Multi-Domain Knowledge Fusion Method for Intelligent Customer Service

Abstract

The present application provides a multi-domain knowledge fusion method for intelligent customer service, which relates to the technical field of data processing. The method includes: determining multi-domain data sources based on big data and performing knowledge extraction, connecting multi-domain knowledge according to association mapping rules, retrieving the target needs of users, formulating a fusion strategy, screening according to the target needs, performing knowledge fusion according to the fusion strategy, complementing knowledge for the fused data according to knowledge associations, then performing knowledge domain adaptation, and optimizing according to feedback information. Through the present application, the limitations existing in the prior art in semantic understanding can be solved. For complex or ambiguous problems, the intention of the user cannot be accurately understood. The user's questions can be understood more comprehensively, the accuracy and comprehensiveness of the answers can be improved, and it can be made closer to the level of human customer service.

Description

Technical Field

[0001] This application relates to the technical field of data processing, and in particular to a multi-domain knowledge fusion method for intelligent customer service. Background Art

[0002] With the popularization of the Internet and the increasing attention of enterprises to customer service, customers' requirements for service quality are constantly improving. Traditional manual customer service has become difficult to meet these needs due to high labor costs, low efficiency, and unstable service quality. Early intelligent customer service systems were mainly based on rules and templates, with limited ability to handle complex problems. Subsequently, intelligent customer service systems based on machine learning began to emerge. These systems can understand user intentions and provide more accurate answers, but there are still problems such as insufficient understanding of context and incomplete answers.

[0003] In summary, the existing technologies still have limitations in semantic understanding and cannot accurately understand the intentions of users for complex or ambiguous questions. Summary of the Invention

[0004] The purpose of this application is to provide a multi-domain knowledge fusion method for intelligent customer service to solve the problem that the existing technologies still have limitations in semantic understanding and cannot accurately understand the intentions of users for complex or ambiguous questions.

[0005] In view of the above problems, this application provides a multi-domain knowledge fusion method for intelligent customer service.

[0006] This application provides a multi-domain knowledge fusion method for intelligent customer service. The method includes: determining multi-domain data sources based on big data, extracting knowledge based on the multi-domain data sources, and establishing an association mapping rule for multi-domain knowledge according to the extracted information; connecting the multi-domain knowledge based on the association mapping rule to construct a first multi-domain knowledge information library; retrieving the conversation data of the intelligent customer service to determine the user's target needs, and formulating a target knowledge fusion strategy according to the user's target need information; screening the multi-domain knowledge according to the user's target needs to obtain a multi-target domain knowledge set, and performing knowledge fusion on the multi-target domain knowledge set by executing the target knowledge fusion strategy to obtain a target fusion knowledge data set; performing knowledge completion on the target fusion knowledge data set based on knowledge association, updating the first multi-domain knowledge information library according to the knowledge completion result to obtain a second multi-domain knowledge information library; traversing the second multi-domain knowledge information library to adapt to the knowledge domain of the intelligent customer service, generating an adaptation feedback information, and optimizing the intelligent customer service according to the adaptation feedback information.

[0007] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0008] By determining multi-domain data sources based on big data, extracting knowledge based on the multi-domain data sources, establishing association mapping rules for multi-domain knowledge according to the extracted information; connecting the multi-domain knowledge based on the association mapping rules to construct a first multi-domain knowledge information library; retrieving the conversation data of the intelligent customer service to determine the user's target needs, formulating a target knowledge fusion strategy according to the user's target needs information; screening the multi-domain knowledge according to the user's target needs to obtain a multi-target domain knowledge set, implementing the target knowledge fusion strategy to perform knowledge fusion on the multi-target domain knowledge set to obtain a target fusion knowledge data set; performing knowledge completion on the target fusion knowledge data set based on knowledge association, updating the first multi-domain knowledge information library according to the knowledge completion result to obtain a second multi-domain knowledge information library; traversing the second multi-domain knowledge information library to adapt to the knowledge domain of the intelligent customer service, generating an adaptation feedback information, and optimizing the intelligent customer service according to the adaptation feedback information, effectively solving the limitations still existing in the prior art in semantic understanding, and being unable to accurately understand the user's intention for complex or ambiguous problems, can understand the user's questions more comprehensively, improve the accuracy and comprehensiveness of the answers, and make it closer to the level of human customer service.

[0009] The above description is only an overview of the technical solution of the present application. In order to be able to understand the technical means of the present application more clearly, it can be implemented according to the content of the specification. And in order to make the above and other purposes, features and advantages of the present application more obvious and understandable, the specific embodiments of the present application are specifically exemplified below. It should be understood that the content described in this part is not intended to identify the key or important features of the embodiments of the present application, nor is it used to limit the scope of the present application. Other features of the present application will become easily understandable through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the technical solutions in the present application or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings described below are only exemplary, and for those of ordinary skill in the art, other drawings can be obtained according to the provided drawings without creative efforts.

[0011] Figure 1 It is a schematic flowchart of the multi-domain knowledge fusion method for the intelligent customer service of the present application;

[0012] Figure 2 It is a schematic flowchart of formulating a target knowledge fusion strategy for the multi-domain knowledge fusion method for the intelligent customer service of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0013] By providing a multi - domain knowledge fusion method for intelligent customer service, this application solves the problem that the prior art still has limitations in semantic understanding. For complex or ambiguous questions, it is unable to accurately understand the user's intention. It can more comprehensively understand the user's questions, improve the accuracy and comprehensiveness of answers, and make them closer to the level of human customer service.

[0014] Next, the technical solutions in this application will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all the embodiments of this application. It should be understood that this application is not limited by the example embodiments described here. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of this application. Additionally, it should be noted that for the sake of description, only the parts related to this application are shown in the drawings rather than all of them.

[0015] Embodiment 1

[0016] Please refer to the attached Figure 1 , this application provides a multi - domain knowledge fusion method for intelligent customer service, where the method specifically includes the following steps:

[0017] S1: Determine multi - domain data sources based on big data, perform knowledge extraction based on the multi - domain data sources, and establish association mapping rules for multi - domain knowledge according to the extracted information;

[0018] Specifically, using big data technology, widely collect data sources from different fields. These data sources include public databases, research reports, academic papers, social media, enterprise internal data, etc. Clean the collected raw data to remove duplicate, incorrect, or irrelevant information. Integrate the cleaned data in a certain format to form a unified data set. Use natural language processing technology to identify entities in the text data, such as names of people, places, organization names, etc. Through semantic analysis and pattern matching technologies, extract the relationships between entities from the text, such as someone is a management level of a certain company, etc. Identify and extract the event information described in the text, such as mergers, acquisitions, new product releases, etc. Extract the attribute information related to the entities, such as the price, color, size of the product, etc. Match and link the extracted entities with the entities in the existing knowledge base to ensure that the same entity in different data sources can be correctly identified and associated. According to the extracted relationship information, establish the relationship mapping rules between entities in different fields.

[0019] S2: Connect the multi - domain knowledge based on the association mapping rules to construct a first multi - domain knowledge information base;

[0020] Specifically, using the association mapping rules, knowledge entities in different fields are connected. For example, if the rule defines the association between an author and a work, the corresponding author entity is connected to its work entity. For knowledge entities with multiple associations, ensure that all necessary connections are correctly established according to the rules. Represent the multi-field knowledge after connection in the form of a graph, where nodes represent knowledge entities such as people, places, concepts, etc., and edges represent the relationships between entities. Use a graph database or a dedicated knowledge graph management tool to store and query this graph. Validate the constructed first multi-field knowledge information repository to check for incorrect connections or missing information.

[0021] S3: Retrieve the dialogue data of the intelligent customer service to determine the user's target needs, and formulate a target knowledge fusion strategy according to the user target need information;

[0022] Specifically, retrieve the historical dialogue data between the user and the customer service from the database of the intelligent customer service system. The dialogue data may include text chat records, text converted from voice, user feedback scores, etc. Use natural language processing techniques to perform text analysis on the dialogue data to identify the topics, intentions, and emotions of the user's questions. Through methods such as keyword extraction and semantic role labeling, deeply understand the user's query content and needs. Analyze the user's feedback scores and comments to understand the user's satisfaction and dissatisfaction with the current service. According to the analysis results of the user target needs, determine the key direction of knowledge fusion. For example, if it is found that users often ask questions about product features, then the integration and update of product information should be the focus. In response to the diversity and complexity of user needs, formulate a flexible knowledge fusion strategy. For common questions, strengthen the integration and presentation of relevant knowledge; for professional or complex questions, introduce more professional knowledge resources or expert systems. Optimize the way and timing of knowledge fusion. For example, relevant knowledge that may be needed can be pre-loaded before the user asks a question to reduce the user's waiting time.

[0023] S4: Screen the multi-field knowledge according to the user target needs to obtain a multi-target field knowledge set, and execute the target knowledge fusion strategy to perform knowledge fusion on the multi-target field knowledge set to obtain a target fusion knowledge data set;

[0024] Specifically, use methods such as keyword matching and semantic similarity calculation to select knowledge entries closely related to the user's needs. Integrate the selected relevant knowledge into a multi-target field knowledge set. This knowledge set will contain relevant knowledge points in multiple fields that meet the user's needs. According to the formulated target knowledge fusion strategy, perform fusion processing on the multi-target field knowledge set. The fusion may include merging duplicate or similar knowledge entries, supplementing missing information, correcting inaccurate data, etc. After the fusion processing, a more refined, accurate, and user-needs-satisfying target fusion knowledge data set is obtained.

[0025] S5: Perform knowledge completion on the target integrated knowledge dataset based on knowledge associations, update the first multi-domain knowledge information base according to the knowledge completion results, and obtain a second multi-domain knowledge information base;

[0026] Specifically, analyze the knowledge entries in the target integrated knowledge dataset and identify potential associations between them. These associations may include relationships between entities, hierarchical structures between concepts, temporal relationships between events, etc. By comparing and analyzing the associations between knowledge entries, identify possible missing knowledge links or information. For example, if there is a clear association between two entities but the direct description of this association is missing in the knowledge dataset, then this is a missing knowledge point. Develop a knowledge completion strategy, which may include using external data sources, reasoning mechanisms, or machine learning algorithms to predict and fill in the missing knowledge. External data sources can be public databases, professional websites, or relevant literature materials. The reasoning mechanism can perform logical inferences based on existing knowledge associations to discover new associations or attribute values. According to the developed strategy, obtain the missing knowledge from external data sources or infer new knowledge points through the reasoning mechanism. Verify the completed knowledge to ensure its authenticity and accuracy. Integrate the verified knowledge with the original target integrated knowledge dataset. Update the first multi-domain knowledge information base: Use the integrated knowledge dataset to update the first multi-domain knowledge information base, replacing or adding corresponding knowledge entries.

[0027] S6: Traverse the second multi-domain knowledge information base and perform knowledge domain adaptation with the intelligent customer service, generate adaptation feedback information, and optimize the intelligent customer service according to the adaptation feedback information.

[0028] Specifically, systematically access each knowledge entry in the second multi-domain knowledge information base. Ensure that various types of knowledge, such as entities, relationships, events, etc., can be identified and processed during the traversal. For each knowledge entry, check whether the intelligent customer service system can understand and process the relevant knowledge. Test the response ability and accuracy of the intelligent customer service system for different domain knowledge. Record the successful and failed cases during the adaptation process, as well as the performance of the intelligent customer service system when processing knowledge. Make corresponding adjustments and optimizations to the intelligent customer service system according to the problems pointed out in the adaptation feedback. This includes improving the natural language processing model, adding domain-specific knowledge bases, optimizing the dialogue strategy, etc. If the knowledge in certain domains is too complex or difficult for the intelligent customer service to understand, consider introducing human assistants to assist in processing. After each optimization, re-perform the adaptation test. Continuously adjust and optimize the system according to the test results until a satisfactory performance level is achieved.

[0029] Furthermore, step S1 of the present application further includes:

[0030] Use big data technology to screen and identify datasets in multiple fields, and determine multi-field data sources according to the identification results;

[0031] Integrate the multi-field data sources to build a multi-field data lake;

[0032] Based on the multi-field data lake, perform entity recognition to generate multiple entities, and use relation extraction technology to identify the association relationships between the multiple entities for formal construction to obtain knowledge triples;

[0033] According to the knowledge triples, use association rule mining algorithms to obtain the association rules between multi-field knowledge in the multi-field knowledge lake;

[0034] Through the similarity calculation of the multi-field knowledge, formulate mapping rules between the multi-field knowledge according to the association rules.

[0035] Specifically, through big data processing frameworks such as Hadoop and Spark, screen and identify datasets from different fields. Determine the data sources according to factors such as data quality, relevance, and reliability. Clean, transform, and integrate the screened multi-field data sources to ensure the unity of data formats and standards. Store all the integrated data in a centralized data lake, which will serve as the basis for subsequent knowledge extraction and association analysis. Use natural language processing technologies, such as named entity recognition, to identify entities from the multi-field data lake, such as person names, place names, company names, etc. Adopt relation extraction technology to identify the association relationships between multiple entities, and represent these relationships in a formal way, such as the triple entity 1, relation, entity 2. Based on the knowledge triples, use association rule mining algorithms such as Apriori and FP-Growth to find the association rules between multi-field knowledge. Analyze and verify the mined association rules to determine which rules are meaningful. Identify cross-field knowledge associations by calculating the similarity between different field knowledge, such as cosine similarity based on text vectors. According to the association rules and similarity calculation results, formulate mapping rules between multi-field knowledge. These rules can be used to associate and integrate knowledge from different fields.

[0036] Further, as Figure 2 , step S3 of this application further includes:

[0037] Analyze the dialogue data to identify multiple user demand information, cluster-analyze the multiple user demand information according to the multi-field knowledge, and extract the largest cluster according to the cluster analysis results;

[0038] Based on the largest cluster, determine the user's target demand, and determine multiple knowledge field fusion levels according to the user's target demand to delimit the knowledge field fusion scope;

[0039] Take the scope of the knowledge domain integration as the integration boundary value, and formulate the target knowledge integration strategy according to the multiple knowledge domain integration levels.

[0040] Specifically, collect the conversation data between the user and the intelligent customer service or related systems. Use natural language processing technology to analyze the conversation data, identify and extract the topics, intents, keywords, etc. asked by the user as the user demand information. Cluster the identified multiple user demand information according to multi-domain knowledge. Here, algorithms such as K-means and hierarchical clustering can be used to cluster similar user demand information into one category. Analyze the clustering results, and find out the clustering cluster that contains the most user demand information. This cluster represents the most concentrated and common needs of the users. Based on the user demand information in the largest clustering cluster, comprehensively summarize the main target needs of the users. According to the user target needs, analyze which knowledge domains need to be integrated and the association levels between these domains. According to the user target needs and the knowledge domain integration levels, clarify the specific scope of the knowledge domains to be integrated, which will be used as the boundary value of knowledge integration. Based on the defined scope of the knowledge domain integration, combined with the integration levels of multiple knowledge domains, formulate specific integration strategies. This includes determining which knowledge needs to be integrated first and how to integrate the knowledge of different domains, etc.

[0041] Furthermore, this application also includes:

[0042] Establish knowledge credibility based on the integration boundary value, traverse the multiple knowledge domain integration levels according to the knowledge credibility, and sequentially judge whether each knowledge domain integration level meets the preset credibility threshold;

[0043] If it is satisfied, assign a priority to the target knowledge domain integration level. If it is not satisfied, reset the priority of the target knowledge domain integration level, and iterate until each knowledge domain integration level has a priority;

[0044] Arrange the multiple knowledge domain integration levels in descending order according to the priority, and formulate the target knowledge integration strategy according to the priority sequence.

[0045] Specifically, based on the fusion boundary values, an initial credibility value is assigned to each knowledge area or knowledge point. This credibility value can be set according to factors such as the reliability of the data source, the update frequency of the knowledge, and the historical usage feedback. For example, a credibility range from 0 to 1 can be set, where 1 represents completely credible and 0 represents completely non-credible. Traverse according to the determined multiple knowledge area fusion levels. For each fusion level, evaluate the credibility of the knowledge points included therein. Set a preset credibility threshold for determining whether the credibility of the knowledge point or knowledge area meets the requirements. If the credibility of a certain knowledge area fusion level is satisfied, that is, greater than or equal to the preset credibility threshold, then this level is considered credible. If the credibility of the knowledge area fusion level meets the preset threshold, assign a priority value to this level. For example, a relatively high priority value can be assigned. If not satisfied, reset the priority of this level. For example, a lower priority value can be assigned, or directly set to the default priority. Repeat the above steps until each knowledge area fusion level is assigned a priority. According to the priority values of each knowledge area fusion level, sort all levels in descending order. The fusion levels with higher priorities will be ranked in the front, indicating that they should be processed preferentially during the knowledge fusion process. Based on the sorted priority sequence, formulate a specific knowledge fusion strategy. The knowledge areas or levels with higher priorities will obtain more fusion resources and attention to ensure the accuracy and effectiveness of the fusion results.

[0046] Furthermore, step S4 of the present application further includes:

[0047] Analyze the user's target requirements to set an expected threshold, and perform knowledge retrieval according to the expected threshold to obtain a retrieved knowledge set;

[0048] Based on knowledge timeliness, screen the retrieved knowledge set to obtain the multi-target domain knowledge set;

[0049] Specifically, based on the user's needs, set appropriate expected thresholds for knowledge retrieval, such as requirements in terms of relevance, timeliness, and authority. Use the set expected thresholds to retrieve relevant information in the knowledge base to obtain a preliminary retrieved knowledge set. Screen the retrieved knowledge to remove outdated or no longer relevant information to ensure the timeliness and accuracy of the knowledge, thereby forming a multi-target domain knowledge set. The domain knowledge closely related to the user's needs screened out is composed into a multi-target domain knowledge set, and each target domain knowledge set corresponds to a specific user need.

[0050] Furthermore, step S4 of the present application further includes:

[0051] Based on the knowledge fusion channel, perform knowledge fusion on each target domain knowledge set within the multi-target domain knowledge set through the target knowledge fusion strategy, and determine whether there are fusion conflicts in the knowledge fusion;

[0052] When there are fusion conflicts, identify the conflict types, formulate conflict resolution strategies according to the conflict types, execute the conflict resolution strategies to complete the knowledge fusion of the multi-target domain knowledge set, and obtain the target fusion knowledge data set.

[0053] Specifically, through the formulated target knowledge fusion strategy, fuse each target domain knowledge in the multi-target domain knowledge set. During the fusion process, detect whether there are fusion conflicts, such as inconsistent, duplicate or contradictory information. Once a fusion conflict is found, first identify the specific type of the conflict, such as data inconsistency, information redundancy, etc. According to the identified conflict type, formulate corresponding resolution strategies, such as conflict resolution, conflict avoidance, conflict ignoring, etc. Implement the formulated conflict resolution strategies to ensure the accuracy and integrity of the knowledge fusion. After the above steps, finally obtain a conflict-free and high-quality target fusion knowledge data set.

[0054] Furthermore, this application also includes:

[0055] If there are fusion conflicts in the knowledge fusion, record the fusion conflict data and analyze the conflict sources;

[0056] Perform knowledge tracing based on the conflict sources, and generate fusion conflict constraints according to the tracing results;

[0057] Integrate the fusion conflict constraints and synchronize them to the knowledge fusion channel.

[0058] Specifically, when conflicts are detected during the knowledge fusion process, automatically record these conflict data. The recorded content includes the conflict knowledge points, conflict types, involved data sources, etc. Conduct in-depth analysis on the recorded conflict data to determine the conflict sources. Possible conflict sources include data source inconsistencies, knowledge representation differences, outdated information, etc. Based on the analyzed conflict sources, perform knowledge tracing. This includes tracing the sources of knowledge points to understand how they are collected, processed and fused. The tracing results reveal problems with data sources, errors in the processing process or inconsistencies in knowledge representation. According to the tracing results, generate specific fusion conflict constraints. These constraints are rules, conditions or strategies aimed at preventing similar conflicts from occurring again in the future. Integrate the generated fusion conflict constraints into the knowledge fusion system. This may include updating the system rule base, strategy base or algorithm logic. Ensure that these new constraints are synchronized to the knowledge fusion channel so that they can be applied and executed during subsequent knowledge fusion processes.

[0059] Further, step S5 of this application further includes:

[0060] Perform knowledge association analysis on the target fused knowledge dataset to obtain knowledge association factors;

[0061] Construct a knowledge association network based on the knowledge association factors;

[0062] Traverse the target fused knowledge dataset through the knowledge association network for missing identification to determine knowledge missing information;

[0063] Generate a knowledge completion instruction according to the knowledge missing information, and perform knowledge completion on the target fused knowledge dataset through the knowledge completion instruction to generate the knowledge completion result.

[0064] Specifically, conduct in-depth analysis on the target knowledge dataset that has been fused to identify potential associations between different knowledge points. By analyzing co-occurrence relationships, causal relationships, similarities, etc. in the dataset, knowledge association factors are extracted. These association factors are the basis for constructing a knowledge association network. Based on the extracted knowledge association factors, a knowledge association network is constructed. In this network, nodes represent different knowledge points, and edges represent the association relationships between these knowledge points. Optimize the network, such as removing weak or redundant edges to highlight the core association structure. By traversing the constructed knowledge association network, check each node and edge in the network to identify possible knowledge missing. During the traversal process, if it is found that some key nodes lack necessary connections or some important associations are not established, it is determined as knowledge missing information. According to the identified knowledge missing information, corresponding knowledge completion instructions are generated. These instructions include adding new associations, improving node information, etc. By executing these knowledge completion instructions, perform a completion operation on the target fused knowledge dataset. After the completion operation, a more perfect and closely associated knowledge base is generated as the final result.

[0065] In summary, the multi-domain knowledge fusion method for intelligent customer service provided by this application has the following technical effects:

[0066] By determining multi-domain data sources based on big data, extracting knowledge based on the multi-domain data sources, establishing association mapping rules for multi-domain knowledge according to the extracted information; connecting the multi-domain knowledge based on the association mapping rules to construct a first multi-domain knowledge information library; retrieving the conversation data of the intelligent customer service to determine the user's target needs, formulating a target knowledge fusion strategy according to the user's target need information; screening the multi-domain knowledge according to the user's target needs to obtain a multi-target domain knowledge set, implementing the target knowledge fusion strategy to perform knowledge fusion on the multi-target domain knowledge set to obtain a target fusion knowledge data set; performing knowledge completion on the target fusion knowledge data set based on knowledge association, updating the first multi-domain knowledge information library according to the knowledge completion result to obtain a second multi-domain knowledge information library; traversing the second multi-domain knowledge information library to perform knowledge domain adaptation with the intelligent customer service, generating adaptation feedback information, and optimizing the intelligent customer service according to the adaptation feedback information, effectively solving the limitation that the prior art still has in semantic understanding. For complex or ambiguous problems, it is unable to accurately understand the user's intention, can understand the user's questions more comprehensively, improve the accuracy and comprehensiveness of the answers, and make it closer to the level of human customer service.

[0067] The foregoing description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0068] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalent technologies, the present application is also intended to include these changes and modifications.

Claims

1. A multi-domain knowledge fusion method for intelligent customer service, characterized in that: The method comprises: Determine multi-domain data sources based on big data, extract knowledge based on the multi-domain data sources, and establish association mapping rules for multi-domain knowledge based on the extracted information; Connecting the multi-domain knowledge based on the association mapping rule to construct a first multi-domain knowledge information base; Retrieve the conversation data of the intelligent customer service to determine the user's target needs, and formulate a target knowledge fusion strategy based on the user's target demand information; Screening the multi-domain knowledge according to the user's target requirements to obtain a multi-target domain knowledge set, executing the target knowledge fusion strategy to fuse the multi-target domain knowledge set to obtain a target fusion knowledge data set; Performing knowledge completion on the target fusion knowledge data set based on knowledge association, updating the first multi-domain knowledge information base according to the knowledge completion result, and acquiring a second multi-domain knowledge information base; Traversing the second multi-domain knowledge information base to perform knowledge domain adaptation with the intelligent customer service, generating adaptation feedback information, and optimizing the intelligent customer service according to the adaptation feedback information; The method of retrieving the conversation data of the intelligent customer service to determine the user's target demand and formulating the target knowledge fusion strategy according to the user's target demand information includes: Analyze the conversation data to identify multiple user demand information, perform cluster analysis on the multiple user demand information according to the multi-domain knowledge, and extract the largest cluster according to the cluster analysis result; Determine the user target requirement based on the largest cluster, determine multiple knowledge domain fusion levels according to the user target requirement, and define the knowledge domain fusion scope; Taking the knowledge domain fusion range as the fusion boundary value, formulating the target knowledge fusion strategy according to the multiple knowledge domain fusion levels; The method of taking the knowledge domain fusion range as the fusion boundary value and formulating the target knowledge fusion strategy according to the multiple knowledge domain fusion levels includes: Establishing knowledge credibility based on the fusion boundary value, traversing the multiple knowledge domain fusion levels according to the knowledge credibility, and determining in turn whether each knowledge domain fusion level meets a preset credibility threshold; If satisfied, the priority of the target knowledge domain fusion level is assigned; if not satisfied, the priority of the target knowledge domain fusion level is reset, and the process is repeated until each knowledge domain fusion level has a priority; The plurality of knowledge domain fusion levels are arranged in descending order according to the priority, and the target knowledge fusion strategy is formulated according to the priority sequence.

2. The method according to claim 1, characterized in that Determine multi-domain data sources based on big data, extract knowledge based on the multi-domain data sources, and establish association mapping rules for multi-domain knowledge based on the extracted information. The method includes: Use big data technology to screen and identify data sets in multiple fields, and determine multi-field data sources based on the identification results; Integrate the multi-domain data sources to build a multi-domain data lake; Perform entity recognition based on the multi-domain data lake to generate multiple entities, use relationship extraction technology to identify the association relationship between the multiple entities for formal construction, and obtain knowledge triples; Acquire association rules between multi-domain knowledge in the multi-domain knowledge lake by using an association rule mining algorithm according to the knowledge triples; By calculating the similarity of the multi-domain knowledge, a mapping rule between the multi-domain knowledge is formulated according to the association rule.

3. The method according to claim 1, characterized in that The multi-domain knowledge is screened according to the user's target requirements to obtain a multi-target domain knowledge set, the method comprising: Analyze the user's target requirements to set an expected threshold, perform knowledge retrieval according to the expected threshold, and obtain a retrieved knowledge set; The retrieval knowledge set is screened based on knowledge timeliness to obtain the multi-target domain knowledge set.

4. The method according to claim 1, characterized in that Executing the target knowledge fusion strategy to fuse the multi-target domain knowledge set to obtain a target fusion knowledge data set, the method includes: Based on the knowledge fusion channel, each target domain knowledge set in the multi-target domain knowledge set is fused through the target knowledge fusion strategy to determine whether there is a fusion conflict in the knowledge fusion; When there is a fusion conflict, the conflict type is identified, a conflict resolution strategy is formulated according to the conflict type, and the conflict resolution strategy is executed to complete the knowledge fusion of the multi-target domain knowledge sets and obtain the target fusion knowledge data set.

5. The method according to claim 4, characterized in that The method also includes: If there is a fusion conflict in the knowledge fusion, the fusion conflict data is recorded and the source of the conflict is analyzed; Perform knowledge tracing based on the conflict source, and generate fusion conflict constraints according to the tracing result; The fusion conflict constraints are integrated and synchronized to the knowledge fusion channel.

6. The method according to claim 1, characterized in that The target fusion knowledge data set is supplemented with knowledge based on knowledge association, the first multi-domain knowledge information base is updated according to the knowledge supplement result, and a second multi-domain knowledge information base is obtained, the method comprising: Performing knowledge association analysis on the target fusion knowledge data set to obtain knowledge association factors; Constructing a knowledge association network based on the knowledge association factors; Traversing the target fusion knowledge data set through the knowledge association network to identify missing information and determine missing knowledge information; A knowledge completion instruction is generated according to the knowledge missing information, and the target fusion knowledge data set is completed through the knowledge completion instruction to generate the knowledge completion result.

Images