CN-122025027-A - Psychological consultation system and method based on artificial intelligence

CN122025027ACN 122025027 ACN122025027 ACN 122025027ACN-122025027-A

Abstract

The invention belongs to the technical field of mental health service, and particularly relates to a mental consultation system based on artificial intelligence, which comprises the steps that after text information is received by a user input interface, an emotion recognition module generates an emotion state through an emotion pattern library and a keyword matching algorithm; the intelligent knowledge base matching module divides the consulting stage according to the memory abstract and matches the professional answer, the crisis intervention recognition module recognizes the potential crisis through the crisis keywords and generates intervention response, the multi-mode response generation module fuses the multi-dimensional data to generate response content, and finally the AI response output module outputs the result according to the language mode. The invention can solve the problem that the psychological consultation technology in the prior art has defects in the aspects of resource coverage, response efficiency and personalized service.

Inventors

LIU ZHIYONG
Fu Ruqin
ZHANG NENGJIN
Xiong Weirui
HUANG TING

Assignees

重庆九五医院有限公司
重庆师范大学

Dates

Publication Date: 20260512
Application Date: 20260130

Claims (10)

1. The psychological consultation system based on artificial intelligence is characterized by comprising a user input interface, an emotion recognition module, a context memory enhancement module, a knowledge base intelligent matching module, a crisis intervention recognition module, a multi-modal response generation module and an AI response output module, wherein: The user input interface is used for receiving text information input by a user; the emotion recognition module is used for recognizing emotion types and intensity levels in texts input by users based on a preset emotion pattern library and a keyword matching algorithm, and generating emotion states; the context memory enhancement module is used for constructing a user memory data structure, recording and updating user keywords, topics and emotion history information, and generating a memory abstract; The knowledge base intelligent matching module is used for dividing a plurality of stages according to the multidimensional factors of the memory abstract, and matching answers from a preset psychological expertise knowledge base after each stage generates a corresponding communication strategy to generate a communication result; The crisis intervention recognition module is used for recognizing potential crisis tendency of the user from the memory abstract through preset crisis keywords and generating intervention response; The multi-mode response generation module is used for combining the emotion states, the memory abstract and the communication results and calling AI to generate response contents; The AI response output module is used for outputting final AI response and intervention response according to the built-in language mode; The knowledge base intelligent matching module comprises a multi-dimensional stage judging unit, wherein: The multi-dimensional stage judging unit is used for matching the dialogue turn, the emotion state, the topic depth and the user participation degree based on the memory abstract with preset multi-stages, wherein the preset multi-stages comprise an initial stage, a exploring stage, a supporting stage and an ending stage.
2. The artificial intelligence based psychological consulting system of claim 1, wherein said emotion recognition module comprises a text preprocessing unit, an emotion pattern library unit and a keyword matching unit, wherein: the text preprocessing unit is used for converting text information input by a user into a character string format and preprocessing the text information to generate preprocessed text data; The emotion pattern library unit establishes an emotion pattern association index according to preset emotion types, keywords and intensity levels, extracts corresponding data from a preset emotion data pool based on the emotion pattern association index and stores the data to generate an emotion pattern library; The keyword matching unit is used for traversing the emotion pattern library, matching emotion keywords in the preprocessed text data, outputting corresponding emotion types and intensity levels, and representing the emotion states of the user; The emotional states are divided into a plurality of emotional grades from low to high according to the positive degree and the negative degree of the emotion, and the lower the emotional grade is, the lower the emotional positive degree is.
3. The artificial intelligence based psychological consulting system of claim 2, wherein the contextual memory enhancement module includes a user memory data storage unit and a memory digest generation unit, wherein: The user memory data storage unit is used for storing a user name, a keyword list, a topic list, an emotion history record, access time and access times and generating a user memory data structure; the memory digest generation unit is used for generating a memory digest containing access rounds, user names, user participation, topic depths and emotional states according to the user memory data structure.
4. The psychological consulting system based on artificial intelligence of claim 3, wherein the multi-dimensional stage judging unit is configured to match the dialogue turns, the emotion states, the topic depths and the user participation based on the memory abstract with the preset multi-stage specifically: The method comprises the steps of obtaining a dialogue round, judging the dialogue round and a preset dialogue round threshold switched in each stage, and generating a corresponding stage; Acquiring and judging an emotion state, and entering a supporting stage in advance if the emotion type and the intensity level in the emotion state reach a negative emotion level; obtaining topic depth in the memory abstract, and judging that the topic depth is larger than a preset threshold value and the dialogue turns meet the dialogue turn threshold value of switching in the exploration phase, then directly entering the exploration phase; and acquiring the user participation degree in the memory abstract, and if the user participation degree is lower than a preset user participation threshold value, directly returning to the initial stage.
5. The psychological consultation system based on artificial intelligence of claim 4, wherein the knowledge base intelligent matching module further comprises a communication result generating unit, the communication result generating unit generates a corresponding communication strategy according to the stage result generated by the multi-dimensional stage judging unit and by combining the preprocessed text data, matches the communication strategy with entries of a preset psychological expertise knowledge base, sorts the entries in descending order according to the matching score, and returns a preset number of answers to generate a communication result.
6. The artificial intelligence based psychological consulting system of claim 1, wherein the crisis intervention identification module includes a crisis keyword storage unit, a crisis detection unit, and a crisis intervention response generation unit, wherein: The crisis keyword storage unit is used for storing preset crisis keywords and converting the crisis keywords into crisis keyword regular expressions; The crisis detection unit is used for matching and detecting whether crisis keywords exist in the preprocessed text data or not according to the crisis keyword regular expression; the crisis intervention response generation unit is used for generating a crisis intervention response after the crisis keyword is detected.
7. The artificial intelligence based psychological consulting system of claim 6, wherein said multimodal response generating module includes a dynamic weight distribution unit, a multimodal fusion unit, and a response optimization unit, wherein: The dynamic weight distribution unit calculates the answer weight of the knowledge base and the AI to generate the answer weight according to the confidence level of the answer of the knowledge base in the communication result and the dialogue stage, the emotion state and the user intention in the memory abstract; The multi-mode fusion unit is used for selecting a single-mode or a multi-mode fusion mode according to the answer weight of the knowledge base and the answer weight generated by the AI, selecting a fusion strategy according to a dialogue stage to generate a fusion response, and outputting one of the answers of the knowledge base or the answers generated by the AI or a response result of fusing the knowledge base file and the answers generated by the AI; the response optimizing unit is used for optimizing the response result according to a preset quality evaluation method.
8. The psychological consulting system based on artificial intelligence of claim 7, wherein the dynamic weight distribution unit is configured to calculate the knowledge base answer weight and the AI generated answer weight according to the knowledge base answer confidence and dialogue phase in the communication result, the emotion state and the user intention in the memory abstract, specifically: Calculating answer confidence coefficient of a psychological expertise base in the communication result, and presetting a first confidence coefficient threshold value and a second confidence coefficient threshold value, wherein the first confidence coefficient threshold value is larger than the second confidence coefficient threshold value; If the knowledge base answer confidence is higher than the first confidence threshold, setting the knowledge base answer weight higher than the AI generation answer weight; if the knowledge base answer confidence coefficient is between the first confidence coefficient threshold value and the second confidence coefficient threshold value, setting the knowledge base answer weight equal to the AI generation answer weight, and starting a fusion mode; if the answer confidence of the knowledge base is lower than the second confidence threshold, setting AI generation answer weight higher than the answer weight of the knowledge base; in the support stage, increasing AI generation answer weight; When the user intention is judged to be information seeking, the answer weight of the knowledge base is increased, and when the user intention is judged to be emotion support, the AI generation answer weight is increased.
9. The psychological consulting system based on artificial intelligence of claim 8, wherein in the multimodal fusion unit, a single modality mode or a multimodal fusion mode is selected according to the knowledge base answer weight and the AI generation answer weight, and a fusion response is generated according to a fusion strategy selected in a dialogue stage, and the response result of outputting one of the knowledge base answers or the AI generation answers or fusing the knowledge base file and the AI generation answer is specifically: Presetting a first weight threshold and a second weight threshold, wherein the first weight threshold is larger than the second weight threshold; When the answer weight of the knowledge base is higher than a first weight threshold and the answer weight generated by AI is higher than a second weight threshold, selecting a fusion mode of the answer preposition of the knowledge base and the answer supplement generated by AI; When the AI generation answer weight is higher than the first weight threshold and the knowledge base answer weight is higher than the second weight threshold, selecting a fusion mode of the AI generation answer preposition and the knowledge base answer supplement; if the answer weight of the knowledge base and the answer weight generated by the AI are both between the first weight threshold and the second weight threshold, or lower than the second weight threshold, or higher than the first weight threshold, selecting a fusion mode of the answer of the knowledge base and the answer generated by the AI; According to the dialogue stage, the knowledge base answer is selected as a dominant in the initial stage and the ending stage, and the AI generation answer is selected as a dominant in the exploration stage and the supporting stage.
10. An artificial intelligence based psychological consultation method applied to the artificial intelligence based psychological consultation system according to any one of the claims 1-9 and characterized by comprising: s1, acquiring text information input by a user; s2, identifying emotion types and intensity levels in texts input by users based on a preset emotion pattern library and a keyword matching algorithm, and generating emotion states; s3, constructing a user memory data structure, recording and updating user keywords, topics and emotion history information, and generating a memory abstract; S4, dividing a plurality of stages according to multidimensional factors of the memory abstract, and after each stage generates a corresponding communication strategy, matching answers from a preset psychological expertise knowledge base to generate a communication result; S5, identifying potential crisis tendency of the user from the memory abstract through preset crisis keywords, and generating an intervention response; S6, calling AI to generate response content by combining the emotion state, the memory abstract and the communication result; and S7, outputting a final AI response and an intervention response according to the built-in language mode.

Description

Psychological consultation system and method based on artificial intelligence Technical Field The invention belongs to the technical field of mental health services, and particularly relates to a mental consultation system and method based on artificial intelligence. Background Along with the rapid development of social economy, the life rhythm of people is continuously accelerated, the pressure of work and life is continuously increased, the incidence rate of psychological health problems is increased year by year, and the requirements for professional psychological consultation services are increased in an explosive manner. However, the existing psychological consultation service system and related technical solutions have a plurality of technical defects to be solved urgently, and the technical defects are as follows: 1. The problem of unbalanced resource allocation is remarkable in that the culture period of professional psychological consultants is long, the quantity is scarce, high-quality resources are excessively concentrated in a core city, so that large-scale and instant consultation demands cannot be met to form supply-demand contradiction, and meanwhile, in remote areas, due to lack of the professional psychological consultation resources, the service coverage is limited, and the equalization supply of psychological consultation services is difficult to realize. 2. The service response timeliness is insufficient, the traditional psychological consultation mode depends on off-line reservation or fixed period communication, a user needs to plan time in advance, 24-hour all-weather instant response cannot be realized, and timely and effective support is difficult to provide for sudden emotion crisis or instant psychological grooming demands. 3. The psychological consultation process relates to personal privacy information and emotion privacy of the user, and partial users do not want to complain to human consultants about the mind of real ideas due to worry about information disclosure or social bias, so that the consultation process is not deep and the consultation effect is affected. 4. The existing AI consultation system has low intelligent degree, wherein the AI consultation related products in the current market mostly adopt a fixed template reply mechanism, lack dynamic perceptibility of emotional states of users, cannot accurately identify complex emotion and intensity changes such as anxiety, depression, anger and the like, and simultaneously lack a contextual memory function, cannot correlate historical dialogue information of the users, so that poor dialogue consistency is caused, personalized consultation service is difficult to provide, and deep trust relationship cannot be established. Therefore, the existing psychological consultation mode and related technical schemes have obvious technical defects in aspects of resource coverage, response efficiency, privacy protection, personalized service, crisis intervention and the like, and the requirements of the current user on high-quality, convenient and personalized psychological consultation service are difficult to meet. Disclosure of Invention The technical problem solved by the invention is to provide a psychological consultation system and a psychological consultation method based on artificial intelligence, so as to solve the problem that the psychological consultation technology in the prior art has defects in terms of resource coverage, response efficiency and personalized service. The basic scheme provided by the invention is as follows: the psychological consultation system based on artificial intelligence comprises a user input interface, an emotion recognition module, a context memory enhancement module, a knowledge base intelligent matching module, a crisis intervention recognition module, a multi-modal response generation module and an AI response output module, wherein: The user input interface is used for receiving text information input by a user; the emotion recognition module is used for recognizing emotion types and intensity levels in texts input by users based on a preset emotion pattern library and a keyword matching algorithm, and generating emotion states; the context memory enhancement module is used for constructing a user memory data structure, recording and updating user keywords, topics and emotion history information, and generating a memory abstract; The knowledge base intelligent matching module is used for dividing a plurality of stages according to the multidimensional factors of the memory abstract, and matching answers from a preset psychological expertise knowledge base after each stage generates a corresponding communication strategy to generate a communication result; The crisis intervention recognition module is used for recognizing potential crisis tendency of the user from the memory abstract through preset crisis keywords and generating intervention response; The multi-mode response generation mo