computing unit 501 may be configured to perform the retrieval enhancement generation method by any other suitable means (e.g., by means of firmware).

[0107] The computer device 500 may be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the above-described retrieval enhancement generation method.

[0108] Those skilled in the art will understand that all or part of the steps in the above methods can be implemented by a program instructing related hardware, and the program can be stored in a computer-readable storage medium, such as a read-only memory, a disk, or an optical disk. Optionally, all or part of the steps in the above embodiments can also be implemented using one or more integrated circuits. Accordingly, each module / unit in the above embodiments can be implemented in hardware or as a software functional module. This invention is not limited to any particular combination of hardware and software.

[0109] Other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of the invention are indicated by the following claims.

[0110] It should be understood that the present invention is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.

Claims

1. A retrieval enhancement generation method, characterized in that, The method includes: In response to receiving a search request, user permission information is obtained. The search request carries question information, and the user permission information is used to indicate the user permissions of the user account that triggered the search request. Based on the user permission information and the data permissions of each candidate data in the plurality of candidate data, a plurality of first data are selected from the plurality of candidate data, wherein the user permission is not lower than the data permission of each first data; Based on the similarity between the question information and each of the first data points, and the reference information for each of the first data points, a score for each of the first data points is determined, wherein the reference information is used to indicate the data quality of the first data points. Based on the score of each of the first data points, context information related to the problem information is determined, wherein the context information includes at least one first data point whose score is not lower than the scores of other first data points among the plurality of first data points; Based on the context information and the question information, an answer is generated.

2. The retrieval enhancement generation method according to claim 1, characterized in that, The method further includes: Obtain permission attributes based on the user identity information associated with the user account; The user permission information is generated based on the permission attributes. The permission attributes include at least one of the following: Role permissions; Data access level; Departmental affiliation; Business scenario permissions.

3. The retrieval enhancement generation method according to claim 1, characterized in that, The step of determining the score for each first data point based on the similarity between the question information and each first data point, and the reference information for each first data point, includes: For each first data point, the similarity and reference information corresponding to the first data point are weighted according to a preset coefficient to obtain a score for the first data point. The preset coefficient includes the coefficient corresponding to the similarity and the coefficient corresponding to the reference information.

4. The retrieval enhancement generation method as described in claim 3, characterized in that, The reference information includes at least one of the following: Credibility weight; Timeliness weight; Content weight; Importance weight; Business relevance weight.

5. The retrieval enhancement generation method according to claim 1, characterized in that, The method further includes: The response information is subjected to permission checks. If sensitive information is found in the response information, a target operation is performed. The target operation includes at least one of the following: Replace the sensitive information; The sensitive information is hidden; Delete the sensitive information; Regenerate the answer information.

6. A retrieval enhancement generation system, said retrieval enhancement generation system being used to implement the retrieval enhancement generation method as described in any one of claims 1 to 5, characterized in that, The retrieval enhancement generation system includes: The permission identification module is used to obtain user permission information in response to receiving a search request. The search request carries question information, and the user permission information is used to indicate the user permissions of the user account that triggered the search request. The permission filtering module is used to select multiple first data from the multiple candidate data based on the user permission information and the data permission of each candidate data in the multiple candidate data, wherein the user permission is not lower than the data permission of each first data; The scoring module is used to determine the score of each first data based on the similarity between the question information and each first data, and the reference information of each first data, wherein the reference information is used to indicate the data quality of the first data; A context determination module is used to determine context information related to the problem information based on the score of each of the first data points, wherein the context information includes at least one first data point whose score is not lower than the scores of other first data points among the plurality of first data points; The generation module is used to generate answer information based on the context information and the question information.

7. The retrieval enhancement generation system as described in claim 6, characterized in that, The retrieval enhancement generation system also includes: The permission detection module is used to perform permission detection on the answer information. In response to the presence of sensitive information in the answer information, it performs a target operation, which includes at least one of the following: Replace the sensitive information; The sensitive information is hidden; Delete the sensitive information; Regenerate the answer information.

8. A retrieval enhancement generation device, characterized in that, The retrieval enhancement generation device includes: The permission identification unit is configured to obtain user permission information in response to receiving a retrieval request, wherein the retrieval request carries question information and the user permission information is used to indicate the user permissions of the user account that triggered the retrieval request. The permission filtering unit is configured to select a plurality of first data from the plurality of candidate data based on the user permission information and the data permission of each candidate data in the plurality of candidate data, wherein the user permission is not lower than the data permission of each first data. The scoring unit is configured to determine a score for each of the first data points based on the similarity between the question information and each of the first data points, and reference information for each of the first data points, wherein the reference information is used to indicate the data quality of the first data points. The context determination unit is configured to determine context information related to the problem information based on the score of each of the first data, wherein the context information includes at least one first data whose score is not lower than the scores of other first data among the plurality of first data; The generation unit is configured to generate answer information based on the context information and the question information.

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 steps of the method according to any one of claims 1 to 5.

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 steps of the method according to any one of claims 1 to 5.

Images