Intelligent customer service question and answer method based on government purchase

Abstract

The invention provides an intelligent customer service question and answer method for government purchase, and the method comprises the steps: analyzing the consultation question of a user, recognizing the intention of the user, and carrying out the comprehensive analysis and comparison according to a weight proportion difference value, so as to determine the real intention range of the user. The method comprises the steps of user consultation receiving, preliminary intention recognition, intention range reduction, user selection confirmation, knowledge base full-text search, intelligent verbal skill service matching, manual service intervention and the like. Through intelligent analysis and processing, the user demand is quickly and accurately identified, manual intervention is reduced, and the service cost is reduced. Meanwhile, through a user feedback and evaluation mechanism, intelligent service performance is continuously optimized, intelligence, individuation and precision of government purchase service are realized, and service efficiency and user satisfaction are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of customer service question and answer, and in particular to a method for intelligent customer service question and answer based on government procurement. Background Art

[0002] With the rapid development of electronic government procurement business, the number of users related to the government procurement system is growing, the problems encountered by users are increasing, and the service requirements are becoming higher and higher.

[0003] Based on existing technologies, relevant government procurement service personnel are under great work pressure, and are unable to effectively manage service response efficiency, service quality, and service standardization, with risks of human factors. They are unable to promptly resolve government procurement users’ problems 24 hours a day, affecting their business activities. Summary of the invention

[0004] The present invention provides an intelligent and efficient government procurement customer service question and answer service; when government procurement related users have questions and need to consult, the present invention performs semantic analysis on customer questions and identifies the user's intention scope, and conducts guided automatic question and answer through unique intelligent speech service, and finally accurately replies to the user; the present invention can effectively improve the efficiency, quality and standardization of the service; it can handle government procurement users' question consultations 24 hours a day, so that government procurement users' questions can be resolved in a timely manner, and their work can be carried out smoothly.

[0005] The technical solution adopted by the present invention is: a method based on intelligent customer service question and answer of government procurement, comprising the following steps:

[0006] Step 1: The government procurement user enters the relevant questions to be consulted through the government procurement management system terminal service consultation function interface and clicks "Send". The user's consultation questions will be sent to the server, which will parse and respond;

[0007] Step 2: After receiving the consultation questions, the unique intelligent analysis service conducts a comprehensive analysis of the government procurement user's consultation questions, conducts a preliminary analysis of the user's intention scope, and analyzes the weight ratio of the user's relevant intention scope through the unique intelligent analysis service;

[0008] Step 3: Based on the preliminary user intent range identified in step 2, traverse its range, and conduct comprehensive analysis and comparison through the weight ratio difference of each intent, and compare it with the set analysis threshold and weight ratio difference. If it is greater than the threshold, it means that the result does not quite match the user's true intent. If it is less than the threshold, it means that it is relatively consistent with the user's intent range. After traversal and comparison, the user intent range is narrowed down;

[0009] Step 4: Based on the user intentions comprehensively analyzed in step 3, if the number of user intentions is greater than 1, the relevant user intentions will be returned to the customer service interface of the government procurement management system terminal, and the user will select questions that meet his or her actual consultation intentions, and then the questions and answers will be conducted based on the user's selections; if the number of user intentions is equal to 1, the questions and answers will be conducted directly;

[0010] Step 5: Determine the user's intention according to step 4, and conduct a full-text search on this user's intention and the knowledge base maintained in the system. If the full-text search results are matched, it means that the user's intention is relatively clear, and the final matching answer will be sent to the customer service consultation interface of the government procurement management system terminal. If there are multiple situations in the full-text search results, it means that the user's intention is relatively vague, and the corresponding speech will be further matched from the knowledge base. The unique intelligent speech service will conduct guided automatic Q&A based on the configured speech process. After multiple automatic interactive Q&A, the user's intention will be finally determined and the user will be accurately replied. If the final reply fails to solve the user's problem, manual service can be selected, and the manual service will open an instant chat window with the user to provide manual answers.

[0011] Step 6: After the government procurement user consultation is completed, the user can evaluate the service. Based on the user's evaluation, the intelligent service will automatically optimize its own neural network based on the questions and responses in the consultation, enrich the corpus, and ensure that the next consultation will be faster and more accurate.

[0012] As a further improvement of the present invention, the comprehensive analysis in step 2 mainly includes syntactic analysis, semantic analysis, named entity recognition, relationship extraction, and sentiment analysis.

[0013] As a further improvement of the present invention, the specific process of the unique intelligent analysis service in step 2 is as follows:

[0014] S1: The received consultation question is stored in a StringBuffer object, and then the service interface is called, and the StringBuffer object is passed into the interface as a calling parameter;

[0015] S2: Call the NLP interface to perform word segmentation and part-of-speech tagging on the consulting questions;

[0016] S3: Perform syntactic analysis through NLP to identify phrases and dependencies in the question;

[0017] S4: Based on the analysis of S2 and S3, further semantic analysis, named entity recognition, relationship extraction, and sentiment analysis are performed through NLP;

[0018] S5: Based on the comprehensive analysis, further understand the consultation question and analyze the user's initial intention. Calculate the matching degree based on NLP, and store the corresponding matching user intention and weight ratio in the LIST object and return it to the server, which will perform subsequent analysis and processing.

[0019] As a further improvement of the present invention, the specific process of the unique intelligent speech service in step 5 is as follows:

[0020] SS1: According to the user's intention, the corresponding speech model is retrieved from the knowledge base and stored in the LIST object;

[0021] SS2: Extract the speech process data stored in the speech model LIST object, extract the first process node data of the speech according to the speech process configuration, and then assemble it into a list mode based on the node data and reply to the user;

[0022] SS3: The user further selects corresponding question information based on the list information replied by the server, and feeds back the selected question information to the server;

[0023] SS4: Based on the user's selection, the server further extracts the process node data that matches the user's selection from the LIST object, assembles it into a list mode based on the node data again, and replies to the user; and so on, and finally determines the user's intention and responds to the user accurately.

[0024] Beneficial effects of the present invention: (1) Significantly improve service efficiency: Through the intelligent customer service question-and-answer system, the present invention enables government procurement users to quickly obtain accurate responses, greatly shortening the waiting time and improving the efficiency of problem solving; at the same time, the intelligent service system can provide services 24 hours a day, ensuring that users can get timely responses at any time, effectively solving the time limitations of traditional manual customer service;

[0025] (2) Optimizing service quality and standardization: The intelligent customer service system of the present invention is based on big data and natural language processing technology, which can more accurately understand user intentions and provide targeted answers; this not only improves the quality of service, but also ensures the standardization of services through standardized question-and-answer processes, reducing service differences caused by human factors;

[0026] (3) Enhance user experience and satisfaction: The intelligent customer service question-and-answer system of the present invention significantly improves the experience of government procurement users through accurate responses and efficient interactions. Users can solve problems in a short period of time and receive satisfactory services, thereby improving their overall satisfaction and trust in the government procurement system. At the same time, the system also provides a user evaluation function, which continuously optimizes services through user feedback, forming a virtuous circle. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 It is an overall flow chart of a method for intelligent customer service question and answer based on government procurement of the present invention. DETAILED DESCRIPTION

[0028] In order to make the technical problems, technical solutions and beneficial effects to be solved by the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0029] The present invention provides a method based on intelligent customer service question and answer of government procurement, comprising the following steps:

[0030] Step 1: The government procurement user enters the relevant questions to be consulted through the government procurement management system terminal service consultation function interface and clicks "Send". The user's consultation questions will be sent to the server, which will parse and respond;

[0031] Step 2: After receiving the consultation questions, the unique intelligent analysis service conducts a comprehensive analysis of the government procurement user's consultation questions. After comprehensive analysis and preliminary identification of the user's intention range, the unique intelligent analysis service analyzes the weight ratio of the user's relevant intention range. The comprehensive analysis mainly includes syntactic analysis, semantic analysis, named entity recognition, relationship extraction, and sentiment analysis.

[0032] The specific process of its unique intelligent analysis service is as follows: the received consulting questions are stored in the StringBuffer object, and then this service interface is called, and the StringBuffer object is passed into its interface as a calling parameter; the NLP interface is called to segment and annotate the consulting questions; syntactic analysis is performed through NLP to identify the phrases and dependencies in the questions; based on the analysis of S2 and S3, semantic analysis, named entity recognition, relationship extraction, and sentiment analysis are further performed through NLP; based on the comprehensive analysis, the consulting questions are further understood, and the user's initial intention is analyzed, the matching degree is calculated based on NLP, and the corresponding matching user intention and weight ratio are stored in the LIST object and returned to the server, and the server performs subsequent analysis and processing;

[0033] Step 3: Based on the preliminary user intent range identified in step 2, traverse its range, and conduct comprehensive analysis and comparison through the weight ratio difference of each intent, and compare it with the set analysis threshold and weight ratio difference. If it is greater than the threshold, it means that the result does not quite match the user's true intent. If it is less than the threshold, it means that it is relatively consistent with the user's intent range. After traversal and comparison, the user intent range is narrowed down;

[0034] Step 4: Based on the user intentions comprehensively analyzed in step 3, if the number of user intentions is greater than 1, the relevant user intentions will be returned to the customer service interface of the government procurement management system terminal, and the user will select questions that meet his or her actual consultation intentions, and then the questions and answers will be conducted based on the user's selections; if the number of user intentions is equal to 1, the questions and answers will be conducted directly;

[0035] Step 5: Determine the user's intention according to step 4, and conduct a full-text search on this user's intention and the knowledge base maintained in the system. If the full-text search results are matched, it means that the user's intention is relatively clear, and the final matching answer will be sent to the customer service consultation interface of the government procurement management system terminal. If there are multiple situations in the full-text search results, it means that the user's intention is relatively vague, and the corresponding speech will be further matched from the knowledge base. The unique intelligent speech service will conduct guided automatic Q&A based on the configured speech process. After multiple automatic interactive Q&A, the user's intention will be finally determined and the user will be accurately replied. If the final reply fails to solve the user's problem, manual service can be selected, and the manual service will open an instant chat window with the user to provide manual answers.

[0036] The specific process of its unique intelligent speech service is as follows: according to the user's intention, the corresponding speech is retrieved from the knowledge base, and the searched speech model is stored in the LIST object; the speech process data stored in the speech model LIST object is extracted, and according to the speech process configuration, the first process node data of the speech is extracted, and then it is assembled into a list mode based on the node data and replied to the user; the user further selects the corresponding question information based on the list information replied by the server, and feeds back the selected question information to the server; the server further extracts the process node data matching the user's selection from the LIST object based on the user's selection, and then it is assembled into a list mode based on the node data and replied to the user; and so on, and finally the user's intention is determined and the user is replied accurately;

[0037] Step 6: After the government procurement user consultation is completed, the user can evaluate the service. Based on the user's evaluation, the intelligent service will automatically optimize its own neural network based on the questions and responses in the consultation, enrich the corpus, and ensure that the next consultation will be faster and more accurate.

[0038] Embodiment 1:

[0039] Embodiment 1 of the present application provides a Figure 1 The government procurement intelligent customer service question and answer method shown includes the following steps:

[0040] Step 1: The government procurement user enters the question he needs to consult through the government procurement management system terminal service consultation function interface, "I am from Changzhou, and I encounter problems with the planned operation. What should I do?" and clicks Send. The user's consultation question will be sent to the server, which will parse and respond to it;

[0041] Step 2: After the server receives the inquiry question "I am from Changzhou and I have encountered a problem with the plan operation. What should I do?", the unique intelligent analysis service conducts a comprehensive analysis of the government procurement user's inquiry question. After comprehensive analysis and preliminary identification of the user's intention range, it is mainly divided into: (1) plan splitting and merging issues; (2) procurement plan synchronization issues; (3) how to cancel procurement plans; (4) how to apply for temporary plans; (5) view procurement plan review opinions; (6) procurement plan process issues; and the unique intelligent analysis service analyzes the weight ratio of the user's relevant intention range as follows:

[0042] User Intent Weight ratio (1) Issues regarding plan splitting and merging; 25％ (2) Purchase plan synchronization issues; 7％ (3) How to cancel the procurement plan; 9％ (4) How to apply for the temporary plan; 25％ (5) Review the procurement plan review opinions; 4％ (6)Problems with the procurement planning process; 30％

[0043] Step 3: Based on the preliminary user intent range identified in step 2, traverse its range and conduct a comprehensive analysis and comparison through the weight ratio difference of each intent. The analysis threshold is set to (10%) and compared with the weight ratio difference. When it is greater than the threshold, it means that the result does not quite match the user's true intent. When it is less than the threshold, it means that it is relatively consistent with the user's intent range. After traversal and comparison, the user intent range is narrowed. After final analysis, the user intent range is determined as follows: (1) plan splitting and merging issues; (4) how to apply for temporary plans; (6) procurement plan process issues;

[0044] Step 4: Based on the user intentions comprehensively analyzed in step 3, if the number of user intentions is greater than 1, the relevant user intentions are returned to the customer service interface of the government procurement management system terminal, and the user selects questions that meet his or her actual consultation intentions, and then questions and answers are conducted based on the user's selections; if the number of user intentions is equal to 1, questions and answers are conducted directly;

[0045] Step 5: Determine the user's intention based on step 4. For example, if the user selects "How to apply for a temporary plan", a full-text search is performed on this user's intention and the knowledge base maintained in the system. If the full-text search results are matched, the operation steps matching "How to apply for a temporary plan" are replied to the customer service consultation interface of the government procurement management system terminal. If the user selects "Problems with procurement plan process", there are multiple situations in the full-text search results, indicating that the user's intention is relatively vague. The "Procurement plan process" speech will be further matched from the knowledge base, and the user's relevant intentions "(1) Procurement plan submission issue; (2) Procurement plan withdrawal issue; (3) Procurement plan adjustment issue" will be continued through the unique intelligent speech service. If the user selects "Problems with procurement plan submission", the speech model will be matched based on the user's selection. If the intention basis of the selection is still unclear, the user will continue to be replied. If it is clear, the user will be accurately replied with the relevant operation instructions of "Problems with procurement plan submission". If the final reply fails to solve the user's problem, the user can choose manual service, which will open an instant chat window with the user for manual answering.

[0046] Step 6: After the government procurement user consultation is completed, the user can evaluate the service. Based on the user's evaluation, the intelligent service will automatically optimize its own neural network based on the consultation questions and responses, enrich the corpus, and ensure that the next consultation is faster and more accurate.

[0047] Embodiment 2:

[0048] Step 2 of the method for intelligent customer service question and answer in government procurement in Example 1 is specifically as follows:

[0049] S1. The received consultation question is stored in a StringBuffer object, and then the service interface is called, and the StringBuffer object is passed into the interface as a calling parameter;

[0050] S2. Call the NLP interface to perform word segmentation and part-of-speech tagging on the consulting questions;

[0051] S3, perform syntactic analysis through NLP to identify phrases and dependency relationships in the question;

[0052] S4, based on the analysis of S2 and S3, further perform semantic analysis, named entity recognition, relationship extraction, and sentiment analysis through NLP;

[0053] S5. Based on the comprehensive analysis, further understand the consultation question, analyze the user's initial intention, calculate the matching degree based on NLP, and store the corresponding matching user intention and weight ratio in the LIST object and return it to the server, which will perform subsequent analysis and processing;

[0054] Step 5 of the method for intelligent customer service question and answer in government procurement in Example 1 is specifically as follows:

[0055] SS1. According to the user's intention, the corresponding speech model is retrieved from the knowledge base and stored in the LIST object;

[0056] SS2. Extract the speech process data stored in the speech model LIST object, extract the first process node data of the speech according to the speech process configuration, and then assemble it into a list mode based on the node data and reply to the user;

[0057] SS3. The user further selects corresponding question information based on the list information replied by the server, and feeds back the selected question information to the server;

[0058] SS4. Based on the user's selection, the server further extracts the process node data that matches the user's selection from the LIST object, assembles it into a list mode based on the node data again and replies to the user; and so on, and finally determines the user's intention and accurately replies to the user.

[0059] To sum up, the method of intelligent customer service question and answer based on government procurement of the present invention can effectively solve the problems encountered by users in the government procurement process, improve service efficiency and user satisfaction; through intelligent analysis and processing, it can quickly and accurately identify user needs, reduce manual intervention, and reduce service costs; at the same time, through user feedback and evaluation mechanisms, the performance of intelligent services is continuously optimized, so that the system is more in line with the actual needs of users, and the government procurement services are intelligent, personalized and precise.

[0060] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method based on intelligent customer service question and answer for government procurement, characterized in that: The following steps are involved: Step 1: The government procurement user enters the relevant questions to be consulted through the government procurement management system terminal service consultation function interface and clicks "Send". The user's consultation questions will be sent to the server, which will parse and respond; Step 2: After receiving the consultation questions, the unique intelligent analysis service conducts a comprehensive analysis of the government procurement user's consultation questions, conducts a preliminary analysis of the user's intention scope, and analyzes the weight ratio of the user's relevant intention scope through the unique intelligent analysis service; Step 3: Based on the preliminary user intent range identified in step 2, traverse its range, and conduct comprehensive analysis and comparison through the weight ratio difference of each intent, and compare it with the set analysis threshold and weight ratio difference. If it is greater than the threshold, it means that the result does not quite match the user's true intent. If it is less than the threshold, it means that it is relatively consistent with the user's intent range. After traversal and comparison, the user intent range is narrowed down; Step 4: Based on the user intentions comprehensively analyzed in step 3, if the number of user intentions is greater than 1, the relevant user intentions will be returned to the customer service interface of the government procurement management system terminal, and the user will select questions that meet his or her actual consultation intentions, and then the questions and answers will be conducted based on the user's selections; if the number of user intentions is equal to 1, the questions and answers will be conducted directly; Step 5: Determine the user's intention based on step 4, and conduct a full-text search on this user's intention and the knowledge base maintained in the system. If the full-text search results are matched, it means that the user's intention is relatively clear, and the final matching answer will be sent to the customer service consultation interface of the government procurement management system terminal. If there are multiple situations in the full-text search results, it means that the user's intention is relatively vague, and the corresponding speech will be further matched from the knowledge base. The unique intelligent speech service will conduct guided automatic Q&A based on the configured speech process. After multiple automatic interactive Q&A, the user's intention is finally determined and the user is accurately responded to; If the final response fails to solve the user's problem, you can choose manual service, which will open an instant chat window with the user to provide a manual answer; Step 6: After the government procurement user consultation is completed, the user can evaluate the service. Based on the user's evaluation, the intelligent service will automatically optimize its own neural network based on the questions and responses in the consultation, enrich the corpus, and ensure that the next consultation will be faster and more accurate.

2. According to claim 1, a method based on intelligent customer service question and answer for government procurement is characterized in that: The comprehensive analysis in step 2 mainly includes syntactic analysis, semantic analysis, named entity recognition, relationship extraction, and sentiment analysis.

3. According to claim 1, a method based on intelligent customer service question and answer for government procurement is characterized in that: The specific process of the unique intelligent analysis service in step 2 is as follows: S1: The received consultation question is stored in a StringBuffer object, and then the service interface is called, and the StringBuffer object is passed into the interface as a calling parameter; S2: Call the NLP interface to perform word segmentation and part-of-speech tagging on the consulting questions; S3: Perform syntactic analysis through NLP to identify phrases and dependencies in the question; S4: Based on the analysis of S2 and S3, further semantic analysis, named entity recognition, relationship extraction, and sentiment analysis are performed through NLP; S5: Based on the comprehensive analysis, further understand the consultation question and analyze the user's initial intention. Calculate the matching degree based on NLP, and store the corresponding matching user intention and weight ratio in the LIST object and return it to the server, which will perform subsequent analysis and processing.

4. According to claim 1, a method based on intelligent customer service question and answer for government procurement is characterized in that: The specific process of the unique intelligent speech service in step 5 is as follows: SS1: According to the user's intention, the corresponding speech model is retrieved from the knowledge base and stored in the LIST object; SS2: Extract the speech process data stored in the speech model LIST object, extract the first process node data of the speech according to the speech process configuration, and then assemble it into a list mode based on the node data and reply to the user; SS3: The user further selects corresponding question information based on the list information replied by the server, and feeds back the selected question information to the server; SS4: Based on the user's selection, the server further extracts the process node data that matches the user's selection from the LIST object, assembles it into a list mode based on the node data again, and replies to the user; and so on, and finally determines the user's intention and responds to the user accurately.

Images