Knowledge processing method, system, storage medium and program product

Abstract

The present disclosure provides a knowledge processing method, system, storage medium and program product, and relates to the field of artificial intelligence. The method comprises: constructing a user knowledge graph according to historical operations of each user on knowledge in a public knowledge base; processing the user knowledge graph to obtain a user feature vector and a knowledge feature vector; obtaining an embedding vector matrix of a plurality of candidate knowledge based on the user feature vector and the knowledge feature vector; and performing knowledge recommendation for a target user based on historical operations of the target user on knowledge and the embedding vector matrix of the plurality of candidate knowledge.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the field of artificial intelligence, and particularly relates to a knowledge processing method, system, storage medium and program product. BACKGROUND

[0002] With the rapid progress of artificial intelligence technology, large language models have become a key technology cornerstone in multiple fields, especially in intelligent question answering systems, they are changing the way of information acquisition and processing with unprecedented capabilities. Especially combined with advanced retrieval enhancement generation framework, in many application scenarios, government intelligent question answering is particularly prominent, how to accurately recommend government knowledge becomes particularly important. SUMMARY

[0003] One of the technical problems to be solved by the present disclosure is to provide a knowledge processing method, system, storage medium and program product, which can improve the accuracy of knowledge recommendation.

[0004] According to an aspect of the present disclosure, a knowledge processing method is provided, comprising: constructing a user knowledge graph according to the historical operation of each user on the knowledge in the public knowledge base; processing the user knowledge graph to obtain a user feature vector and a knowledge feature vector; obtaining an embedding vector matrix of a plurality of candidate knowledge based on the user feature vector and the knowledge feature vector; and recommending knowledge to a target user based on the historical operation of the target user on the knowledge and the embedding vector matrix of the plurality of candidate knowledge.

[0005] In some embodiments, in response to at least one of an editing operation and an updating operation of the target user on the recommended knowledge, the public knowledge base is updated.

[0006] In some embodiments, recommending knowledge to the target user based on the historical operation of the target user on the knowledge and the embedding vector matrix of the plurality of candidate knowledge comprises: performing information enhancement on the historical operation of the target user on the knowledge to obtain a first embedding vector matrix corresponding to the target user; performing self-attention processing on the first embedding vector matrix to obtain a second embedding vector matrix; scoring each candidate knowledge in the plurality of candidate knowledge using the second embedding vector matrix and the embedding vector matrix of the plurality of candidate knowledge; and recommending knowledge to the target user based on the score of each candidate knowledge.

[0007] In some embodiments, the information enhancement on the historical operation of the target user on the knowledge to obtain the first embedding vector matrix corresponding to the target user comprises: averaging a plurality of contexts of each session in the historical operation of the target user on the knowledge to obtain a first fusion embedding vector representation; performing time-varying information fusion on each session in the historical operation of the target user on the knowledge to obtain a second fusion embedding vector representation; performing weighted calculation on the first fusion embedding vector representation and the second fusion embedding vector representation to obtain a third fusion embedding vector representation corresponding to each session; performing a concatenation operation on the third fusion embedding vector representation corresponding to each session and the time of each session, and processing by using a feedforward neural network to obtain an embedding vector matrix corresponding to each session; and obtaining the first embedding vector matrix corresponding to the target user according to the embedding vector matrix corresponding to each session.

[0008] In some embodiments, the self-attention processing on the first embedding vector matrix to obtain the second embedding vector matrix comprises: performing normalized exponential processing on the first embedding vector matrix by using a query parameter matrix, a key parameter matrix and a value parameter matrix of a self-attention algorithm to obtain the second embedding vector matrix.

[0009] In some embodiments, the processing on the user knowledge graph to obtain the user feature vector and the knowledge feature vector comprises: processing the user knowledge graph by using a bipartite graph to obtain the user feature vector and the knowledge feature vector.

[0010] In some embodiments, the updating of the public knowledge base further comprises: in response to the recommended knowledge being passed by the audit, forming public knowledge; and updating the public knowledge base by using the public knowledge.

[0011] In some embodiments, the forming of the public knowledge in response to the recommended knowledge being passed by the audit comprises: classifying the recommended knowledge; and forming the public knowledge after the recommended knowledge is passed by the audit according to the category.

[0012] In some embodiments, the knowledge processing method further comprises: distributed collection of original knowledge; data conversion and preprocessing of the original knowledge; knowledge extraction of the original knowledge after the data conversion and preprocessing around a business domain to form a structured knowledge representation; and collation of the structured knowledge representation to obtain the public knowledge base.

[0013] According to another aspect of the present disclosure, there is also provided a knowledge processing system, comprising a knowledge recommendation module, the knowledge recommendation module comprising: a construction unit configured to construct a user knowledge graph according to historical operations of each user on knowledge in a public knowledge base; a processing unit configured to process the user knowledge graph to obtain a user feature vector and a knowledge feature vector; a candidate unit configured to obtain an embedding vector matrix of a plurality of candidate knowledge based on the user feature vector and the knowledge feature vector; and a recommendation unit configured to perform knowledge recommendation on a target user based on historical operations of the target user on knowledge, using the user feature vector and the knowledge feature vector.

[0014] In some embodiments, the knowledge processing system further comprises a sharing module configured to update the public knowledge base in response to at least one of an editing operation and an updating operation performed by the target user on the recommended knowledge.

[0015] In some embodiments, the knowledge processing system further comprises a collection module configured to collect raw knowledge in a distributed manner, perform data conversion and preprocessing on the raw knowledge, perform knowledge extraction on the raw knowledge that has been data-converted and preprocessed around a business domain to form a structured knowledge representation, and organize the structured knowledge representation to obtain the public knowledge base.

[0016] According to another aspect of the present disclosure, there is also provided a knowledge processing system, comprising: a memory; and a processor coupled to the memory, the processor being configured to perform the knowledge processing method as described above based on instructions stored in the memory.

[0017] According to another aspect of the present disclosure, there is also provided a computer-readable storage medium having computer program instructions stored thereon, the instructions being executed by a processor to implement the knowledge processing method as described above.

[0018] According to another aspect of the present disclosure, there is also provided a computer program product comprising a computer program or instructions, the computer program or instructions being executed by a processor to implement the knowledge processing method as described above.

[0019] Other features and advantages of the present disclosure will become apparent from the following detailed description of illustrative embodiments thereof, which proceeds with reference to the accompanying drawings. BRIEF DESCRIPTION OF DRAWINGS

[0020] The accompanying drawings, which form a part of the specification, illustrate embodiments of the present disclosure and, together with the description, serve to explain the principles of the present disclosure.

[0021] The present disclosure can be more clearly understood with reference to the following detailed description in conjunction with the accompanying drawings, in which:

[0022] Figure 1 Flowchart of some embodiments of the knowledge processing method of the present disclosure;

[0023] Figure 2 schematic diagram of some embodiments of the knowledge processing method of the present disclosure;

[0024] Figure 3 schematic diagram of some embodiments of the knowledge processing method of the present disclosure;

[0025] Figure 4 schematic diagram of some embodiments of the knowledge processing method of the present disclosure;

[0026] Figure 5 schematic diagram of some embodiments of the knowledge recommendation module of the knowledge processing system of the present disclosure;

[0027] Figure 6 schematic diagram of some embodiments of the knowledge processing system of the present disclosure;

[0028] Figure 7 schematic diagram of some embodiments of the knowledge processing system of the present disclosure;

[0029] Figure 8 schematic diagram of some embodiments of the government affair knowledge management system of the present disclosure. DETAILED DESCRIPTION

[0030] Various exemplary embodiments of the present disclosure will now be described in detail with reference to the accompanying drawings. It should be noted that the relative arrangement of the components and steps set forth in these embodiments, numerical expressions, and numerical values do not limit the scope of the present disclosure unless specifically stated otherwise.

[0031] It should be understood, of course, that the various embodiments of the disclosure are merely examples of implementations and are not intended to limit the scope of the disclosure in any way.

[0032] The following description of at least one example embodiment is merely exemplary in nature and is in no way intended to limit the scope of the disclosure, its application, or uses.

[0033] Techniques, methods, and devices known to those of ordinary skill in the relevant art can not be discussed in detail herein. However, where appropriate, such techniques, methods, and devices can be considered part of the description of the present disclosure.

[0034] In all of the examples shown and discussed herein, any specific values should be interpreted as merely exemplary and not as a limitation. Thus, other examples of the exemplary embodiments can have different values.

[0035] It should be noted that like references and characters herein relate to like items throughout the figures, and once an item is defined in one figure, it need not be discussed further in subsequent figures.

[0036] To make the objectives, technical solutions, and advantages of this disclosure clearer, the following detailed description is provided in conjunction with specific embodiments and the accompanying drawings.

[0037] It should be noted that the collection, gathering, updating, analysis, processing, use, transmission, storage, and sharing of user personal information involved in this disclosed technical solution all comply with relevant laws and regulations, are used for legitimate purposes, and do not violate public order and good morals. Necessary measures are taken to prevent unauthorized access to user personal information data and to safeguard user personal information security, network security, and national security.

[0038] Figure 1 This is a flowchart illustrating some embodiments of the knowledge processing method disclosed herein, which includes steps S11-S14.

[0039] In step S11, a user knowledge graph is constructed based on each user's historical operations on knowledge in the public knowledge base.

[0040] This knowledge includes, for example, public knowledge such as government affairs knowledge. Government affairs knowledge mainly refers to question-and-answer pairs formed from structured information about business personnel, policy documents, and matters, including questions and answers.

[0041] In some embodiments, a user-knowledge graph is constructed based on the salesperson's historical actions on question-and-answer pairs. For example... Figure 2 As shown, Figure 2 This is a schematic diagram of some embodiments of the knowledge processing flow disclosed herein, whereby the user uses u i This indicates that the question-and-answer pair uses k n It means that the weight k i,n The value is 0 or 1, where 1 indicates that the user has clicked on the question-and-answer pair and 0 indicates that the user has not clicked on the question-and-answer pair. The constructed user-knowledge graph is represented by G.

[0042] In step S12, the user knowledge graph is processed to obtain user feature vectors and knowledge feature vectors.

[0043] In some embodiments, a bipartite graph is used to process the user knowledge graph to obtain user feature vectors and knowledge feature vectors.

[0044] Bipartite graphs can fully model user-knowledge interactions. Learning using bipartite graphs preserves second-order proximity, making its conditional probability close to its empirical distribution, thus obtaining implicit user features and knowledge features.

[0045] User characteristics include, for example, a user's job title, business attributes, and interactive behaviors.

[0046] At step S13, based on the user feature vector and the knowledge feature vector, an embedding vector matrix of multiple candidate knowledge is obtained.

[0047] In some embodiments, the user feature vector and the knowledge feature vector are recalled according to a predetermined number, and an embedding vector matrix of multiple candidate knowledge is obtained.

[0048] For example, based on the learned user feature vector and the knowledge feature vector, recall is performed, for example, the initial setting of the recall pool is 50. Those skilled in the art should understand that the setting of the recall pool here is only for example, and can be set according to actual conditions.

[0049] At step S14, based on the historical operation of the target user on the knowledge and the embedding vector matrix of the multiple candidate knowledge, knowledge recommendation is performed on the target user.

[0050] For example, the softmax algorithm is used to score the candidate knowledge, so as to recommend the knowledge to the target user according to the score size.

[0051] In the above embodiments, by constructing the user knowledge graph, a high-quality embedding vector matrix of multiple candidate knowledge is obtained, and then based on the historical operation of the target user on the knowledge and the embedding vector matrix of the multiple candidate knowledge, personalized knowledge recommendation is provided for the target user, so as to realize accurate knowledge recommendation, thereby improving service quality and efficiency.

[0052] In some embodiments of the present disclosure, step S14, based on the historical operation of the target user on the knowledge and the embedding vector matrix of the multiple candidate knowledge, knowledge recommendation is performed on the target user, as shown in Figure 3 , including steps S141-S144.

[0053] At step S141, the historical operation of the target user on the knowledge is information enhanced, and a first embedding vector matrix corresponding to the target user is obtained.

[0054] As shown in Figure 2 , the historical interaction of the target user u i is represented by a vector Subsequently, through a context enhancement layer, a first embedding vector matrix

[0055] This step fuses the periodicity and context granularity information of each session in the historical operation, which can reduce the influence of time fluctuation and timestamp division on the session embedding representation.

[0056] In some embodiments, the first fused embedding vector representation is obtained by averaging multiple contexts of each session in the target user's historical operations on knowledge; the second fused embedding vector representation is obtained by time-varying information fusion of each session in the target user's historical operations on knowledge; the first fused embedding vector representation and the second fused embedding vector representation are weighted to obtain the third fused embedding vector representation corresponding to each session; the third fused embedding vector representation corresponding to each session and the time of each session are concatenated and processed by a feedforward neural network to obtain the embedding vector matrix corresponding to each session; and the first embedding vector matrix corresponding to the target user is obtained based on the embedding vector matrix corresponding to each session.

[0057] like Figure 2 As shown, according to Obtain multiple operation records of the target user (u i c j k n , t j ), (u i c j k n , t j ) represents user u i At time t j For category c j Knowledge k n Perform the operation. For each session. The embedding vector representation is obtained by averaging multiple contexts. For each session, time-varying information is fused to obtain an embedding vector representation. ts represents the timestamp. The time-varying nature here includes the basic characteristics of time, including year, month, day, hour, minute, and second, as well as time period characteristics, including the number of hours in a day, the day of the week in a month, and the number of months in a year.

[0058] Each session is obtained through a balanced contextual and time-aware information enhancement paradigm. expression α is used to balance the weights of contextual and time-varying information. Applying the aforementioned information enhancement strategy, a third fused embedding vector representation can be obtained. Finally, aggregation is performed according to ts to obtain Where FFN stands for Feedforward Neural Network, and Concat represents the concatenation operation of the embedding vectors.

[0059] For target user u i A set of historical operations The embedding vector matrix E( hist )={a1,a2,...,a n} is used to represent this.

[0060] At step S142, the first embedding vector matrix is subjected to self-attention processing to obtain a second embedding vector matrix.

[0061] In this step, the self-attention mechanism can effectively capture the sequential dependency between the operation records in the user interaction. For example, the embedding expression after the self-attention layer is E' = {a'1, a'2,..., a'N} (hist) n}.

[0062] In some embodiments, the first embedding vector matrix is subjected to normalized exponential processing by using the query parameter matrix, the key parameter matrix and the value parameter matrix of the self-attention algorithm to obtain the second embedding vector matrix.

[0063] As shown in Figure 2 , the history represents the historical operations of the target user, which can be used to construct the query parameter matrix, the key parameter matrix and the value parameter matrix of the self-attention algorithm, and then the first embedding vector matrix is subjected to normalized exponential processing by using the query parameter matrix, the key parameter matrix and the value parameter matrix to obtain the second embedding vector matrix E' = {a'1, a'2,..., a'N} (hist) n}.

[0064] At step S143, each candidate knowledge in the plurality of candidate knowledge is scored by using the second embedding vector matrix and the embedding vector matrix of the plurality of candidate knowledge.

[0065] In some embodiments, the embedding vector matrix of the plurality of candidate knowledge can be represented as The candidate knowledge is scored by using the softmax algorithm to obtain wherein each represents the score value of the model for the knowledge ki, and the higher the value, the more likely the model considers that the user will perform an operation on this knowledge.

[0066] For example, the score value of the candidate knowledge is obtained by using the formula , wherein Sum(·) represents the weighted sum of the last dimension, and d is determined by parameter experiment. Those skilled in the art should understand that a plurality of algorithms can be used to score the subsequent knowledge, and the above formula is only used for example.

[0067] At step S144, the target user is recommended knowledge based on the score value of each candidate knowledge.

[0068] For example, the target user is recommended the candidate knowledge with the highest score value. For another example, the top-K (k highest score value) knowledge is recommended to the business staff in order according to the score value. ​​

[0069] In the above embodiment, the contextual enhanced conversation recommendation can improve the accuracy of knowledge recommendation.

[0070] In some embodiments of the present disclosure, the public knowledge base is updated in response to at least one of the editing operation and the updating operation on the recommended knowledge by the target user.

[0071] For example, as the business system develops, the information updates at a faster speed, and new knowledge emerges and deposits continuously. However, the traditional government knowledge management mechanism often cannot keep up with these changes in time, resulting in that the intelligent question and answer system driven by the large model cannot provide the latest information. In the embodiment, the user can edit or update the recommended knowledge, so that the knowledge in the public knowledge base is the latest knowledge, and the knowledge in the public knowledge base can be shared.

[0072] In some embodiments, the recommended knowledge is reviewed and passed to form public knowledge, and the public knowledge is used to update the public knowledge base.

[0073] The construction of the knowledge base usually depends on a large number of data sources, including office guides, tutorials, books, articles, etc. However, the quality of these data sources is uneven, and may contain incorrect and biased information. Ensuring the reliability of these data sources is a great challenge. In the embodiment, the recommended knowledge is reviewed to improve the quality of the knowledge, and thus the accuracy of the knowledge recommendation is improved.

[0074] For example, for the collected and feedback knowledge, an effective collaboration mechanism between artificial intelligence and human experts is established to ensure the quality of the knowledge. For example, the accuracy of the knowledge automatically entered by the machine or the recommended knowledge needs to be reviewed and calibrated by experts, and the government knowledge is formed after the review.

[0075] In some embodiments, the recommended knowledge is classified, and the recommended knowledge is reviewed and passed according to the category to form public knowledge.

[0076] In the embodiment, different categories of knowledge are reviewed according to the category, which can improve the quality of the knowledge and ensure the authority of the knowledge.

[0077] In some embodiments of the present disclosure, the collection of public knowledge is the basis for subsequent recommendation, and therefore, the collection of multi-source data is also very important. In the context of government multi-source data, the real-time and accuracy of data collection need to be solved. For example Figure 4 As shown in Figure 4 FIG. 4 is a flowchart of another embodiment of the knowledge processing method of the present disclosure, including steps S41-S44.

[0078] In step S41, the distributed collection of raw knowledge is performed.

[0079] The raw data can exist in various formats, including txt, word, excel, pdf, and json types, covering six categories of data such as policy documents, work guidelines, operation details, training materials, question and answer records, and open letters.

[0080] In step S42, data conversion and preprocessing are performed on the raw knowledge.

[0081] In some embodiments, the raw data in various formats and structures is converted into a unified format, and then quality assessment, cleaning, and other preprocessing are performed.

[0082] Specifically, the raw data is processed through data conversion, data segmentation, stop word removal, and feature construction, etc. Since the general segmentation does not have high precision for government affairs, the government knowledge is divided based on the constructed government dictionary.

[0083] In step S43, knowledge extraction is performed on the raw knowledge that has been data-converted and preprocessed, around the business domain, to form a structured knowledge representation.

[0084] For example, question and answer pairs are extracted around the government business domain and the matters belonging to it, and the question and answer pair information includes the business domain, the question, and the corresponding answer. Through this process, valuable knowledge points can be extracted from the preprocessed data to form a structured knowledge representation.

[0085] In step S44, the structured knowledge representation is organized to obtain a public knowledge base.

[0086] For example, on the basis of knowledge extraction, the knowledge is classified, labeled, stored, and other core links are organized and summarized to form a knowledge system that is easy to search and use.

[0087] During the use of the knowledge base, knowledge from different sources is often scattered in different departments or business systems, which easily forms knowledge islands. These knowledge islands lack effective connection and integration, which leads to the fact that the system cannot fully utilize these scattered knowledge resources. In this embodiment, through the distributed knowledge collection pipeline, automatic structured processing and segmentation of large-scale data, and knowledge extraction, the consistency and redundancy of data processing in a distributed environment can be ensured.

[0088] Figure 5 A structural schematic diagram of some embodiments of a knowledge recommendation module of the knowledge processing system of the present disclosure is shown. The knowledge recommendation module 500 includes a construction unit 510, a processing unit 520, a candidate unit 530, and a recommendation unit 540.

[0089] The constructing unit 510 is configured to construct a user knowledge graph according to historical operations of each user on knowledge in a public knowledge base.

[0090] The knowledge is, for example, public knowledge such as government affairs knowledge.

[0091] The processing unit 520 is configured to process the user knowledge graph to obtain a user feature vector and a knowledge feature vector.

[0092] In some embodiments, the user knowledge graph is processed by using a bipartite graph to obtain the user feature vector and the knowledge feature vector.

[0093] The candidate unit 530 is configured to obtain an embedding vector matrix of a plurality of candidate knowledge based on the user feature vector and the knowledge feature vector.

[0094] In some embodiments, the user feature vector and the knowledge feature vector are recalled according to a predetermined number to obtain the embedding vector matrix of the plurality of candidate knowledge.

[0095] The recommendation unit 540 is configured to perform knowledge recommendation for a target user by using the user feature vector and the knowledge feature vector based on historical operations of the target user on knowledge.

[0096] In some embodiments, the recommendation unit 540 is configured to perform information enhancement on the historical operations of the target user on knowledge to obtain a first embedding vector matrix corresponding to the target user; perform self-attention processing on the first embedding vector matrix to obtain a second embedding vector matrix; score each candidate knowledge in the plurality of candidate knowledge by using the second embedding vector matrix and the embedding vector matrix of the plurality of candidate knowledge; and perform knowledge recommendation for the target user based on a score of each candidate knowledge.

[0097] The information enhancement on the historical operations of the target user on knowledge to obtain the first embedding vector matrix corresponding to the target user includes: averaging a plurality of contexts of each session in the historical operations of the target user on knowledge to obtain a first fused embedding vector representation; performing time-varying information fusion on each session in the historical operations of the target user on knowledge to obtain a second fused embedding vector representation; performing weighted calculation on the first fused embedding vector representation and the second fused embedding vector representation to obtain a third fused embedding vector representation corresponding to each session; performing concatenation operation on the third fused embedding vector representation corresponding to each session and a time of each session, and processing by using a feedforward neural network to obtain an embedding vector matrix corresponding to each session; and obtaining the first embedding vector matrix corresponding to the target user according to the embedding vector matrix corresponding to each session.

[0098] The self-attention processing on the first embedding vector matrix obtains a second embedding vector matrix, including: performing normalization index processing on the first embedding vector matrix by using a query parameter matrix, a key parameter matrix and a value parameter matrix of a self-attention algorithm to obtain the second embedding vector matrix.

[0099] In the above embodiment, by constructing a user knowledge graph, a plurality of high-quality candidate knowledge embedding vector matrices are obtained, and then personalized knowledge recommendation is provided for a target user based on historical operations of the target user on knowledge and the embedding vector matrices of the plurality of candidate knowledge, so that accurate knowledge recommendation is realized, thereby improving service quality and efficiency.

[0100] In some embodiments of the present disclosure, as shown in Figure 6 Figure 6 The structure diagram of some embodiments of the knowledge processing system of the present disclosure includes a sharing module 610 configured to update the public knowledge base in response to at least one of an editing operation and an updating operation of the recommended knowledge by the target user.

[0101] For example, the sharing module 610 is configured to form public knowledge in response to the recommended knowledge being approved; and update the public knowledge base by using the public knowledge.

[0102] In some embodiments, the recommended knowledge is classified; and the recommended knowledge is approved according to the category to form public knowledge.

[0103] Through the sharing module, the knowledge in the public knowledge base can be kept updated and of higher quality, thereby improving the accuracy of knowledge recommendation.

[0104] In other embodiments, the knowledge processing system further includes a collection module 620 configured to collect original knowledge in a distributed manner, perform data conversion and preprocessing on the original knowledge, perform knowledge extraction on the original knowledge subjected to data conversion and preprocessing around a business domain, form a structured knowledge representation, and organize the structured knowledge representation to obtain the public knowledge base.

[0105] In this embodiment, by using a distributed knowledge collection pipeline, automatic structured processing and segmentation of large-scale data and knowledge extraction can ensure data consistency and redundancy in a distributed environment.

[0106] Figure 7 ​For structural schematic diagram of some embodiments of the knowledge processing system of the present disclosure, the knowledge processing system comprises a memory 710 and a processor 720. Among them: the memory 710 can be a disk, a flash memory or any other non-volatile storage medium. The memory is used to store the instructions in the above embodiments. The processor 720 is coupled to the memory 710 and can be implemented as one or more integrated circuits, such as a microprocessor or a microcontroller. The processor 720 is used to execute the instructions stored in the memory.

[0107] In some embodiments, the processor 720 is coupled to the memory 710 through the BUS bus 730. The knowledge processing system can also be connected to an external storage device 750 through a storage interface 740 to call external data, and can also be connected to a network or another computer system (not shown) through a network interface 760. Details are not introduced here.

[0108] In the following, the scheme of the present disclosure will be introduced taking the processing of government affair knowledge as an example. The core challenge of the government affair intelligent question and answer system lies in how to process and respond to the policy interpretation and consulting service, which requires the system to have a high method capability of government affair knowledge acquisition, updating and sharing.

[0109] Figure 8 Structural schematic diagram of some embodiments of the government affair knowledge management system of the present disclosure.

[0110] The government affair knowledge management system comprises a distributed knowledge acquisition module 810, a feedback-based knowledge updating and recommendation module 820, and a knowledge review and sharing module 830.

[0111] The distributed knowledge acquisition is the basis of the government affair knowledge management system, and the distributed knowledge acquisition module 810 can perform data conversion, knowledge extraction and knowledge acquisition. For example, different parts of government affair knowledge are acquired respectively. The distributed knowledge acquisition module 810 supports automatic structured processing and segmentation of large-scale data, knowledge extraction, and ensures data consistency and redundancy in a distributed environment.

[0112] The feedback-based knowledge updating and recommendation module 820 can make accurate knowledge recommendation based on the government affair knowledge system, and provide editing and updating capabilities. Throughout the process, the goal is to solve the problems of knowledge distribution, knowledge updating, version management, knowledge evaluation, and real-time and accuracy of the recommendation system.

[0113] The knowledge review and sharing module 830 can review and share the acquired and feedback knowledge. For example, an effective collaboration mechanism between artificial intelligence and human experts is established. Different parts of government affair knowledge are reviewed by experts of the corresponding part, and after the review is passed, the knowledge is shared as government affair knowledge, so as to ensure the authority and final quality of the knowledge.

[0114] The above-mentioned government affair knowledge management system can improve the quality and efficiency of government service through efficient knowledge management and intelligent technology.

[0115] In some embodiments, a computer program product is protected, including a computer program or instructions, which, when executed by a processor, implement the above-mentioned method. The computer program product includes a computer program carried on a computer readable medium, which contains program codes for executing the method shown in the flowchart. In such embodiments, the computer program can be downloaded and installed from a network through a communication device, or installed from a storage device, or installed from a ROM. When the computer program is executed by a CPU, the above-mentioned functions defined in the method of the embodiments of the present disclosure are executed.

[0116] In some embodiments, a computer program product is protected, including a computer program or instructions, which, when executed by a processor, implement the above-mentioned method. The computer program product includes a computer program carried on a computer readable medium, which contains program codes for executing the method shown in the flowchart. In such embodiments, the computer program can be downloaded and installed from a network through a communication device, or installed from a storage device, or installed from a ROM. When the computer program is executed by a CPU, the above-mentioned functions defined in the method of the embodiments of the present disclosure are executed.

[0117] The present disclosure is described with reference to the flowcharts and / or block diagrams according to the embodiments of the present disclosure. It should be understood that each flow and / or block in the flowcharts and / or block diagrams and the combination of the flows and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to produce a machine, so that the instructions executed by the computer or other programmable data processing devices produce a device that implements the functions specified in the flowcharts and / or block diagrams. Figure 1 The device that implements the functions specified in one or more flows and / or blocks. Figure 1 The device that implements the functions specified in one or more flows and / or blocks.

[0118] These computer program instructions can also be stored in a computer readable memory that can guide the computer or other programmable data processing devices to work in a specific way, so that the instructions stored in the computer readable memory produce a manufactured product that includes instruction devices, which implement the functions specified in the flowcharts and / or block diagrams. Figure 1 The device that implements the functions specified in one or more flows and / or blocks. Figure 1 The device that implements the functions specified in one or more flows and / or blocks.

[0119] These computer program instructions can also be loaded into a computer or other programmable data processing devices, so that a series of operational steps are performed on the computer or other programmable data processing devices, to generate computer-implemented processes, thus the instructions executed on the computer or other programmable data processing devices provide the function implemented in the flow Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0120] So far, the present disclosure has been described in detail. In order to avoid obscuring the concept of the present disclosure, some details known in the art are not described. Those skilled in the art can fully understand how to implement the technical solutions disclosed herein according to the above description.

[0121] The method and device of the present disclosure can be implemented in many ways. For example, the method and device of the present disclosure can be implemented by software, hardware, firmware or any combination of software, hardware and firmware. The above order of steps for the method is only for illustration, and the steps of the method of the present disclosure are not limited to the above specific description, unless otherwise specifically described. In addition, in some embodiments, the present disclosure can also be implemented as programs recorded in recording media, which include machine-readable instructions for implementing the method according to the present disclosure. Therefore, the present disclosure also covers the recording media storing the programs for executing the method according to the present disclosure.

[0122] Although some specific embodiments of the present disclosure have been described in detail through examples, those skilled in the art should understand that the above examples are only for illustration, not for limiting the scope of the present disclosure. Those skilled in the art should understand that the above embodiments can be modified without departing from the scope and spirit of the present disclosure. The scope of the present disclosure is defined by the appended claims.

Claims

1. A knowledge processing method, comprising: constructing a user knowledge graph according to historical operations of each user on knowledge in a public knowledge base; processing the user knowledge graph to obtain a user feature vector and a knowledge feature vector; obtaining an embedding vector matrix of a plurality of candidate knowledge based on the user feature vector and the knowledge feature vector; averaging a plurality of contexts of each session in historical operations of a target user on knowledge to obtain a first fused embedding vector representation; performing information fusion on time variation of each session in the historical operations of the target user on knowledge to obtain a second fused embedding vector representation; performing weighted calculation on the first fused embedding vector representation and the second fused embedding vector representation to obtain a third fused embedding vector representation corresponding to each session; performing a concatenation operation on the third fused embedding vector representation corresponding to each session and time of each session, and processing by using a feedforward neural network to obtain an embedding vector matrix corresponding to each session; and obtaining a first embedding vector matrix corresponding to the target user according to the embedding vector matrix corresponding to each session; performing self-attention processing on the first embedding vector matrix to obtain a second embedding vector matrix; scoring each candidate knowledge in the plurality of candidate knowledge based on a softmax algorithm by using the second embedding vector matrix and the embedding vector matrix of the plurality of candidate knowledge; and performing knowledge recommendation for the target user based on the score of each candidate knowledge. 2.The knowledge processing method of claim 1, further comprising: updating the public knowledge base in response to at least one of an editing operation and an updating operation on the recommended knowledge by the target user.

3. The knowledge processing method according to Claim 1, wherein The self-attention processing on the first embedding vector matrix to obtain a second embedding vector matrix comprises: performing a normalized exponential processing on the first embedding vector matrix by using a query parameter matrix, a key parameter matrix and a value parameter matrix of a self-attention algorithm to obtain the second embedding vector matrix.

4. The knowledge processing method according to Claim 1, wherein The processing of the user knowledge graph to obtain a user feature vector and a knowledge feature vector comprises: processing the user knowledge graph by using a bipartite graph to obtain the user feature vector and the knowledge feature vector.

5. The knowledge processing method according to Claim 2, wherein The updating of the public knowledge base further comprises: forming a public knowledge in response to the recommended knowledge being approved; and updating the public knowledge base by using the public knowledge.

6. The knowledge processing method according to Claim 5, wherein The forming of the public knowledge in response to the recommended knowledge being approved comprises: classifying the recommended knowledge; and forming the public knowledge after the recommended knowledge is approved according to the belonging category. 7.The knowledge processing method of any one of claims 1 to 6, further comprising: collecting original knowledge in a distributed manner; performing data conversion and preprocessing on the original knowledge; performing knowledge extraction on the original knowledge after data conversion and preprocessing around a business domain to form a structured knowledge representation; and organizing the structured knowledge representation to obtain the public knowledge base. ​ 8. A knowledge processing system comprising a knowledge recommendation module, the knowledge recommendation module comprising: a constructing unit configured to construct a user knowledge graph according to historical operations of each user on knowledge in a public knowledge base; a processing unit configured to process the user knowledge graph to obtain a user feature vector and a knowledge feature vector; a candidate unit configured to obtain an embedding vector matrix of a plurality of candidate knowledge based on the user feature vector and the knowledge feature vector; and a recommendation unit configured to average a plurality of contexts of each session in historical operations of a target user on knowledge, to obtain a first fused embedding vector representation, to perform time-varying information fusion on each session in the historical operations of the target user on knowledge, to obtain a second fused embedding vector representation, to perform weighted calculation on the first fused embedding vector representation and the second fused embedding vector representation, to obtain a third fused embedding vector representation corresponding to each session, to perform concatenation operation on the third fused embedding vector representation corresponding to each session and time of each session, and to process using a feedforward neural network to obtain an embedding vector matrix corresponding to each session, and to obtain a first embedding vector matrix corresponding to the target user according to the embedding vector matrix corresponding to each session; to perform self-attention processing on the first embedding vector matrix to obtain a second embedding vector matrix; to score each candidate knowledge in the plurality of candidate knowledge based on a softmax algorithm using the second embedding vector matrix and the embedding vector matrix of the plurality of candidate knowledge; and to recommend knowledge to the target user based on a score of each candidate knowledge.

9. The knowledge processing system of claim 8, further comprising: a sharing module configured to update the public knowledge base in response to at least one of an editing operation and an updating operation on recommended knowledge by the target user.

10. The knowledge processing system of claim 8 or 9, further comprising: a collection module configured to collect raw knowledge in a distributed manner, to perform data conversion and preprocessing on the raw knowledge, to perform knowledge extraction on the raw knowledge that has undergone data conversion and preprocessing around a business domain to form a structured knowledge representation, to organize the structured knowledge representation to obtain the public knowledge base.

11. A knowledge processing system comprising: a memory; and a processor coupled to the memory, the processor being configured to perform a knowledge processing method as claimed in any one of claims 1 to 7 based on instructions stored in the memory.

12. A computer-readable storage medium having computer program instructions stored thereon, the instructions being executed by a processor to implement a knowledge processing method as claimed in any one of claims 1 to 7.

13. A computer program product comprising computer programs or instructions, the computer programs or instructions being executed by a processor to implement a knowledge processing method as claimed in any one of claims 1 to 7. ​

Images