Business question-answering system and method, electronic equipment and medium

Abstract

One or more embodiments of the invention provide a business question-answering system and method, electronic equipment and a medium. The system comprises a question receiving module, a public knowledge acquisition module and an answer generation module. The question receiving module is used for receiving a business question asked by a user; the public knowledge obtaining module is used for obtaining first knowledge matched with the business problem from a public knowledge base based on the business problem, and the first knowledge is obtained by summarizing multiple pieces of public knowledge with association in the public knowledge base; and the answer generation module is used for generating answer information for the business question according to the business question and the public knowledge.

Description

Technical Field

[0001] One or more embodiments of the present specification relate to the field of artificial intelligence technology, and in particular, to a business question-answering system and method, electronic device, and medium. Background Art

[0002] The use of AI to automatically answer questions with users is becoming more and more widespread in e-commerce, customer service, financial consulting and other fields. For example, in the insurance field, AI can automatically communicate with users and answer their questions about insurance after-sales consultation. For another example, in the e-commerce field, AI can also be used to handle users' after-sales consultation questions about products. This automatic question-and-answer method reduces dependence on manual customer service, helps reduce service costs and improve service efficiency.

[0003] However, the automatic question answering system also faces some challenges. Sometimes the answers to user questions are not accurate enough and cannot solve the user's questions well. Therefore, how to improve the quality of the answers generated by the automatic question answering system is a problem that needs to be solved. Summary of the invention

[0004] In view of this, one or more embodiments of the present specification provide a business question-and-answer system and method, an electronic device, and a medium to improve the quality of answer generation in a business question-and-answer scenario.

[0005] To achieve the above objectives, one or more embodiments of this specification provide the following technical solutions:

[0006] According to a first aspect of an embodiment of this specification, a business question-answering system is provided, the system comprising: a question receiving module, a public knowledge acquisition module, and an answer generation module;

[0007] The question receiving module is used to receive business questions asked by users;

[0008] The public knowledge acquisition module is used to acquire first knowledge matching the business problem from a public knowledge base based on the business problem, where the first knowledge is summarized based on a plurality of public knowledge related to each other in the public knowledge base;

[0009] The answer generation module is used to generate answer information for the business question based on the business question and the public knowledge.

[0010] According to a second aspect of an embodiment of this specification, a business question and answer method is provided. The method is executed by any business question and answer system described in this specification, and the method includes:

[0011] Receive business questions from users;

[0012] Based on the business problem, first knowledge matching the business problem is acquired from a public knowledge base, where the first knowledge is summarized based on a plurality of related public knowledge in the public knowledge base;

[0013] Based on the business problem and the public knowledge, answer information for the business problem is generated.

[0014] According to a third aspect of the embodiments of this specification, an electronic device is provided, including:

[0015] processor;

[0016] a memory for storing processor-executable instructions;

[0017] The processor implements the method described in any embodiment of this specification by running the executable instructions.

[0018] According to a fourth aspect of the embodiments of this specification, a computer-readable storage medium is provided, on which computer instructions are stored, and when the instructions are executed by a processor, the method described in any embodiment of this specification is implemented.

[0019] According to a fifth aspect of the embodiments of this specification, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the method described in any embodiment of this specification is implemented.

[0020] The business question-and-answer system and method, electronic device and medium of the embodiments of this specification store first knowledge in a public knowledge base, and the first knowledge is summarized based on multiple public knowledge with association in the public knowledge base, that is, the first knowledge is obtained by considering the association between the various knowledge in the knowledge base. Compared with the storage of independent knowledge points, this contains more comprehensive information, which can help users obtain all information related to business issues in the knowledge base, thereby providing a more comprehensive perspective and helping to make the generated answers more complete and accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in one or more embodiments of the present disclosure or related technologies, the drawings required for use in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in one or more embodiments of the present disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0022] Figure 1 It is a system architecture diagram of a business question-and-answer system provided by an exemplary embodiment.

[0023] Figure 2It is a schematic diagram of constructing a public knowledge base provided by an exemplary embodiment.

[0024] Figure 3 It is a flow chart of a business question-and-answer method provided by an exemplary embodiment.

[0025] Figure 4 is a flow chart of a modeling algorithm provided by an exemplary embodiment.

[0026] Figure 5 It is a schematic diagram of graph structure clustering provided by an exemplary embodiment.

[0027] Figure 6 It is a schematic diagram of a business question-and-answer system provided by an exemplary embodiment.

[0028] Figure 7 It is a schematic diagram of constructing a private knowledge base provided by an exemplary embodiment.

[0029] Figure 8 is a schematic diagram of an electronic device provided by an exemplary embodiment. DETAILED DESCRIPTION

[0030] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The implementations described in the following exemplary embodiments do not represent all implementations consistent with one or more embodiments of this specification. Instead, they are merely examples of devices and methods consistent with some aspects of one or more embodiments of this specification as detailed in the appended claims.

[0031] It should be noted that: in other embodiments, the steps of the corresponding method are not necessarily performed in the order shown and described in this specification. In some other embodiments, the steps included in the method may be more or less than those described in this specification. In addition, a single step described in this specification may be decomposed into multiple steps for description in other embodiments; and multiple steps described in this specification may be combined into a single step for description in other embodiments.

[0032] Figure 1 is a system architecture diagram of a business question-answering system provided by an exemplary embodiment, such as Figure 1 As shown, the business question-answering system may include: a question receiving module 11, a public knowledge acquisition module 12 and an answer generation module 13.

[0033] The question receiving module 11 is used to receive business questions asked by users.

[0034] This embodiment does not limit what specific business questions are asked. For example, it can be a policy after-sales consultation question in the insurance field, or it can also be a disease-related consultation question in the medical field, etc.

[0035] The public knowledge acquisition module 12 can be used to: based on the above business problem, acquire first knowledge matching the business problem from the public knowledge base, and the first knowledge can be summarized based on multiple related public knowledge in the public knowledge base.

[0036] Specifically, you can combine Figure 2 Describe the public knowledge base.

[0037] like Figure 2 As shown, the construction of the public knowledge base can be processed by the public knowledge base construction module included in the business question and answer system. Specifically, the public knowledge base construction module can receive multiple public knowledge documents 22 uploaded to the public knowledge base 21, which can be uploaded by an administrator. These public knowledge documents may include business-related public knowledge. The public knowledge can be understood as some general knowledge in the business field. For example, taking the after-sales service in the insurance field as an example, some public knowledge documents related to insurance after-sales can be uploaded to the public knowledge base. Exemplarily, the public knowledge in these public knowledge documents may include, but is not limited to: the insurance claims process, the determination of insurance premiums, matters needing attention for insurance renewal, the rights and responsibilities of insurance, the understanding of insurance contracts and terms, etc.

[0038] These public knowledge in the public knowledge base may be related to each other. For example, some knowledge is about the insurance claim process, and some knowledge is about insurance renewal, etc. In other words, even if the knowledge in the public knowledge base is relatively scattered and independent when uploaded, some knowledge may be related to each other, such as knowledge belonging to the same content theme.

[0039] For example, the following pieces of public knowledge are related:

[0040] 1) “In the event of an accident, the policyholder should immediately notify the insurance company.”

[0041] 2) “The insurance company will conduct an investigation upon receipt of the notification.”

[0042] 3) “If the investigation confirms liability, the insurance company will compensate for the losses.”

[0043] Among them, the above-mentioned knowledge 1) to 3) are all related to the insurance processing process after the accident occurs, and these knowledge can be considered as multiple related public knowledge.

[0044] It is understandable that this embodiment does not limit the specific type of "association" in the "multiple public knowledge with association". For example, the above-mentioned content may belong to the same topic. For another example, multiple public knowledge may be related to the same entity. For example, in the automatic business question-and-answer scenario in the medical field, if the user wants to consult the knowledge related to leukemia, then multiple public knowledge related to "leukemia" can be queried from the public knowledge base in the medical field. These knowledge are all related to the disease "leukemia". The disease is equivalent to an "entity", and these knowledge also belong to multiple public knowledge with association. Other types of association are not listed in detail.

[0045] like Figure 2 As shown, in this embodiment, a plurality of public knowledge with associations can be obtained from a plurality of public knowledge documents in a public knowledge base through a modeling algorithm. For example, Figure 2 The knowledge A, knowledge B and knowledge C in are a group of public knowledge with association. It can be understood that more than one group of associated knowledge can be obtained through the modeling algorithm, and multiple groups of knowledge can be obtained, and each group of knowledge can include multiple public knowledge with association.

[0046] The corresponding first knowledge can also be generated based on the generalization. The first knowledge can be a combination of multiple related knowledge listed together to form a whole; or a coherent general description can be generated based on multiple related public knowledge.

[0047] Here are some examples:

[0048] Assume that there are three related pieces of knowledge as follows: "Failure to pay premiums on time may cause the policy to become invalid", "Insurance policy invalidation may affect claims", and "Insurance cannot be enjoyed when policy is invalid". Based on this, the first piece of knowledge generated may be "Failure to pay premiums on time may cause the policy to become invalid, which in turn affects claims and protection".

[0049] Each group of related knowledge can generate a corresponding first knowledge.

[0050] As described above, the public knowledge base stores first knowledge, which can be summarized based on multiple related public knowledge in the public knowledge base. In this embodiment, the public knowledge acquisition module 12 can acquire first knowledge matching the business question from the public knowledge base based on the business question asked by the user.

[0051] like Figure 2As shown, in order to facilitate the retrieval of the first knowledge, a corresponding retrieval element can usually be set for each first knowledge in the public knowledge base. For example, the retrieval element can be: vector, triple, etc. Taking the vector as an example, the first knowledge can be vectorized and its corresponding vector can be used as the retrieval element. Then in the public knowledge base, the first knowledge and its corresponding vector can be stored, and when retrieving the first knowledge, the similarity between the vectors can be calculated to match.

[0052] Specifically, when retrieving the first knowledge corresponding to the business question, the corresponding search element can be extracted according to the business question asked by the user, for example, the business question asked by the user can be converted into a corresponding vector. Then, the vector corresponding to the business question can be matched with the candidate vectors in the public knowledge base, and the similarity between the vectors can be calculated. The candidate vectors can be all the vectors in the public knowledge base, or a part of the vectors selected according to the set conditions, which is not limited in this embodiment.

[0053] When the similarity meets the preset conditions, such as reaching the similarity threshold, or obtaining the similarities ranked in the top N positions (N is a natural number, and is sorted from high to low), the first knowledge corresponding to the vectors that meet the similarity conditions can be determined as the first knowledge that matches the business problem.

[0054] For example, the business question asked by the user is "What is the insurance claims process?" Then, if a first knowledge K1 in the public knowledge base is related to the claims process, and another first knowledge K2 is related to premium renewal, when calculating the similarity between the vector corresponding to the business question and the vector corresponding to the first knowledge, the similarity of K1's vector will be higher than the similarity of K2's vector, thereby determining that the first knowledge K1 is the first knowledge that matches the business question.

[0055] After obtaining the first knowledge, the public knowledge acquisition module 12 can return the first knowledge to the question receiving module 11, and the question receiving module 11 can send the business question asked by the user and the first knowledge obtained from the public knowledge base to the answer generation module 13. The answer generation module 13 can generate answer information for the business question based on the business question asked by the user and the first knowledge.

[0056] For example, the answer generation module 13 can be implemented by LLM (Large Language Model). LLM can combine the first knowledge obtained from the query to output a more accurate answer to the business question. Moreover, the answer generated by LLM can be more natural and coherent, providing users with a better interactive experience.

[0057] As described above, the business question and answer system can provide more accurate answer information to the business questions asked by users by executing the business question and answer method, thereby meeting the consultation needs of users. Figure 3 is a flowchart of a business question-and-answer method provided by an exemplary embodiment. Figure 3 As shown, the business question and answer method may include the following processing:

[0058] In step 300, a business question asked by a user is received.

[0059] In step 302, based on the business problem, first knowledge matching the business problem is acquired from a public knowledge base, where the first knowledge is summarized based on a plurality of related public knowledge in the public knowledge base.

[0060] In step 304, answer information for the business question is generated based on the business question and public knowledge.

[0061] The business question answering system of this embodiment has the following advantages: the first knowledge stored in the public knowledge base of this embodiment is obtained by summarizing multiple public knowledge with association in the public knowledge base, that is, the first knowledge is obtained by considering the association between the knowledge in the knowledge base. Compared with the independent knowledge point storage, this has the following advantages:

[0062] First, the generalized first knowledge contains more comprehensive information, which can help users obtain all the information related to the business problem in the knowledge base, thereby providing a more comprehensive perspective and helping to make the generated answers more complete; and the general first knowledge can better capture the semantic relationship between knowledge points. For example, when multiple knowledge points are clustered together, the causal relationship, conditional relationship, etc. between them can be expressed more clearly. This helps to better understand the nature of the problem and provide more accurate answers. However, a single knowledge point may not be fully obtained when it is retrieved, and relevant information may be lost, thus failing to cover the required information.

[0063] Second, the first knowledge can integrate the related public knowledge into a coherent narrative, which is more fluent. When storing individual knowledge points, independent knowledge points are obtained during retrieval, which is not conducive to understanding. In addition, this first knowledge with better coherence makes the logical relationship between knowledge clearer. When the answer generation module generates an answer based on the first knowledge, it also helps to better understand the background and context of the question, thereby giving a more accurate answer.

[0064] Third, in terms of retrieval speed, if it is an independent knowledge point, then each knowledge point needs to be matched one by one, but for the first knowledge, only one search is needed to match the overview of multiple knowledge points, which is faster and more convenient.

[0065] Next, the embodiment of this specification provides an exemplary modeling algorithm for modeling knowledge in a public knowledge base: Figure 4 is a flowchart of a modeling algorithm provided by an exemplary embodiment, which can be modeled by GraphRAG (Graph-based Reinforced Aggregative Model for Abstractive Summarization), such as Figure 4 As shown, the following processing may be included:

[0066] In step 400, each public knowledge document uploaded to the public knowledge base is divided into blocks to obtain a plurality of text blocks.

[0067] In this step, the documents in the public knowledge base may be divided into blocks, for example, by sentences or paragraphs.

[0068] In step 402, for each text block, corresponding entities and relations are extracted.

[0069] For example, entities and relationships can be extracted in this step for use in graph construction in subsequent steps.

[0070] For example, LLM can be used to extract triples of each text block. Assuming that one of the text blocks is "failure to pay premiums on time may cause the policy to expire", the triple extracted from the text block may be {failure to pay premiums on time may cause, policy to expire}, then in the triple, "failure to pay premiums on time" and "policy to expire" will be entities, and "may cause" will be the relationship between the two entities.

[0071] In step 404, a graph structure is constructed based on entities and relationships extracted from public knowledge documents in the public knowledge base.

[0072] In this step, a graph structure may be constructed, and the graph structure may include the entities and relationships extracted in step 402. For example, in the above example, "failure to pay premiums on time" and "policy expiration" are two entities in the graph, and there is a directed edge between the two entities, and the directed edge points from the entity "failure to pay premiums on time" to the entity "policy expiration", and the meaning of this edge indicates that the relationship between the two entities is: failure to pay premiums on time "may lead to" the expiration of the policy.

[0073] In step 406, the graph structure is clustered by a clustering algorithm to obtain a plurality of public knowledge with associations belonging to the same category.

[0074] In this step, clustering analysis can be performed on entities and relationships in the graph structure by using a clustering algorithm, for example, the Leiden algorithm can be used. Through clustering, multiple public knowledge with associations belonging to the same category can be obtained, and these public knowledge with associations can be called a cluster cluster.

[0075] For example, the Leiden algorithm finds that the following triples belong to the same category, that is, the knowledge represented by these triples is related and can be called a cluster:

[0076] 1) {Failure to pay the premium on time may result in the policy becoming invalid}

[0077] 2) {The policy has expired and cannot be enjoyed or guaranteed}

[0078] 3) {Insurance policy expires, may affect claims}

[0079] The above triples are all about the same topic, that is, they are all about the knowledge of "policy expiration", for example, what circumstances will cause the policy to expire, and what impact will it have if the policy expires.

[0080] In step 408, corresponding first knowledge is generated based on the plurality of associated public knowledge.

[0081] In this step, the corresponding first knowledge can be summarized based on the multiple public knowledge of the same category obtained by clustering the Leiden algorithm. For example, in the above example, the first knowledge can be generated by the model as follows: "Failure to pay the premium on time may cause the policy to become invalid, and thus affect the protection and claims."

[0082] In this example, the public knowledge base may store various clusters and their corresponding first knowledge.

[0083] For the first knowledge in the public knowledge base, when searching for the first knowledge matching the question based on the business question, the following method can be used: extract the triple of the business question asked by the user through LLM, and match the triple of the business question with the cluster in the public knowledge base. When matching, the triple of the business question can be compared with the triple in the cluster for similarity, and then the similarity between the triple of the business question and the cluster is obtained.

[0084] For example, assuming that a triple is extracted based on a business problem, and assuming that a cluster in the library includes three triples, then during the matching calculation, the similarity between the triple extracted from the business problem and each triple in the cluster can be calculated separately, and these three similarities can be aggregated (for example, they can be averages), and the aggregated similarity can be used as the similarity between the cluster and the triple extracted from the business problem.

[0085] By calculating the similarity between the triples of the business problem and the various clusters in the public knowledge base, the similarities can be sorted, the clusters ranked in the top N positions can be selected, and the first knowledge corresponding to these clusters can be used as the first knowledge matching the business problem.

[0086] In the above example, the modeling algorithm is based on the method of obtaining a clustering cluster including multiple triples through a clustering algorithm, and querying the clustering cluster matching the business problem by calculating the similarity between the triples. In this example, the retrieval element can be understood as a triple, and the first knowledge matching the business problem is obtained by calculating the similarity between the triples.

[0087] In other examples, the modeling algorithm may be in other forms, and the search element may be other elements besides triples, which is not limited in the embodiments of this specification. For example, after clustering the graph structure by a clustering algorithm, an entity description may be obtained, and the entity description may be obtained based on entities and relationships associated with the same entity and belonging to the same cluster. Figure 5 The hint, Figure 5 The diagram shows part of the structure (only for illustration). Figure 5 The graph structure in can also be constructed by extracting entities and relationships from documents. The graph structure can include entities such as "Xiao Ming", "Xiao Hua", "Boy", "Girl", etc., as well as relationships between entities. For example, the relationship between Xiao Hua and diving is "hobby", that is, Xiao Hua's hobby is diving.

[0088] for Figure 5 The graph structure shown in the figure, when the graph structure is clustered by a clustering algorithm, entities and relationships belonging to the same category can be obtained. These interrelated entities and relationships belonging to the same category obtained by the clustering algorithm can constitute a "community". For example, community S1 may include four entities "Xiao Ming", "Xiao Hua", "Boy", "Hope Primary School" and their relationships. These entities and relationships are all related to the entity "Xiao Ming", so they belong to the same category of related entities and relationships. In addition, each pair of entities and their relationships also corresponds to a public knowledge. For example, the entities "Xiao Ming", "Xiao Hua" and their relationships in community S1 correspond to the knowledge {Xiao Ming and Xiao Hua are friends}, and the entities "Xiao Ming", "Hope Primary School" and their relationships correspond to the knowledge {Xiao Ming is a student of Hope Primary School}. Therefore, it can also be said that the above-mentioned community S1 includes multiple public knowledge with associations, and these knowledge are all associated with the entity "Xiao Ming".

[0089] Based on the knowledge of community S1, the following description of the entity "Xiaoming" can be generated:

[0090] "Xiao Ming is a boy, his friend is Xiao Hua, and he is a student of Hope Primary School."

[0091] Similarly, for each entity in the graph structure, the entity description corresponding to the entity can be obtained in the above manner. These entity descriptions can be called first knowledge, which is generated based on multiple public knowledge profiles associated with the entity.

[0092] For the convenience of subsequent retrieval, the entity description can also be converted into a corresponding vector. In subsequent retrieval, the business question asked by the user can also be converted into a corresponding vector. By calculating the similarity between vectors, the vector in the knowledge base that matches the business question is obtained, and the entity description corresponding to the vector is used as the first knowledge that matches the business knowledge. In this example, the retrieval element is a vector, and the similarity is evaluated by the vector distance.

[0093] In other examples, the business question-and-answer system of the embodiment of this specification can not only answer users' inquiries about general questions through the above-mentioned public knowledge base, but also meet the questions about private questions. The private questions mentioned here can be non-general questions related to the user's private business. For example, taking insurance after-sales as an example, the user's after-sales consultation about his or her own insurance policy is a private question. For example, the user asks "When is the effective date of my insurance policy ***?" This is a consultation about the user's specific insurance policy and is not a general question.

[0094] In order to meet the above-mentioned needs for answering users' private questions, the business question-answering system of this embodiment also adds a private knowledge acquisition module in the system. Figure 6 As shown, Figure 6 It is a schematic diagram of a business question-and-answer system provided by an exemplary embodiment. The private knowledge acquisition module 14 is connected to the question receiving module 11. The private knowledge acquisition module 14 can obtain second knowledge matching the business question from the private knowledge base based on the business question received by the question receiving module. The second knowledge is obtained based on the user's private business knowledge in the private knowledge base.

[0095] Similarly, you can combine Figure 7 , the construction of the private knowledge base is explained as follows: the private knowledge base 72 can be constructed by the private knowledge base construction module in the business question and answer system. Specifically, the private knowledge base construction module can receive the user's new business document 71. For example, taking insurance after-sales as an example, the private knowledge base construction module can monitor the policy issuance message and obtain the new policy purchased by the user. The new policy is the new business document. The new business document includes the user's private business knowledge. For the new policy purchased by the user, the information in the new policy is the user's private business knowledge.

[0096] Then, the new business document can be split into text blocks, and the text blocks obtained after the splitting can be vectorized. The text blocks obtained after the splitting can be called the second knowledge.

[0097] For example, for a newly purchased insurance policy, the structured information of the insurance policy can be converted into understandable text, electronic insurance policy, and other forms of text, and then the converted text can be split. Taking the conversion into understandable text as an example, a description template can be set for each structured information field, and then the corresponding field value in the insurance policy can be filled into the template, such as the remaining number of periods in the term insurance policy field (the template is that the remaining number of periods to be paid for the policy is ${remainPhase} periods), and then the description of each field is combined into a set of text. For example, for the comprehensible text obtained by conversion, it can be split according to sentences or paragraphs, and then the split text blocks can be converted into vectors. For example, the Word2Vec model can be used to convert text blocks into vectors. The second knowledge and the corresponding vectors can be stored in a private knowledge base.

[0098] The purpose of converting private knowledge into vectors as described above is to facilitate subsequent knowledge retrieval. For example, when retrieving the second knowledge, the business question asked by the user can be vectorized, and the vector corresponding to the business question can be calculated with the vector in the private knowledge base for similarity, and the text blocks ranked in the top N positions in similarity can be selected as the second knowledge matching the business question.

[0099] In addition, there can be multiple private knowledge bases. For example, taking the insurance scenario as an example, assuming that the user has multiple policies, a corresponding private knowledge base can be established for each policy. For example, the user's multiple policies include policy D1 and policy D2, then the text block of policy D1 can be vectorized and stored in the first private knowledge base, and the text block of policy D2 can be vectorized and stored in the second private knowledge base. In a specific implementation, when a policy issuance message is monitored, a private knowledge base corresponding to the new policy can be established. You can also monitor policy update information. If the policy is updated, the private knowledge base must also be updated accordingly to ensure the accuracy of the queried knowledge.

[0100] When querying the second knowledge, you first need to determine the private knowledge base to be queried. For example, in actual implementation, users can directly start business questions and answers on the policy details page of the corresponding policy, so that the business question and answer system can directly lock which policy the user is querying, thereby determining the private knowledge base corresponding to the policy. For another example, you can also let the user select the policy they want to query, and then determine the corresponding private knowledge base based on the policy selected by the user.

[0101] After acquiring the second knowledge, the private knowledge acquisition module can return the second knowledge to the question receiving module 11, and the question receiving module 11 can send the business problem and the acquired knowledge to the answer generation module, so that the answer generation module can generate answer information for the business problem based on the business problem and the acquired knowledge.

[0102] It should be noted that, for the first knowledge obtained from the public knowledge base and the second knowledge obtained from the private knowledge base, no matter what business question the user asks, the first knowledge and the second knowledge will be obtained from the above two knowledge bases based on the business question, and then the answer generation module will generate answer information for the business question based on "business question, first knowledge and second knowledge".

[0103] Another method may be: after receiving the business question asked by the user, the question receiving module may identify the business question and identify whether the question is a general question or a private question. If it is a general question, the first knowledge may be acquired only through the public knowledge acquisition module, and the answer generation module generates answer information based on the business question and the first knowledge. Alternatively, if it is a private question, the second knowledge may be acquired only through the private knowledge acquisition module, and the answer generation module generates answer information based on the business question and the second knowledge.

[0104] In the above-mentioned embodiment, a private knowledge acquisition module is added to the business question and answer system, and the module obtains second knowledge related to the user's personalized questions from the private knowledge base, which helps to better answer the user's private questions, thereby improving the quality of answer information generated by the business question and answer system.

[0105] In combination with any of the foregoing embodiments, an embodiment of this specification provides a business question-and-answer method, which can be performed by a business question-and-answer system of any embodiment of this specification. The method may include: the business question-and-answer system receives a business question asked by a user, and based on the business question, obtains first knowledge matching the business question from a public knowledge base, wherein the first knowledge is summarized based on multiple public knowledge related to each other in the public knowledge base; and the business question-and-answer system can generate answer information for the business question based on the business question and the public knowledge.

[0106] The detailed processing process of each step in the business question and answer method can be specifically referred to in the relevant implementation process in the above-mentioned business question and answer system embodiment, which will not be repeated here. It can also be understood that the business question and answer method executed by the business question and answer system, in addition to the above, can also include any method that the system can execute mentioned in the system embodiment.

[0107] For example, the business question and answer method may also include: the business question and answer system may also receive multiple uploaded public knowledge documents through a public knowledge base construction module, wherein the public knowledge documents include business-related public knowledge, and obtain multiple associated public knowledge from the multiple public knowledge documents through a modeling algorithm, and summarize and generate corresponding first knowledge based on the multiple associated public knowledge.

[0108] Among them, the public knowledge base construction module, when used to obtain multiple public knowledge with association from the multiple public knowledge documents through a modeling algorithm, can specifically include: dividing each public knowledge document into blocks to obtain multiple text blocks; for each text block, extracting corresponding entities and relationships; constructing a graph structure based on the entities and relationships extracted from the public knowledge documents in the public knowledge base; clustering the graph structure through a clustering algorithm to obtain multiple public knowledge with association belonging to the same category.

[0109] For another example, the business question-and-answer method may further include: a public knowledge acquisition module in the system, when used to acquire first knowledge matching the business problem from a public knowledge base based on the business problem, may specifically include: extracting corresponding search elements according to the business problem; matching the search elements with alternative search elements in the public knowledge base to obtain the similarity between the search elements, wherein each first knowledge in the public knowledge base has a corresponding search element; and using the first knowledge corresponding to the search element whose similarity meets the conditions as the first knowledge matching the business problem.

[0110] For example, the business question and answer method may also include: the system obtains second knowledge matching the business problem from a private knowledge base based on the business problem through a private knowledge acquisition module, and the second knowledge is obtained based on the user's private business knowledge in the private knowledge base; the system's answer generation module can generate answer information for the business problem based on the business problem and the acquired knowledge, and the knowledge at least includes the private knowledge.

[0111] For another example, the business question-and-answer method may also include: the system receives a new business document of a user through a private knowledge base construction module, wherein the new business document includes the user's private business knowledge; performs text splitting on the new business document, and uses the text blocks obtained after the splitting as the second knowledge; vectorizes the text blocks, and stores the second knowledge and the corresponding vectors into the private knowledge base.

[0112] For the method embodiment, since it basically corresponds to the system embodiment, the relevant parts can refer to the partial description of the system embodiment. The system embodiment described above is only schematic, wherein the modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical modules, that is, they may be located in one place, or they may be distributed on multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this specification. A person of ordinary skill in the art can understand and implement it without paying creative labor.

[0113] like Figure 8 As shown, Figure 8 A hardware structure diagram of an electronic device according to an embodiment of the present specification is shown, and the device may include: a processor 810, a memory 820, an input / output interface 830, a communication interface 840, and a bus 880. The processor 810, the memory 820, the input / output interface 830, and the communication interface 840 are connected to each other in communication within the device through the bus 880.

[0114] The processor 810 can be implemented by a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, etc., and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this specification. The processor implements the above method by running executable instructions.

[0115] The memory 820 for storing processor executable instructions can be implemented in the form of ROM (Read Only Memory), RAM (Random Access Memory), static storage device, dynamic storage device, etc. The memory 820 can store operating systems and other application programs. When the technical solutions provided in the embodiments of this specification are implemented by software or firmware, the relevant program codes are stored in the memory 820.

[0116] The input / output interface 830 is used to connect the input / output module to realize information input and output. The input / output module can be configured as a component in the device (not shown in the figure), or it can be externally connected to the device to provide corresponding functions. The input device may include a keyboard, a mouse, a touch screen, a microphone, various sensors, etc., and the output device may include a display, a speaker, a vibrator, an indicator light, etc.

[0117] The communication interface 840 is used to connect a communication module (not shown) to realize communication interaction between the device and other devices. The communication module can realize communication through a wired mode (such as USB, network cable, etc.) or a wireless mode (such as mobile network, WIFI, Bluetooth, etc.).

[0118] The bus 880 comprises a pathway that transmits information between the various components of the device (eg, the processor 810, the memory 820, the input / output interface 830, and the communication interface 840).

[0119] It should be noted that, although the above device only shows the processor 810, the memory 820, the input / output interface 830, the communication interface 840 and the bus 880, in the specific implementation process, the device may also include other components necessary for normal operation. In addition, it can be understood by those skilled in the art that the above device may also only include the components necessary for implementing the embodiments of the present specification, and does not necessarily include all the components shown in the figure.

[0120] The embodiments of this specification also provide a computer-readable storage medium on which a computer program is stored, and when the program is executed by a processor, the above-mentioned business question and answer method is implemented.

[0121] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0122] The embodiments of the present specification also provide a computer program, which, when executed, is used to implement the business question-and-answer method described in any embodiment of the present specification.

[0123] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0124] The above is a description of a specific embodiment of the specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0125] The user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of relevant countries and regions, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0126] The above description is merely a preferred embodiment of one or more embodiments of the present specification and is not intended to limit one or more embodiments of the present specification. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of one or more embodiments of the present specification shall be included in the scope of protection of one or more embodiments of the present specification.

Claims

1. A business question-answering system, characterized in that: The system comprises: a question receiving module, a public knowledge acquisition module, and an answer generating module; The question receiving module is used to receive business questions asked by users; The public knowledge acquisition module is used to acquire first knowledge matching the business problem from a public knowledge base based on the business problem, where the first knowledge is summarized based on a plurality of public knowledge related to each other in the public knowledge base; The answer generation module is used to generate answer information for the business question based on the business question and the public knowledge.

2. The system according to claim 1, characterized in that The system further comprises: a public knowledge base construction module; The public knowledge base construction module is used to: Receiving a plurality of uploaded public knowledge documents, wherein the public knowledge documents include business-related public knowledge; Obtaining a plurality of public knowledge with associations from the plurality of public knowledge documents through a modeling algorithm; Based on the plurality of associated public knowledge, corresponding first knowledge is generated through summarization.

3. The system according to claim 2, characterized in that The public knowledge base construction module, when used to obtain a plurality of public knowledge with associations from the plurality of public knowledge documents through a modeling algorithm, comprises: Each public knowledge document is divided into blocks to obtain multiple text blocks; For each text block, extract the corresponding entities and relations; Building a graph structure based on entities and relationships extracted from public knowledge documents in the public knowledge base; The graph structure is clustered by a clustering algorithm to obtain a plurality of public knowledge with associations belonging to the same category.

4. The system according to claim 1, characterized in that The public knowledge acquisition module, when used to acquire first knowledge matching the business problem from a public knowledge base based on the business problem, includes: According to the business problem, extract corresponding search elements; Matching the search element with candidate search elements in a public knowledge base to obtain similarities between the search elements, wherein each first knowledge in the public knowledge base has a corresponding search element; The first knowledge corresponding to the search element whose similarity satisfies the condition is taken as the first knowledge matching the business question.

5. The system according to any one of claims 1 to 4, characterized in that: The business question-answering system further includes: a private knowledge acquisition module; The private knowledge acquisition module is used to: based on the business problem, acquire second knowledge matching the business problem from a private knowledge base, where the second knowledge is obtained based on the user's private business knowledge in the private knowledge base; The answer generation module is further used to generate answer information for the business question based on the business question and the acquired knowledge, wherein the knowledge at least includes the private knowledge.

6. The system according to claim 5, characterized in that The system further comprises: a private knowledge base construction module; The private knowledge base construction module is used to: receiving a new business document of a user, wherein the new business document includes private business knowledge of the user; Performing text segmentation on the new business document, and using the text blocks obtained after the segmentation as the second knowledge; The text block is vectorized, and the second knowledge and the corresponding vector are stored in a private knowledge base.

7. A business question and answer method, characterized in that: The method is performed by the business question-answering system according to any one of claims 1 to 6, and the method comprises: Receive business questions from users; Based on the business problem, first knowledge matching the business problem is acquired from a public knowledge base, where the first knowledge is summarized based on a plurality of related public knowledge in the public knowledge base; Based on the business problem and the public knowledge, answer information for the business problem is generated.

8. An electronic device, characterized in that: include: processor; a memory for storing processor-executable instructions; The processor implements the method according to claim 7 by running the executable instructions.

9. A computer-readable storage medium having computer instructions stored thereon, characterized in that: When the instruction is executed by a processor, the method according to claim 7 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method according to claim 7 is implemented.

Images