Intelligent consultation method and consultation platform in combination with demand semantic analysis

By pre-processing word segmentation and semantic analysis and annotation of user consulting information, a consulting semantic system is built, and a semantic retrieval model is constructed using recurrent neural networks to build a semantic retrieval model, which solves the problem of inaccurate user consulting in the existing technology and improves the consulting effect.

CN119129609AInactive Publication Date: 2024-12-13NANTONG TIMES INFORMATION TECH CO LTD

Patent Information

Application Number
CN202411620791.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-11-14
Publication Date
2024-12-13
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

In the prior art, the degree of labeling management of user consultation is low, resulting in inaccurate understanding of user needs by the consulting platform, which affects the consulting effect.

Method used

By collecting user consultation information database, perform word segmentation preprocessing and semantic analysis and annotation, build a consulting semantic system, use recurrent neural network structure to build a semantic search model, conduct semantic analysis and search, and optimize the consulting results based on user feedback to improve the accuracy of label management Spend.

Benefits of technology

It realizes a more accurate understanding and label management of user consultation, improves the consulting effect, and ensures the accurate satisfaction of user needs.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119129609A_ABST
    Figure CN119129609A_ABST
Patent Text Reader

Abstract

The invention provides an intelligent consultation method and consultation platform combined with demand semantic analysis, and relates to the technical field of semantic analysis, the method comprises the following steps: acquiring a user consultation information database, carrying out word segmentation preprocessing to obtain an available consultation database, and building a consultation semantic system according to a consultation application scene; performing semantic analysis labeling and classification integration on the available consultation database to obtain a consultation semantic tag set and a consultation semantic tag database; training the consultation semantic label database, constructing a consultation semantic retrieval model, and performing semantic analysis retrieval on the consultation request information to obtain a target consultation retrieval result; and obtaining user consultation interaction feedback data, and carrying out business consultation optimization control. According to the method and the device, the technical problem of inaccurate understanding of a consultation platform on user requirements in the prior art can be solved, the technical target of improving the accuracy degree of labeled management on user consultation is achieved, and the technical effect of improving the consultation effect is achieved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the technical field of semantic analysis, and in particular to an intelligent consulting method and consulting platform combined with demand semantic analysis. Background Art

[0002] The consulting platform is an online platform that provides consulting services, aiming to provide users with professional answers and advice in various fields.

[0003] At present, although the existing technologies have achieved automatic classification and sentiment analysis of user needs to a certain extent, in practical applications, due to the diversity and complexity of user needs, and most of the user's consultations are not labeled, it is often difficult to fully meet the actual needs. For example, for some vague or ambiguous questions, the consultation platform may not be able to accurately understand the user's intention, resulting in inaccurate answers or failure to meet the user's needs. Therefore, a method is needed to solve the above problems.

[0004] In summary, the existing technology has a technical problem that due to the low degree of labeling management of user consultations, the consultation platform has an inaccurate understanding of user needs, which further affects the consultation effect. Summary of the invention

[0005] The purpose of this application is to provide an intelligent consulting method and consulting platform combined with demand semantic analysis, so as to solve the technical problem in the prior art that the consulting platform has an inaccurate understanding of user needs due to the low degree of labeling management of user consultations, which further affects the consulting effect.

[0006] In view of the above problems, the present application provides an intelligent consulting method and consulting platform combined with demand semantic analysis.

[0007] In a first aspect, the present application provides an intelligent consulting method combined with demand semantic analysis, wherein the method is implemented through an intelligent consulting platform combined with demand semantic analysis, wherein the method comprises: S1: acquiring a user consulting information database, wherein the user consulting information database comprises historical user consulting data and corresponding consulting result data, performing word segmentation preprocessing on the user consulting information database to obtain an available consulting database, and at the same time building a consulting semantic system according to a consulting application scenario; S2: performing semantic analysis and annotation on the available consulting database through the consulting semantic system to obtain a consulting semantic label set, and classifying and integrating the available consulting database based on the consulting semantic label set to obtain a consulting semantic label database; S3: training the consulting semantic label database using a recurrent neural network structure, constructing a consulting semantic retrieval model, obtaining consulting request information of a target user, performing semantic analysis and retrieval on the consulting request information based on the consulting semantic retrieval model, and obtaining a target consulting retrieval result; S4: performing business scenario interaction based on the target consulting retrieval result, obtaining user consulting interaction feedback data, and performing business consulting optimization management and control through the user consulting interaction feedback data.

[0008] In the second aspect, the present application also provides an intelligent consulting platform combined with demand semantic analysis, which is used to execute the intelligent consulting method combined with demand semantic analysis as described in the first aspect, wherein the platform includes: a consulting semantic system building module, the consulting semantic system building module is used to collect and obtain a user consulting information database, the user consulting information database includes historical user consulting data and corresponding consulting result data, the user consulting information database is pre-processed with word segmentation to obtain a usable consulting database, and a consulting semantic system is built according to the consulting application scenario; a consulting semantic label database acquisition module, the consulting semantic label database acquisition module is used to perform semantic analysis and annotation on the usable consulting database through the consulting semantic system to obtain consulting semantics A tag set is provided, and the available consulting database is classified and integrated based on the consulting semantic tag set to obtain a consulting semantic tag database; a target consulting retrieval result obtaining module is used to train the consulting semantic tag database using a recurrent neural network structure, build a consulting semantic retrieval model, obtain the consulting request information of the target user, perform semantic analysis and retrieval on the consulting request information based on the consulting semantic retrieval model, and obtain the target consulting retrieval result; a business consulting optimization management and control module is used to perform business scenario interaction based on the target consulting retrieval result, obtain user consulting interaction feedback data, and perform business consulting optimization management and control through the user consulting interaction feedback data.

[0009] One or more technical solutions provided in this application have at least the following technical effects or advantages: A user consultation information database is acquired by collecting and preprocessing the user consultation information database, which includes historical user consultation data and corresponding consultation result data. A usable consultation database is obtained, and a consultation semantic system is built according to the consultation application scenario. The usable consultation database is semantically analyzed and annotated by the consultation semantic system to obtain a consultation semantic tag set, and the usable consultation database is classified and integrated based on the consultation semantic tag set to obtain a consultation semantic tag database. The consultation semantic tag database is trained by using a recurrent neural network structure to construct a consultation semantic retrieval model, and the consultation request information of the target user is obtained. The consultation request information is semantically analyzed and retrieved based on the consultation semantic retrieval model to obtain the target consultation retrieval result. Business scenario interaction is performed based on the target consultation retrieval result to obtain user consultation interaction feedback data, and business consultation optimization management and control are performed through the user consultation interaction feedback data, thereby achieving the technical goal of improving the accuracy of labeling management of user consultations and achieving the technical effect of improving consultation results.

[0010] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented according to the contents of the specification, and in order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are specifically cited below. It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present application, nor is it intended to limit the scope of the present application. Other features of the present application will become easy to understand through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In order to more clearly illustrate the technical solutions in the present application or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only exemplary, and for ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0012] Figure 1 A flowchart of an intelligent consulting method combined with demand semantic analysis for this application; Figure 2 This is a schematic diagram of the structure of the intelligent consulting platform combined with demand semantic analysis for this application.

[0013] Description of reference numerals: A consulting semantic system building module 11, a consulting semantic tag database obtaining module 12, a target consulting search result obtaining module 13, and a business consulting optimization management and control module 14. DETAILED DESCRIPTION

[0014] This application solves the technical problem in the prior art that the consultation platform has an inaccurate understanding of user needs due to the low degree of labeling management of user consultations, which further affects the consultation effect by providing an intelligent consultation method and consultation platform combined with demand semantic analysis. The technical goal of improving the accuracy of labeling management of user consultations is achieved, and the technical effect of improving the consultation effect is achieved.

[0015] Below, the technical solutions in the present application will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments of the present application. It should be understood that the present application is not limited to the example embodiments described herein. Based on the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present application. It should also be noted that, for the convenience of description, only the parts related to the present application are shown in the accompanying drawings, rather than all of them.

[0016] Embodiment 1 Please see attached Figure 1 The present application provides an intelligent consulting method combined with demand semantic analysis, wherein the method is applied to an intelligent consulting platform combined with demand semantic analysis, and the method specifically includes the following steps: S1: Acquire a user consultation information database, which includes historical user consultation data and corresponding consultation result data, perform word segmentation preprocessing on the user consultation information database to obtain a usable consultation database, and build a consultation semantic system according to the consultation application scenario; Specifically, based on user authorization, the user consultation information database is collected and obtained. Among them, the user consultation information database includes historical user consultation data and corresponding consultation result data. The continuous text in the user consultation information database is divided into independent vocabulary units through word segmentation, which helps to extract key information in the text and obtain a usable consultation database. At the same time, a consultation semantic system is built according to the consultation application scenario. For example, the consultation system of the telecommunications application scenario.

[0017] S2: performing semantic analysis and annotation on the available consulting database through the consulting semantic system to obtain a consulting semantic tag set, and classifying and integrating the available consulting database based on the consulting semantic tag set to obtain a consulting semantic tag database; Specifically, semantic analysis is a natural language processing technology used to understand and interpret the meaning of text. Through the established consulting semantic system, in-depth semantic analysis is performed on the available consulting database to identify key concepts, entities, and the relationship between key concepts and entities in the consulting information, and then the corresponding semantic tags are marked for each piece of consulting information. After semantic analysis and annotation, the set of consulting semantic tags is obtained. Among them, the semantic tags of each piece of consulting information in the available consulting database are included.

[0018] S3: Using a recurrent neural network structure to train the consultation semantic tag database, construct a consultation semantic retrieval model, obtain consultation request information of a target user, perform semantic analysis and retrieval on the consultation request information based on the consultation semantic retrieval model, and obtain a target consultation retrieval result; Specifically, the consultation semantic label database is trained using a recurrent neural network structure, and the user consultation data corresponding to the label is used as the input data. After keyword filtering and retrieval semantic recognition, the matching labeled consultation result data is obtained as the output data, and then a consultation semantic retrieval model is constructed. After the model training is completed, the consultation request information of the target user is obtained. Among them, the consultation request information of the target user can be a text input in real time, or it can be a collection of historical consultation data. Based on the trained consultation semantic retrieval model, the consultation request information of the target user is subjected to semantic analysis and retrieval. Through the model's processing of the consultation request information, a target consultation retrieval result that matches the target user's consultation intention is obtained.

[0019] Step 4: S4: Perform business scenario interaction based on the target consultation retrieval results, obtain user consultation interaction feedback data, and perform business consultation optimization management and control through the user consultation interaction feedback data.

[0020] Specifically, after obtaining the target consulting search results, they are presented to the user and business scenario interaction is performed. The search results are displayed to the user in an appropriate manner, such as through a chat interface, email notification, web page display, etc., so that the user can view and understand them. At the same time, necessary interactive functions are provided, such as comments, likes, sharing, etc., so that the user can express satisfaction with the search results or make suggestions for improvement. In the process of business scenario interaction, the user's consulting interaction feedback data is collected. For example, it can include the user's direct evaluation of the search results, such as satisfaction rating, good / bad reviews, etc., and other user behavior data, such as the number of clicks, browsing time, bounce rate, etc., which are used to reflect the user's acceptance and satisfaction with the search results, and provide an important basis for subsequent optimization. The user's consulting interaction feedback data is analyzed to obtain the user's satisfaction with the search results, demand characteristics, and possible problems. By adding interactive feedback data to the consulting semantic retrieval model, the consulting semantic retrieval model is adjusted to provide a specific direction for optimization.

[0021] The intelligent consultation method combining demand semantic analysis is applied to an intelligent consultation platform combining demand semantic analysis, which can achieve the technical goal of improving the accuracy of tagging management of user consultations and achieve the technical effect of improving the consultation effect.

[0022] Furthermore, this application also includes the following steps: S11: Obtain abnormal factors of consultation data, where the abnormal factors of consultation data include duplicate values, missing values, irrelevant symbols, and stop words; S12: Perform preprocessing analysis on the abnormal factors of the consultation data to obtain a data preprocessing program, where the data preprocessing program is mapped and matched with the abnormal factors of the consultation data; S13: Based on the abnormal factors of the consultation data and the data preprocessing program, perform abnormal data identification and abnormal data preprocessing on the user consultation information database in sequence to obtain a standard consultation information database; S14: Perform word segmentation processing on each piece of consultation information in the standard consultation information database through TF-IDF to determine the available consultation database.

[0023] Specifically, obtaining the abnormal factors of consultation data is used to determine the possible abnormal factors in the consultation data. The abnormal factors include duplicate values, missing values, irrelevant symbols, and stop words. Duplicate values refer to duplicate records or data items existing in the user consultation information database. Missing values refer to values that are not filled or cannot be obtained for some fields or data items. Irrelevant symbols refer to punctuation marks, special characters, and irrelevant information that are irrelevant to the consultation content. Removing irrelevant symbols can avoid interference that may be caused by irrelevant symbols in subsequent text processing. Stop words refer to words that frequently appear in the text but contribute little to the meaning of the text, such as "de", "shi", "zai", etc. By removing stop words, the noise of the data set can be reduced and the efficiency of subsequent processing can be improved.

[0024] Then, after identifying the abnormal factors, it is necessary to determine a solution for processing the abnormal factors as a data preprocessing program. Among them, the data preprocessing program is mapped and matched with the abnormal factors of the consultation data. For example, for duplicate values, duplicate records may need to be deleted or merged, and for missing values, they may need to be filled with the average value, median, or other methods. For irrelevant symbols and stop words, text cleaning techniques may need to be used to remove them.

[0025] Next, according to the determined abnormal factors of the consultation data and the preprocessing program, operate on the actual user consultation information database. Identify the abnormal data in the user consultation information database and apply the preprocessing program to process the abnormal data. After processing, a standard consultation information database is obtained, and the data in the standard consultation information database is clean, complete, and standardized.

[0026] Next, TF-IDF, or term frequency-inverse document frequency, is a text mining technique used to evaluate the importance of words in a document set or corpus. Each consulting information in the standard consulting information database is segmented by TF-IDF to extract keywords or key phrases to represent the main content of the consulting information, which is helpful for subsequent text analysis or information retrieval. After the processing is completed, a usable consulting database is obtained for subsequent analysis or application.

[0027] By preprocessing and standardizing the user consultation information database, useful and standardized information can be extracted to facilitate subsequent analysis or application.

[0028] Furthermore, the present application also includes the following steps: S21: extracting knowledge attributes from the consulting semantic system to obtain a consulting knowledge attribute type set, and using the consulting knowledge attribute type set as a consulting knowledge entity set; S22: Filling content based on the consulting knowledge entity set to obtain a consulting knowledge entity content set, and using the consulting knowledge entity content set as a knowledge graph node; S23: constructing a consulting knowledge graph based on the consulting knowledge entity set and the consulting knowledge entity content set; S24: Using the consulting knowledge graph, label and classify each consulting result data in the available consulting database to obtain the consulting semantic label set.

[0029] Specifically, key knowledge attributes are extracted from the established consulting semantic system. For example, it can include various concepts, classifications, attributes, etc. related to the consulting content. By extracting knowledge attributes, a set of consulting knowledge attribute types can be formed. For example, the consulting types of the telecommunications consulting system include operation consulting, telephone fee consulting, and communication consulting.

[0030] Then, fill in specific content or description for each consulting knowledge entity to form a consulting knowledge entity content set. The content can be definitions, explanations, examples, etc. related to the entity. For example, phone bill consultation specifically includes phone bill balance, broadband charges, etc. After completing the content filling, the consulting knowledge entity content set is used as a node of the knowledge graph. Each node represents a consulting knowledge entity with specific content and meaning.

[0031] Next, the knowledge graph is a graphical knowledge representation method that can display the association and hierarchy between knowledge entities through the relationship between nodes and edges. The consulting knowledge graph is constructed through the consulting knowledge entity set and the consulting knowledge entity content set. The consulting knowledge entity nodes will be connected according to the relationship and attributes between entities to form a complete knowledge graph, which is used to clearly display various knowledge entities in the consulting field and their relationship, providing convenience for subsequent applications.

[0032] Next, we will use the established consulting knowledge graph to label and classify each consulting result data in the available consulting database. By comparing the consulting result data with the nodes and relationships in the knowledge graph, we will identify the corresponding label for each consulting result data to reflect the main content and category of the consulting result data, thereby forming a consulting semantic label set to better understand and organize consulting data and provide support for subsequent data analysis and application.

[0033] By constructing a consulting knowledge graph and achieving labeled classification of consulting result data, it helps to improve the efficiency and accuracy of consulting processing, and can also provide strong support for subsequent consulting analysis and optimization.

[0034] Furthermore, the present application also includes the following steps: S31: Perform keyword screening and annotation on each user's consultation data in the available consultation database to obtain a consultation data keyword annotation set; S32: Performing semantic filtering layer training based on the consulting data keyword annotation set to obtain a keyword filtering annotation network; S33: performing feature mapping annotation on the consultation semantic label database through the consultation data keyword annotation set to obtain a consultation semantic feature label data set; S34: Performing semantic recognition training on the consultation semantic feature label data set using a recurrent neural network structure to construct a consultation semantic recognition network; S35: Connect and merge the keyword filtering and annotation network with the consulting semantic recognition network to generate the consulting semantic retrieval model.

[0035] Specifically, each user consultation data in the available consultation database will be subjected to keyword screening and annotation, so that key information in the consultation data can be extracted, providing a basis for the subsequent construction of a semantic retrieval model. After the annotation is completed, a consultation data keyword annotation set is obtained.

[0036] Then, the semantic filtering layer is trained based on the consultation data keyword annotation set to preliminarily screen and filter the consultation data according to the keywords and exclude data that is irrelevant to the query intent. Among them, supervised learning methods such as neural networks or support vector machines can be used to build a keyword filtering and annotation network to learn to classify and filter user consultation data according to keywords to improve the accuracy and efficiency of subsequent semantic retrieval.

[0037] Next, the consultation semantic tag database is feature mapped and annotated using the consultation data keyword annotation set, converting the raw data into features that can be understood and processed by the machine learning algorithm. Based on the keyword annotation set, each piece of data in the consultation semantic tag database is mapped to a corresponding feature vector. Feature mapping can be achieved through methods such as TF-IDF weighting, which is used to convert keywords in the consultation data into numerical feature representations to facilitate subsequent machine learning operations.

[0038] Next, after obtaining the consultation semantic feature label dataset, the recurrent neural network structure is used for semantic recognition training. During the training process, the consultation semantic feature label dataset is used as input, and the semantic representation and category information of the data are learned through the recurrent neural network structure. By continuously adjusting the network parameters and optimizing the loss function, a recognition network that can accurately recognize the consultation semantics is constructed.

[0039] In addition, the keyword filtering and annotation network is connected and merged with the consulting semantic recognition network to generate the final consulting semantic retrieval model. The functions of keyword filtering and semantic recognition are integrated to quickly and accurately retrieve relevant consulting information according to the user's query intention.

[0040] The generated consulting semantic retrieval model can provide users with efficient and accurate consulting services.

[0041] Furthermore, the present application also includes the following steps: Performing word segmentation extraction on each user's consultation data in the available consultation database to obtain a consultation data word segmentation result set; Perform keyword screening on each user consultation data based on the consultation data word segmentation result set to obtain a user consultation data keyword set; The keyword annotation of each user consultation data is performed using the user consultation data keyword set to obtain the consultation data keyword annotation set.

[0042] Specifically, each user consultation data in the available consultation database is pre-processed by word segmentation. The process of dividing the continuous text into independent vocabulary units helps to extract the key information in the text. Existing word segmentation tools or algorithms, such as dictionary-based word segmentation methods, statistical-based word segmentation methods, etc., are used to ensure the accuracy and efficiency of word segmentation. After word segmentation pre-processing, each user consultation data is converted into a set of vocabulary units. These word segmentation results are organized into a set, namely the consultation data segmentation result set, for subsequent keyword screening and labeling.

[0043] Then, based on the consultation data word segmentation result set, keyword screening is performed on each user's consultation data to extract words that can represent the core information of the consultation content, and obtain the user consultation data keyword set.

[0044] Next, keyword annotation is performed on each user consultation data through the user consultation data keyword set, and the screened keywords are matched and marked with the user consultation data to form a consultation data keyword annotation set, so that each user consultation data is accompanied by corresponding keyword annotation information, which provides convenience for subsequent semantic analysis and retrieval.

[0045] By obtaining a set of keyword annotations for consulting data, important data support is provided for subsequent tasks such as building consulting semantic retrieval models, performing semantic analysis and retrieval, etc.

[0046] Furthermore, the present application also includes the following steps: Performing frequency statistics on each word segmentation information in the consultation data word segmentation result set to obtain a consultation word segmentation frequency information set; Performing inverse frequency calculation based on the consultation word segmentation frequency information set and the corpus database to obtain an inverse text frequency index set; The product of the consultation word segmentation frequency information set and the inverse text frequency index set is used as the consultation word segmentation key coefficient set; Based on the consultation word segmentation key degree coefficient set, word segmentation screening within a preset key degree threshold is performed to obtain a keyword screening result set, and each user consultation data is marked by the keyword screening result set to obtain the user consultation data keyword set.

[0047] Specifically, the frequency of occurrence of each word segmentation information in the consultation data word segmentation result set is counted, and then the frequency of occurrence of each word segmentation in the overall consultation data is obtained, providing basic data for the subsequent calculation of the inverse text frequency index and the key coefficient. By accessing the word segmentation result set in sequence, the number of occurrences of each word segmentation is counted, and a consultation word segmentation frequency information set is generated to obtain each word segmentation and its corresponding frequency of occurrence, which is the TF value.

[0048] Then, the inverse text frequency, or IDF value, is calculated based on the consultation word segmentation frequency information set and the corpus database. IDF is a measurement method used to evaluate the importance of a word in the entire corpus, which can filter out words that frequently appear in a few documents, i.e. keywords. The calculation formula for IDF is: IDF=log(N / 1+df), where N is the total number of documents in the corpus and df is the number of documents. By calculating the IDF value of each word segmentation, the inverse text frequency index set is obtained.

[0049] Next, the product of the consultation word segmentation frequency information set and the inverse text frequency index set is used as the consultation word segmentation criticality coefficient, namely TF-IDF, which is used to evaluate the importance of the word in the document or corpus. The calculation formula of TF-IDF is: TF-IDF(t, d)=TF(t, d)×IDF(t), where TF(t, d) is the word frequency of term t in document d, and IDF(t) is the inverse document frequency of term t. Among them, the larger the product of TF and IDF, the greater the criticality of the word, and vice versa. By calculating the TF-IDF value of each word segmentation, the consultation word segmentation criticality coefficient set is obtained.

[0050] Next, based on the set of key coefficients of the consultation word segmentation, a preset key threshold is set to filter the word segmentation. The word segmentation with a key coefficient exceeding the threshold is considered as a keyword, and a keyword screening result set is obtained, which contains all the word segmentations identified as keywords. Each user consultation data is marked through the keyword screening result set to generate a user consultation data keyword set. Each user consultation data is associated with its corresponding keyword, which provides key information for subsequent semantic analysis and retrieval.

[0051] By screening out keywords in user consultation data and marking them, it provides important data support for the subsequent construction of a consultation semantic retrieval model.

[0052] Furthermore, the present application also includes the following steps: S41: If the user consultation interaction feedback data does not meet the consultation result expectations, the consultation request information and the target consultation retrieval result are used as consultation negative samples; S42: establishing a consultation retrieval penalty mechanism, and calculating a penalty factor for the consultation negative sample based on the consultation retrieval penalty mechanism to obtain negative sample penalty factor information; S43: Optimizing the consulting semantic retrieval model based on the negative sample penalty factor information, generating a consulting semantic optimization retrieval model for business consulting optimization management and control.

[0053] Specifically, the consultation request information that does not meet the expected consultation results and the corresponding target consultation retrieval results are used as consultation negative samples to represent cases where the model fails to correctly understand and meet user needs.

[0054] Then, according to the characteristics and quantity of negative consultation samples, a consultation retrieval penalty mechanism is established, and the penalty factor is calculated for each negative consultation sample to obtain the negative sample penalty factor information. Among them, it can be designed based on multiple factors, such as the number of negative samples, the degree of incorrect matching of keywords in negative samples, the degree of negativity of user feedback, etc., to quantify the impact of negative samples on model performance and provide a basis for subsequent optimization.

[0055] Next, the consultation semantic retrieval model is optimized using the calculated negative sample penalty factor information. This can be achieved by adding items related to the penalty factor to the loss function. The model is trained using a loss function with a penalty factor so that the model can better understand and match the user's consultation request and improve the accuracy and relevance of the retrieval results. After optimization, a consultation semantic optimization retrieval model is generated. The consultation semantic optimization retrieval model is adjusted and improved based on the feedback from negative samples and can better meet the user's consultation needs.

[0056] By utilizing user consultation interaction feedback data to optimize the consultation semantic retrieval model, the accuracy of the model and user satisfaction can be improved, thereby improving the quality and efficiency of business consultation.

[0057] In summary, the intelligent consultation method combined with demand semantic analysis provided by this application has the following technical effects: A user consultation information database is acquired by collecting and preprocessing the user consultation information database, which includes historical user consultation data and corresponding consultation result data. A usable consultation database is obtained, and a consultation semantic system is built according to the consultation application scenario. The usable consultation database is semantically analyzed and annotated by the consultation semantic system to obtain a consultation semantic tag set, and the usable consultation database is classified and integrated based on the consultation semantic tag set to obtain a consultation semantic tag database. The consultation semantic tag database is trained by using a recurrent neural network structure to construct a consultation semantic retrieval model, and the consultation request information of the target user is obtained. The consultation request information is semantically analyzed and retrieved based on the consultation semantic retrieval model to obtain the target consultation retrieval result. Business scenario interaction is performed based on the target consultation retrieval result to obtain user consultation interaction feedback data, and business consultation optimization management and control are performed through the user consultation interaction feedback data, thereby achieving the technical goal of improving the accuracy of labeling management of user consultations and achieving the technical effect of improving consultation results.

[0058] Embodiment 2 Based on the intelligent consulting method combined with demand semantic analysis in the aforementioned embodiment, the present application also provides an intelligent consulting platform combined with demand semantic analysis, please refer to the attached Figure 2 , the platform comprises: A consultation semantic system building module 11 is used to collect and obtain a user consultation information database, which includes historical user consultation data and corresponding consultation result data, perform word segmentation preprocessing on the user consultation information database to obtain a usable consultation database, and build a consultation semantic system according to the consultation application scenario; A consultation semantic tag database obtaining module 12, wherein the consultation semantic tag database obtaining module 12 is used to perform semantic analysis and annotation on the available consultation database through the consultation semantic system to obtain a consultation semantic tag set, and classify and integrate the available consultation database based on the consultation semantic tag set to obtain a consultation semantic tag database; A target consultation retrieval result obtaining module 13 is used to train the consultation semantic tag database using a recurrent neural network structure, construct a consultation semantic retrieval model, obtain consultation request information of a target user, and perform semantic analysis and retrieval on the consultation request information based on the consultation semantic retrieval model to obtain a target consultation retrieval result; The business consultation optimization management and control module 14 is used to perform business scenario interaction based on the target consultation retrieval results, obtain user consultation interaction feedback data, and perform business consultation optimization management and control through the user consultation interaction feedback data.

[0059] Furthermore, the consultation semantic system building module 11 in the platform is also used for: Acquire abnormal factors of consulting data, wherein the abnormal factors of consulting data include repeated values, missing values, irrelevant symbols and stop words; Performing preprocessing analysis on the abnormal factors of the consulting data to obtain a data preprocessing program, wherein the data preprocessing program is mapped and matched with the abnormal factors of the consulting data; Based on the abnormal factors of the consultation data and the data preprocessing program, the user consultation information database is sequentially subjected to abnormal data identification and abnormal data preprocessing to obtain a standard consultation information database; Each consulting information in the standard consulting information database is segmented by TF-IDF to determine the available consulting database.

[0060] Furthermore, the consultation semantic tag database obtaining module 12 in the platform is also used for: Extracting knowledge attributes from the consulting semantic system to obtain a consulting knowledge attribute type set, and using the consulting knowledge attribute type set as a consulting knowledge entity set; Filling content based on the consulting knowledge entity set to obtain a consulting knowledge entity content set, and using the consulting knowledge entity content set as a knowledge graph node; Constructing a consulting knowledge graph based on the consulting knowledge entity set and the consulting knowledge entity content set; The consulting knowledge graph is used to label and classify each consulting result data in the available consulting database to obtain the consulting semantic label set.

[0061] Furthermore, the target consultation retrieval result obtaining module 13 in the platform is also used for: Perform keyword screening and annotation on each user's consultation data in the available consultation database to obtain a consultation data keyword annotation set; Performing semantic filtering layer training based on the consulting data keyword annotation set to obtain a keyword filtering annotation network; Performing feature mapping annotation on the consultation semantic tag database through the consultation data keyword annotation set to obtain a consultation semantic feature tag data set; Using a recurrent neural network structure to perform semantic recognition training on the consultation semantic feature label data set to construct a consultation semantic recognition network; The keyword filtering and annotation network is connected and merged with the consulting semantic recognition network to generate the consulting semantic retrieval model.

[0062] Furthermore, the target consultation retrieval result obtaining module 13 in the platform is also used for: Performing word segmentation extraction on each user's consultation data in the available consultation database to obtain a consultation data word segmentation result set; Perform keyword screening on each user consultation data based on the consultation data word segmentation result set to obtain a user consultation data keyword set; The keyword annotation of each user consultation data is performed using the user consultation data keyword set to obtain the consultation data keyword annotation set.

[0063] Furthermore, the target consultation retrieval result obtaining module 13 in the platform is also used for: Performing frequency statistics on each word segmentation information in the consultation data word segmentation result set to obtain a consultation word segmentation frequency information set; Performing inverse frequency calculation based on the consultation word segmentation frequency information set and the corpus database to obtain an inverse text frequency index set; The product of the consultation word segmentation frequency information set and the inverse text frequency index set is used as the consultation word segmentation key coefficient set; Based on the consultation word segmentation key degree coefficient set, word segmentation screening within a preset key degree threshold is performed to obtain a keyword screening result set, and each user consultation data is marked by the keyword screening result set to obtain the user consultation data keyword set.

[0064] Furthermore, the business consultation optimization management and control module 14 in the platform is also used for: If the user consultation interaction feedback data does not meet the consultation result expectations, the consultation request information and the target consultation retrieval results are used as consultation negative samples; Establishing a consultation retrieval penalty mechanism, calculating a penalty factor for the consultation negative sample based on the consultation retrieval penalty mechanism, and obtaining negative sample penalty factor information; The consulting semantic retrieval model is optimized based on the negative sample penalty factor information, and a consulting semantic optimization retrieval model is generated to perform business consulting optimization management and control.

[0065] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The intelligent consulting method combined with demand semantic analysis and the specific examples in the aforementioned embodiment one are also applicable to the intelligent consulting platform combined with demand semantic analysis in this embodiment. Through the aforementioned detailed description of the intelligent consulting method combined with demand semantic analysis, those skilled in the art can clearly know the intelligent consulting platform combined with demand semantic analysis in this embodiment, so for the sake of brevity of the specification, it will not be described in detail here. For the platform disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part description.

[0066] The above 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 may 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 will conform to the widest scope consistent with the principles and novel features disclosed herein.

[0067] 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 belong to the scope of the present application and its equivalent technology, the present application is also intended to include these modifications and variations.

Claims

1. An intelligent consulting method combined with demand semantic analysis is characterized by: The method comprises: S1: Acquire a user consultation information database, which includes historical user consultation data and corresponding consultation result data, perform word segmentation preprocessing on the user consultation information database to obtain a usable consultation database, and build a consultation semantic system according to the consultation application scenario; S2: performing semantic analysis and annotation on the available consulting database through the consulting semantic system to obtain a consulting semantic tag set, and classifying and integrating the available consulting database based on the consulting semantic tag set to obtain a consulting semantic tag database; S3: Using a recurrent neural network structure to train the consultation semantic tag database, construct a consultation semantic retrieval model, obtain consultation request information of a target user, perform semantic analysis and retrieval on the consultation request information based on the consultation semantic retrieval model, and obtain a target consultation retrieval result; S4: Perform business scenario interaction based on the target consultation retrieval result, obtain user consultation interaction feedback data, and perform business consultation optimization management and control through the user consultation interaction feedback data.

2. The method according to claim 1, characterized in that The available consulting database obtained in S1 includes: S11: Obtaining abnormal factors of consulting data, wherein the abnormal factors of consulting data include repeated values, missing values, irrelevant symbols, and stop words; S12: Preprocessing and analyzing the abnormal factors of the consulting data to obtain a data preprocessing program, wherein the data preprocessing program is mapped and matched with the abnormal factors of the consulting data; S13: Based on the abnormal factors of the consultation data and the data preprocessing program, abnormal data identification and abnormal data preprocessing are performed on the user consultation information database in sequence to obtain a standard consultation information database; S14: performing word segmentation processing on each consulting information in the standard consulting information database by using TF-IDF to determine the available consulting database.

3. The method according to claim 1, characterized in that The consulting semantic tag set obtained in S2 includes: S21: extracting knowledge attributes from the consulting semantic system to obtain a consulting knowledge attribute type set, and using the consulting knowledge attribute type set as a consulting knowledge entity set; S22: Filling content based on the consulting knowledge entity set to obtain a consulting knowledge entity content set, and using the consulting knowledge entity content set as a knowledge graph node; S23: constructing a consulting knowledge graph based on the consulting knowledge entity set and the consulting knowledge entity content set; S24: Using the consulting knowledge graph, label and classify each consulting result data in the available consulting database to obtain the consulting semantic label set.

4. The method according to claim 3, characterized in that The consulting semantic retrieval model is constructed in S3, including: S31: Perform keyword screening and annotation on each user's consultation data in the available consultation database to obtain a consultation data keyword annotation set; S32: Performing semantic filtering layer training based on the consulting data keyword annotation set to obtain a keyword filtering annotation network; S33: performing feature mapping annotation on the consultation semantic label database through the consultation data keyword annotation set to obtain a consultation semantic feature label data set; S34: Performing semantic recognition training on the consultation semantic feature label data set using a recurrent neural network structure to construct a consultation semantic recognition network; S35: Connect and merge the keyword filtering and annotation network with the consulting semantic recognition network to generate the consulting semantic retrieval model.

5. The method according to claim 4, characterized in that The consulting data keyword annotation set in S31 includes: Performing word segmentation extraction on each user's consultation data in the available consultation database to obtain a consultation data word segmentation result set; Perform keyword screening on each user consultation data based on the consultation data word segmentation result set to obtain a user consultation data keyword set; The keyword annotation of each user consultation data is performed using the user consultation data keyword set to obtain the consultation data keyword annotation set.

6. The method according to claim 5, characterized in that The obtained user consultation data keyword set includes: Performing frequency statistics on each word segmentation information in the consultation data word segmentation result set to obtain a consultation word segmentation frequency information set; Performing inverse frequency calculation based on the consultation word segmentation frequency information set and the corpus database to obtain an inverse text frequency index set; The product of the consultation word segmentation frequency information set and the inverse text frequency index set is used as the consultation word segmentation key coefficient set; Based on the consultation word segmentation key degree coefficient set, word segmentation screening within a preset key degree threshold is performed to obtain a keyword screening result set, and each user consultation data is marked by the keyword screening result set to obtain the user consultation data keyword set.

7. The method according to claim 1, characterized in that The business consultation optimization management and control through the user consultation interaction feedback data includes: S41: if the user consultation interaction feedback data does not meet the consultation result expectations, the consultation request information and the target consultation retrieval result are used as consultation negative samples; S42: establishing a consultation retrieval penalty mechanism, and calculating a penalty factor for the consultation negative sample based on the consultation retrieval penalty mechanism to obtain negative sample penalty factor information; S43: Optimizing the consulting semantic retrieval model based on the negative sample penalty factor information, generating a consulting semantic optimization retrieval model for business consulting optimization management and control.

8. An intelligent consulting platform combined with demand semantic analysis is characterized by: For implementing the steps of the method according to any one of claims 1 to 7, the platform comprises: A consultation semantic system building module, which is used to collect and obtain a user consultation information database, which includes historical user consultation data and corresponding consultation result data, perform word segmentation preprocessing on the user consultation information database to obtain a usable consultation database, and build a consultation semantic system according to the consultation application scenario; A consultation semantic tag database obtaining module, the consultation semantic tag database obtaining module is used to perform semantic analysis and annotation on the available consultation database through the consultation semantic system to obtain a consultation semantic tag set, and classify and integrate the available consultation database based on the consultation semantic tag set to obtain a consultation semantic tag database; A target consultation retrieval result obtaining module, wherein the target consultation retrieval result obtaining module is used to train the consultation semantic tag database using a recurrent neural network structure, construct a consultation semantic retrieval model, obtain consultation request information of a target user, and perform semantic analysis and retrieval on the consultation request information based on the consultation semantic retrieval model to obtain a target consultation retrieval result; A business consultation optimization management and control module is used to perform business scenario interaction based on the target consultation retrieval results, obtain user consultation interaction feedback data, and perform business consultation optimization management and control through the user consultation interaction feedback data.

Citation Information

Patent Citations

  • Account security checking management method based on artificial intelligence

    CN116402512A

  • Enterprise consultation data management method and system based on cloud computing

    CN117435613A

  • Railway engineering consultation system and consultation method based on BIM and large model

    CN118536950A

  • Internet business-based enterprise affair consultation dialogue analysis method and big data system

    CN118779471A

Cited By

  • Information technology retrieval consultation system and method

    CN120086428A