Psychological counseling intelligent assistant system and service method based on large model

Abstract

The invention discloses a psychological counseling intelligent assistant system based on a large model and a service method. According to the method, communication and butt joint are carried out through the digital-intelligent branch and the user, after communication is finished, early warning vocabularies or phrases appearing in communication content of the user are recognized and analyzed through the digital-intelligent branch monitoring engine, and when it is recognized that the user has negative emotion, corresponding early warning information is sent to the expert workbench in time, so that the user is prevented from suffering from the negative emotion. And the psychological consultation experts can conveniently intervene and process in time. Meanwhile, the psychological consultation expert can request the consultation scheme through the expert workbench, and the big model is combined with the therapy-related data in the local knowledge base, the user portrait and the user communication record, so that the personalized consultation scheme is provided for the user, and the efficiency of the psychological consultation expert in performing interventional consultation on the user is greatly improved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology and machine learning algorithm technology, and in particular to a psychological counseling intelligent assistant system and service method based on a large model. Background Art

[0002] In today's society, people are paying more and more attention to their personal mental health. In the existing technology, it is generally done by self-testing through psychological test scales, or seeking consultation from professional psychological counselors. Psychological counselors use psychological methods and theories to help individuals discover their own problems, improve self-cognition, and find solutions to problems by establishing trust and cooperation with individuals. In the above two methods, due to the lack of personal professional ability, the self-test results are prone to deviations, while consulting and answering questions through professionals, although their professional ability is strong, the analysis process is too dependent on manual work. Limited by the lack of professionals, psychological monitoring is difficult to promote on a large scale. Therefore, with the development of artificial intelligence, psychological counseling platforms have also emerged. Users enter the questions they need to consult in the platform and communicate with artificial intelligence, or directly communicate with artificial intelligence through voice and video. In the process of talking with individuals, the artificial intelligence system will analyze the individual's situation through a large model, ask corresponding questions, and give corresponding feedback. However, only through artificial intelligence, it is impossible to give more appropriate feedback according to different individual situations like professional psychological counselors. In addition, the psychological counseling agents based on large models on the market are just pure chat robots and cannot be combined with offline psychological counselor resources. If a professional psychological counselor is needed to interact, the counselor needs to read the previous chat content one by one, which does not significantly improve the efficiency of the expert side.

[0003] Therefore, the prior art has defects and needs to be improved. Summary of the invention

[0004] The purpose of the present invention is to overcome the deficiencies of the prior art and provide a large model-based psychological counseling intelligent assistant system and service method.

[0005] The technical solution of the present invention is as follows: Provide a psychological consultation intelligent assistant system based on a large model, including: an interaction layer, a service layer and a basic layer, wherein:

[0006] Interaction layer: includes a digital avatar for communicating with users and an expert workbench for serving psychological counseling experts;

[0007] Service layer: including the digital twin service engine, digital twin monitoring engine, user portrait generation engine, and consulting solution recommendation engine;

[0008] Basic layer: includes basic large models, psychological counseling professional field datasets, and fine-tuned large models.

[0009] Furthermore, in the interaction layer:

[0010] The digital twin is used to communicate with users, send the user's communication records to the service layer for keyword monitoring, and reflect the information fed back by the service layer in the communication with the user;

[0011] The expert workbench includes four sub-modules: monitoring panel, user portrait, service takeover, and consulting plan. The user's emotions are analyzed at the basic level. When the service level determines that the user's emotions trigger the warning rules, the user's situation is reflected at the expert workbench and the psychological counseling experts are notified to intervene.

[0012] Furthermore, in the service layer:

[0013] The digital twin service engine is used to receive and send real-time data with the digital twin, receive the user's communication records and return questions or communication words to the user;

[0014] The digital twin monitoring engine is used to query the user's latest communication records and warning words during the communication process. It requests the basic layer to identify and analyze the user's real-time emotional state and emotional changes through the prompt word engineering. When the basic layer reports that the user's emotions trigger the warning rules, it notifies the psychological counseling experts to intervene.

[0015] After the user portrait generation engine analyzes the user's communication records and warning words in the basic layer digital intelligence twin monitoring engine, it will periodically calculate the duration of negative emotions and the statistics of emotion sources generated by the basic layer's emotional trend analysis and emotional source analysis, and save the analysis result data;

[0016] The counseling solution recommendation engine generates personalized counseling solutions based on user portraits and the latest communication records, combined with specific psychotherapy operation steps, through the prompt word engineering request base layer.

[0017] The present invention also provides a large model-based psychological counseling intelligent assistant service method, which uses the large model-based psychological counseling intelligent assistant system as described above to perform data processing, and includes the following steps:

[0018] Step 1: The user requests a consultation, the Digital Avatar Service Engine starts and establishes a dialogue, receives the user's consultation, and saves the user's communication records in the database;

[0019] Step 2: Based on the user's inquiry content, request feedback from the big model in the base layer, and reply to the user's inquiry based on the feedback content of the big model;

[0020] Step 3: After completing the reply, the Digital Clone Monitoring Engine analyzes the user's emotions during the consultation process to determine whether the user's emotions trigger warning rules.

[0021] Step 4: If the user's emotions trigger the warning rules, the user's emotional status is analyzed and the analysis results are sent to the expert workstation for psychological counseling experts to use as a reference for judgment;

[0022] Step 5: Generate a user profile based on the user's emotional status. Based on the user profile and the latest communication records, combined with specific psychological therapy operation steps, generate a personalized psychological counseling plan through the prompt word engineering request base layer;

[0023] Step 6: Recommend the personalized psychological counseling plan feedback to the psychological counseling experts, who will provide consultation and guidance to the users.

[0024] Furthermore, the specific steps of step 3 are:

[0025] Step 3.1: The Digital Avatar Service Engine forwards the user communication records to the Digital Avatar Monitoring Engine;

[0026] Step 3.2: The digital twin monitoring engine uses a large model to identify the user's communication records, determine and extract warning words or phrases in the communication records;

[0027] Step 3.3: After the big model recognizes the communication records, it organizes the extracted warning words or phrases into warning information and returns the corresponding warning information to the digital twin monitoring engine;

[0028] Step 3.3: The digital clone monitoring engine receives the warning information issued by the large model, and issues warnings for the extremely negative emotions in the user's current communication records and the duration of the user's negative emotions over a period of time, and sends the warning information to the expert workstation.

[0029] Furthermore, the warning process of extremely negative emotions is as follows:

[0030] Step a.1: The digital twin monitoring engine receives user communication records;

[0031] Step a.2: Combine the warning word library to assemble relevant prompt words;

[0032] Step a.3: Analyze the user's communication records based on the prompt words, and identify the warning words or phrases in the consultation message raised in the user's communication records;

[0033] Step a.4: If there are no identical or similar warning words or phrases in the communication record, the warning process ends;

[0034] Step a.5: If the same or similar warning words or phrases exist in the communication record, a warning is sent to the expert workbench;

[0035] Step a.6: After receiving the warning, the expert workstation notifies the psychological counseling expert to handle the warning;

[0036] Step a.7: Determine whether the psychological counseling expert has completed processing the warning. If not processed within the time threshold of the set warning timer, return to step a.5 and initiate a warning to the expert workstation again; if the psychological counseling expert has completed processing the warning, end the warning process.

[0037] Furthermore, the specific process of the negative emotion duration is as follows:

[0038] Step b.1: The Digital Intelligence Avatar monitoring engine polls the user;

[0039] Step b.2: When a user has a negative emotion warning record, the frequency and duration of the user's negative emotion warning are identified and analyzed to determine whether the duration of the user's negative emotion exceeds the set time value;

[0040] Step b.3: If it is determined that the duration of the user's negative emotion does not exceed the set time value, the process ends;

[0041] Step b.4: If it is determined that the user's negative emotions last longer than the set time, a warning message is generated based on the duration of the user's negative emotions and sent to the expert workstation, which is received and processed by the psychological counseling expert.

[0042] Furthermore, the specific steps of generating the user portrait in step 5 are:

[0043] Step c.1: After the user's consultation process is completed, the Digital Avatar Service Engine requests the User Portrait Generation Engine to analyze the emotional information of this round of consultation;

[0044] Step c.2: The user portrait generation engine analyzes the user's emotional information through the large model and saves the analysis results in the database;

[0045] Step c.3: The user portrait generation engine regularly polls the emotion analysis results of the target users in the database, and calculates and counts the user's daily emotion information;

[0046] Step c.4: When the psychological counseling expert requests a user portrait through the expert workbench, the user portrait generation engine queries the user's latest communication records from the database and requests the big model to analyze the user's latest emotional information in combination with the prompt words;

[0047] Step c.5: Based on the latest sentiment information, combine the most recent sentiment analysis results of the user queried by the user portrait generation engine to generate a user portrait.

[0048] Furthermore, the specific steps of generating and recommending the personalized psychological counseling plan are as follows:

[0049] Step d.1: The psychological counseling expert requests a counseling solution from the counseling solution recommendation engine;

[0050] Step d.2: The consultation solution recommendation engine queries the user's consultation communication records from the database, and queries the user's sentiment analysis data in the user portrait generation engine;

[0051] Step d.3: Based on the query results, retrieve therapy-related information from the local knowledge base;

[0052] Step d.4: Combine the communication records, user portraits, and therapy-related information through prompt words, and request the big model to recommend a consultation plan;

[0053] Step d.5: Review the counseling plan and then return the reviewed counseling plan to the psychological counseling expert.

[0054] By adopting the above scheme, the present invention communicates with the user through the digital avatar. When the communication is over, the digital avatar monitoring engine identifies and analyzes the warning words or phrases that appear in the user's communication content. When it is identified that the user has negative emotions, the corresponding warning information is sent to the expert workbench in time, so that the psychological counseling experts can intervene in time. At the same time, the psychological counseling experts can request a consultation plan through the expert workbench. The large model combines the therapy-related information in the local knowledge base, the user portrait, and the user communication record to provide the user with a personalized consultation plan, which greatly improves the efficiency of the psychological counseling experts in intervening in the consultation with the user. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 This is a functional architecture diagram of the platform of the present invention.

[0056] Figure 2 This is a schematic diagram of the display interface of the Digital Clone monitoring panel.

[0057] Figure 3 Displays a schematic diagram of the interface for the user portrait.

[0058] Figure 4 This is a schematic diagram of the display interface for intelligent recommendation of consulting solutions. DETAILED DESCRIPTION

[0059] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments.

[0060] See also Figure 1 The present invention provides a psychological consultation intelligent assistant system based on a large model, comprising: an interaction layer, a service layer and a basic layer, wherein:

[0061] Interaction layer: includes digital avatars for communicating with users, and expert workstations for serving psychological counseling experts.

[0062] The digital clone is used to communicate with users, send the user's communication records to the service layer for keyword monitoring, and reflect the information fed back by the service layer in the communication with the user.

[0063] The digital avatar is a highly anthropomorphic virtual digital image that is realized through technologies such as motion capture, three-dimensional modeling, and speech synthesis. It can highly restore the image and facial expressions of real humans, and can even synchronize lip movements with sounds in real time, and has the ability to express emotions and communicate.

[0064] The expert workbench includes four sub-modules: monitoring panel, user portrait, service takeover, and consulting plan. The user's emotions are analyzed at the basic level. When the service level determines that the user's emotions trigger the warning rules, the user's situation is reflected at the expert workbench and the psychological counseling experts are notified to intervene.

[0065] Service layer: includes the digital twin service engine, digital twin monitoring engine, user portrait generation engine, and consulting solution recommendation engine.

[0066] The digital twin service engine is used to receive and send real-time data with the digital twin, receive the user's communication records and return questions or communication words to the user.

[0067] The digital clone monitoring engine is used to query the user's latest communication records and warning words during the communication process. It requests the basic layer to identify and analyze the user's real-time emotional state and emotional changes through the prompt word engineering. When the basic layer feedback shows that the user's emotions trigger the warning rules, it notifies the psychological counseling experts to intervene.

[0068] After the user portrait generation engine analyzes the user's communication records and warning words in the basic layer digital intelligent clone monitoring engine, it will perform the emotional trend analysis and emotional source analysis generated by the basic layer to calculate the duration of negative emotions and the statistics of emotional sources, and save the analysis result data.

[0069] The counseling solution recommendation engine generates personalized counseling solutions based on user portraits and the latest communication records, combined with specific psychotherapy operation steps, through the prompt word engineering request base layer.

[0070] Basic layer: includes basic large models, psychological counseling professional field datasets, and fine-tuned large models.

[0071] The basic large model includes a large model for identifying and analyzing warning words or phrases in the user's communication records. It can judge the warning words through the corresponding prompt words, and analyze the user's emotional state and emotional changes during the communication process. In this way, if the digital clone monitoring engine detects that there is a problem with the user's emotions, it can send corresponding warning information to the psychological counseling expert and request the psychological counseling expert to intervene.

[0072] Communicate with users through digital avatars, accept users' consultations, and monitor users' emotional states. When users have negative emotions during the communication process, the service layer will feedback the negative emotions to the expert workbench and request psychological counseling experts to intervene. At the same time, it will provide psychological counseling experts with corresponding user portraits and recommend corresponding consultation plans, so that psychological counseling experts can quickly understand the problems that users need to consult and the users' emotional states, and communicate with users and provide psychological counseling. By combining online and offline methods, the big model recommends personalized consultation plans, which greatly improves the efficiency of psychological counseling experts' understanding of users and service consultation.

[0073] The present invention also provides a large model-based psychological counseling intelligent assistant service method, which uses the large model-based psychological counseling intelligent assistant system as described above to perform data processing, and includes the following steps:

[0074] Step 1: The user requests consultation, the Digital Avatar Service Engine starts and establishes a dialogue, receives the user's consultation, and saves the user's communication records in the database.

[0075] Step 2: Based on the user's inquiry content, request feedback from the big model of the base layer, and reply to the user's inquiry based on the feedback content of the big model.

[0076] Step 3: After completing the reply, the Digital Avatar monitoring engine analyzes the user's emotions during the consultation process to determine whether the user's emotions trigger the warning rules. Figure 2 The specific steps are:

[0077] Step 3.1: The Digital Twin Service Engine forwards the user communication records to the Digital Twin Monitoring Engine.

[0078] Step 3.2: The digital twin monitoring engine uses a large model to identify the user's communication records, determine and extract warning words or phrases in the communication records.

[0079] Step 3.3: After the big model identifies the communication records, it organizes the extracted warning words or phrases into warning information and returns the corresponding warning information to the digital clone monitoring engine.

[0080] Step 3.3: The digital clone monitoring engine receives the warning information issued by the large model, and issues warnings for the extremely negative emotions in the user's current communication records and the duration of the user's negative emotions over a period of time, and sends the warning information to the expert workstation.

[0081] The digital twin monitoring engine queries the user's latest communication records and warning words during the communication process, and uses the prompt word engineering to request a large model to analyze and identify the user's real-time emotional state and emotional changes. When the user's emotions trigger the warning rules, the digital twin monitoring engine will notify the psychological counseling expert to intervene.

[0082] The warning process of very negative emotions is as follows:

[0083] Step a.1: The digital twin monitoring engine receives user communication records;

[0084] Step a.2: Combine the warning word library to assemble relevant prompt words;

[0085] Step a.3: Analyze the user's communication records based on the prompt words, and identify the warning words or phrases in the consultation message raised in the user's communication records;

[0086] Step a.4: If there are no identical or similar warning words or phrases in the communication record, the warning process ends;

[0087] Step a.5: If the same or similar warning words or phrases exist in the communication record, a warning is sent to the expert workbench;

[0088] Step a.6: After receiving the warning, the expert workstation notifies the psychological counseling expert to handle the warning;

[0089] Step a.7: Determine whether the psychological counseling expert has completed processing the warning. If not processed within the time threshold of the set warning timer, return to step a.5 and initiate a warning to the expert workstation again; if the psychological counseling expert has completed processing the warning, end the warning process.

[0090] Specifically, when the digital clone service engine receives the user communication record, it immediately forwards it to the digital clone monitoring engine, combines the local warning vocabulary and the user consultation message through the prompt word engineering, and then requests the large model to identify whether the user communication record hits the warning vocabulary or is semantically similar to the warning vocabulary. If so, an alert is initiated to the expert.

[0091] The specific process of the negative emotion duration is as follows:

[0092] Step b.1: The Digital Intelligence Avatar monitoring engine polls the user;

[0093] Step b.2: When a user has a negative emotion warning record, the frequency and duration of the user's negative emotion warning are identified and analyzed to determine whether the duration of the user's negative emotion exceeds the set time value;

[0094] Step b.3: If it is determined that the duration of the user's negative emotion does not exceed the set time value,

[0095] Then the process ends;

[0096] Step b.4: If it is determined that the user's negative emotions last longer than the set time, a warning message is generated based on the duration of the user's negative emotions and sent to the expert workstation, which is received and processed by the psychological counseling expert.

[0097] Specifically, the digital clone monitoring engine regularly polls users who have expressed negative emotions every day to inquire about how long the user's negative emotions last. If the user's negative emotions last for more than a certain number of days, in general, if the user's negative emotions last for more than 14 days, a corresponding warning will be issued to the psychological counseling expert, requesting the psychological counseling expert to intervene in a timely manner.

[0098] Step 4: If the user's emotions trigger the warning rules, the user's emotional state is analyzed and the analysis results are sent to the expert workstation for psychological counseling experts to use as a reference for judgment.

[0099] Step 5: Generate a user profile based on the user's emotional status. Based on the user profile and the latest communication records, combined with specific psychotherapy operation steps, generate a personalized psychological counseling plan through the prompt word engineering request base layer.

[0100] After the user consultation is over, the user sentiment analysis engine is notified. After reading the user's latest consultation records and warning keywords and other data, the user sentiment analysis engine uses big data to perform sentiment trend analysis, sentiment source analysis, etc., and regularly calculates the duration of negative emotions and statistics on sentiment sources, and finally saves the analysis result data. When the psychological counseling expert queries the user portrait, the historical sentiment analysis data is combined with the sentiment analysis data of the latest session to automatically generate the latest user portrait.

[0101] The specific steps to generate a user portrait are:

[0102] Step c.1: After the user's consultation process is completed, the Digital Avatar Service Engine requests the User Portrait Generation Engine to analyze the emotional information of this round of consultation;

[0103] Step c.2: The user portrait generation engine analyzes the user's emotional information through a large model.

[0104] And save the analysis results in the database;

[0105] Step c.3: The user portrait generation engine regularly polls the emotion analysis results of the target users in the database, and calculates and counts the user's daily emotion information;

[0106] Step c.4: When the psychological counseling expert requests a user portrait through the expert workbench, the user portrait generation engine queries the user's latest communication records from the database and requests the big model to analyze the user's latest emotional information in combination with the prompt words;

[0107] Step c.5: Based on the latest sentiment information, combine the most recent sentiment analysis results of the user queried by the user portrait generation engine to generate a user portrait.

[0108] User consultation sentiment analysis: When the user ends the consultation or switches to the digital twin, the digital twin service engine initiates a user sentiment analysis request to the user portrait generation engine. The user portrait generation engine uses the prompt word engineering, combined with the user consultation information record and sentiment analysis prompt, to request the big model to analyze the user sentiment information of this round of consultation and save the results in the database.

[0109] Calculation and statistics of user daily emotional information: At 0:00 every day, the user portrait generation engine queries the emotional information of each round of user consultation in the previous day, calculates the average emotional index / polarity, accumulates the frequency of each emotional source, calculates the duration of the user's negative emotions, and saves the results in the database.

[0110] User portrait generation: When an expert queries a user portrait, the user portrait generation engine queries the user's latest communication records, and combines the prompt word to request the big model to analyze the user's latest emotional information. Then, the user portrait generation engine combines the user's cleaning analysis results of the last 10 days to generate the latest user portrait. Please refer to Figure 3 .

[0111] The specific steps of generating and recommending the personalized psychological counseling program are as follows:

[0112] Step d.1: The psychological counseling expert requests a counseling solution from the counseling solution recommendation engine;

[0113] Step d.2: The consulting solution recommendation engine queries the user's consulting communication records from the database.

[0114] And query the user's sentiment analysis data in the user portrait generation engine;

[0115] Step d.3: Based on the query results, retrieve therapy-related information from the local knowledge base;

[0116] Step d.4: Combine the communication records, user portraits, and therapy-related information through prompt words, and request the big model to recommend a consultation plan;

[0117] Step d.5: Review the counseling plan and return the reviewed counseling plan to the counseling expert. Figure 4 .

[0118] When a psychological counseling expert requests a recommended counseling plan through the expert workbench, the counseling plan recommendation engine queries the user's latest communication records and user portraits, and then recalls relevant information on "cognitive behavioral therapy" or other therapies in the local knowledge base based on the query results. It combines the communication records with the user portraits and therapy-related information through the prompt word engineering, and requests the big model to recommend a personalized counseling plan, including questions and recommended words for each step. The recommended plan is output to the expert workbench after passing the system content review.

[0119] Step 6: Recommend the personalized psychological counseling plan feedback to the psychological counseling experts, who will provide consultation and guidance to the users.

[0120] When psychological counseling experts provide consultation and guidance to users, user portraits can be used to identify the user's negative emotional state during the current consultation, as well as whether the user's negative emotional state has lasted for more than 14 days, so as to quickly understand the user's psychological condition and save the psychological counseling expert's time in understanding the consultation content. In addition, the platform can provide personalized consultation plans for the user, thereby effectively improving the efficiency of psychological counseling experts in providing psychological consultation and guidance to users.

[0121] When a user logs in to the platform for consultation, the digital avatar communicates with the user, and the digital avatar service engine provides the digital avatar with the corresponding communication plan, so as to provide consultation and communication for the user. When the digital avatar service engine receives the user's communication record, it immediately forwards it to the digital avatar monitoring engine, and queries the user's latest communication record and the warning words in the communication process through the digital avatar monitoring engine. The local warning word library and the user consultation message are combined through the prompt word project, and then the big model is requested to identify whether the user's communication record hits the warning word or is semantically similar to the warning word. If so, an alert is issued to the expert. After the user consultation is completed, the user sentiment analysis engine is notified. After the user's sentiment analysis engine reads the user's latest consultation record and warning keywords and other data, it uses big data to perform sentiment trend analysis, sentiment source analysis, etc., and regularly calculates the duration of negative emotions and statistics on sentiment sources, and finally saves the analysis result data. When the psychological counseling expert queries the user portrait, the historical sentiment analysis data is combined with the sentiment analysis data of the latest session to automatically generate the latest user portrait. After receiving the warning notification, the psychological counseling expert can understand the user's situation through the user portrait. At the same time, when the psychological counseling expert requests a recommended counseling plan through the expert workbench, the counseling plan recommendation engine queries the user's latest communication records and user portraits, and then recalls relevant information on "cognitive behavioral therapy" or other therapies in the local knowledge base based on the query results. The communication records, user portraits, and therapy-related information are combined through the prompt word engineering, and the big model is requested to recommend a personalized counseling plan, including questions and recommended words for each step, so as to effectively improve the efficiency of counseling communication between psychological counseling experts and users.

[0122] In summary, the present invention communicates with the user through the digital avatar. When the communication is over, the digital avatar monitoring engine identifies and analyzes the warning words or phrases that appear in the user's communication content. When it is identified that the user has negative emotions, the corresponding warning information is sent to the expert workbench in a timely manner, so that the psychological counseling experts can intervene in time. At the same time, the psychological counseling experts can request a consultation plan through the expert workbench. The large model combines the therapy-related information in the local knowledge base, the user portrait, and the user communication record to provide the user with a personalized consultation plan, which greatly improves the efficiency of the psychological counseling experts in intervening in the consultation with the user.

[0123] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A psychological consultation intelligent assistant system based on a large model, characterized in that: include: Interaction layer, service layer and basic layer, including: Interaction layer: includes a digital avatar for communicating with users and an expert workbench for serving psychological counseling experts; Service layer: including the digital twin service engine, digital twin monitoring engine, user portrait generation engine, and consulting solution recommendation engine; Basic layer: includes basic large models, psychological counseling professional field datasets, and fine-tuned large models.

2. The large model-based intelligent assistant system for psychological counseling according to claim 1 is characterized in that: In the interaction layer: The digital twin is used to communicate with users, send the user's communication records to the service layer for keyword monitoring, and reflect the information fed back by the service layer in the communication with the user; The expert workbench includes four sub-modules: monitoring panel, user portrait, service takeover, and consulting plan. The user's emotions are analyzed at the basic level. When the service level determines that the user's emotions trigger the warning rules, the user's situation is reflected at the expert workbench and the psychological counseling experts are notified to intervene.

3. The large model-based psychological counseling intelligent assistant system according to claim 1 is characterized in that: In the service layer: The digital twin service engine is used to receive and send real-time data with the digital twin, receive the user's communication records and return questions or communication words to the user; The digital twin monitoring engine is used to query the user's latest communication records and warning words during the communication process. It requests the basic layer to identify and analyze the user's real-time emotional state and emotional changes through the prompt word engineering. When the basic layer reports that the user's emotions trigger the warning rules, it notifies the psychological counseling experts to intervene. After the user portrait generation engine analyzes the user's communication records and warning words in the basic layer digital intelligence twin monitoring engine, it will periodically calculate the duration of negative emotions and the statistics of emotion sources generated by the basic layer's emotional trend analysis and emotional source analysis, and save the analysis result data; The counseling solution recommendation engine generates personalized counseling solutions based on user portraits and the latest communication records, combined with specific psychotherapy operation steps, through the prompt word engineering request base layer.

4. A method for providing a psychological consultation intelligent assistant service based on a large model, using the psychological consultation intelligent assistant system based on a large model as described in any one of claims 1 to 3 for data processing, characterized in that: The steps include: Step 1: The user requests a consultation, the Digital Avatar Service Engine starts and establishes a dialogue, receives the user's consultation, and saves the user's communication records in the database; Step 2: Based on the user's inquiry content, request feedback from the big model in the base layer, and reply to the user's inquiry based on the feedback content of the big model; Step 3: After completing the reply, the Digital Clone Monitoring Engine analyzes the user's emotions during the consultation process to determine whether the user's emotions trigger warning rules. Step 4: If the user's emotions trigger the warning rules, the user's emotional status is analyzed and the analysis results are sent to the expert workstation for psychological counseling experts to use as a reference for judgment; Step 5: Generate a user profile based on the user's emotional status. Based on the user profile and the latest communication records, combined with specific psychological therapy operation steps, generate a personalized psychological counseling plan through the prompt word engineering request base layer; Step 6: Recommend the personalized psychological counseling plan feedback to the psychological counseling experts, who will provide consultation and guidance to the users.

5. The psychological consultation intelligent assistant service method according to claim 4, characterized in that: The specific steps of step 3 are: Step 3.1: The Digital Avatar Service Engine forwards the user communication records to the Digital Avatar Monitoring Engine; Step 3.2: The digital twin monitoring engine uses a large model to identify the user's communication records, determine and extract warning words or phrases in the communication records; Step 3.3: After the big model recognizes the communication records, it organizes the extracted warning words or phrases into warning information and returns the corresponding warning information to the digital twin monitoring engine; Step 3.3: The digital clone monitoring engine receives the warning information issued by the large model, and issues warnings for the extremely negative emotions in the user's current communication records and the duration of the user's negative emotions over a period of time, and sends the warning information to the expert workstation.

6. The psychological consultation intelligent assistant service method according to claim 5, characterized in that: The warning process of very negative emotions is as follows: Step a.1: The digital twin monitoring engine receives user communication records; Step a.2: Combine the warning word library to assemble relevant prompt words; Step a.3: Analyze the user's communication records based on the prompt words, and identify the warning words or phrases in the consultation message raised in the user's communication records; Step a.4: If there are no identical or similar warning words or phrases in the communication record, the warning process ends; Step a.5: If the same or similar warning words or phrases exist in the communication record, a warning is sent to the expert workbench; Step a.6: After receiving the warning, the expert workstation notifies the psychological counseling expert to handle the warning; Step a.7: Determine whether the psychological counseling expert has completed processing the warning. If not processed within the time threshold of the set warning timer, return to step a.5 and initiate a warning to the expert workstation again; if the psychological counseling expert has completed processing the warning, end the warning process.

7. The psychological consultation intelligent assistant service method according to claim 5, characterized in that: The specific process of the negative emotion duration is as follows: Step b.1: The Digital Intelligence Avatar monitoring engine polls the user; Step b.2: When a user has a negative emotion warning record, the frequency and duration of the user's negative emotion warning are identified and analyzed to determine whether the duration of the user's negative emotion exceeds the set time value; Step b.3: If it is determined that the duration of the user's negative emotion does not exceed the set time value, the process ends; Step b.4: If it is determined that the user's negative emotions last longer than the set time, a warning message is generated based on the duration of the user's negative emotions and sent to the expert workstation, which is received and processed by the psychological counseling expert.

8. The psychological consultation intelligent assistant service method according to claim 4, characterized in that: The specific steps of generating the user portrait in step 5 are: Step c.1: After the user's consultation process is completed, the Digital Avatar Service Engine requests the User Portrait Generation Engine to analyze the emotional information of this round of consultation; Step c.2: The user portrait generation engine analyzes the user's emotional information through the large model and saves the analysis results in the database; Step c.3: The user portrait generation engine regularly polls the emotion analysis results of the target users in the database, and calculates and counts the user's daily emotion information; Step c.4: When the psychological counseling expert requests a user portrait through the expert workbench, the user portrait generation engine queries the user's latest communication records from the database and requests the big model to analyze the user's latest emotional information in combination with the prompt words; Step c.5: Based on the latest sentiment information, combine the most recent sentiment analysis results of the user queried by the user portrait generation engine to generate a user portrait.

9. The psychological consultation intelligent assistant service method according to claim 4, characterized in that: The specific steps of generating and recommending the personalized psychological counseling program are as follows: Step d.1: The psychological counseling expert requests a counseling solution from the counseling solution recommendation engine; Step d.2: The consultation solution recommendation engine queries the user's consultation communication records from the database, and queries the user's sentiment analysis data in the user portrait generation engine; Step d.3: Based on the query results, retrieve therapy-related information from the local knowledge base; Step d.4: Combine the communication records, user portraits, and therapy-related information through prompt words, and request the big model to recommend a consultation plan; Step d.5: Review the counseling plan and then return the reviewed counseling plan to the psychological counseling expert.

Images