Interactive psychological consultation session system

Abstract

The invention relates to the technical field of psychological consultation, in particular to an interactive psychological consultation session system, which comprises a dialogue management module used for receiving consultation information of a current user and determining a context process and sequence when the user consults; the language processing module is used for analyzing the consultation information and understanding the emotional tendency expressed by the current user; the consultation knowledge module is used for acquiring historical data consulted by the user and determining an emotion mode and consultation degree information consulted by the user in combination with the emotion tendency of the consultation information; the consultation recommendation module is used for determining user preferences according to an emotion mode and consultation degree information consulted by the user; recommending a consultation scheme according to user preferences; the more personalized and appropriate consultation service is realized, and the psychological consultation efficiency and accuracy of the user are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of psychological consultation, and in particular to an interactive psychological consultation conversation system. Background Art

[0002] Psychological counseling is the main professional way to maintain people's mental health. Unlike biomedicine, psychological counseling mainly works through good interpersonal communication between counselors and clients. Therefore, assessing the client's ability to participate in psychological counseling, and assessing and selecting appropriate counseling methods and counselors for them (hereinafter referred to as "psychological counseling decision assessment") directly affects the counseling effect and is the core element of psychological counseling. Psychological counseling decision assessment is conducive to counseling agencies to select the most suitable counselors for specific clients, and to psychological counselors to select the most appropriate counseling methods and styles for clients. It also avoids iatrogenic injuries caused by mismatches in counselors and / or counseling methods, which undermines the effectiveness of counseling. Psychological counseling decision assessment is also an important support means for the development of mental health services in the era of precision medicine.

[0003] For example, Chinese patent publication number CN112885433A discloses a psychological counseling service matching method, system, device and storage medium, the method comprising: obtaining first information of a user, and obtaining a psychological counseling suitability index of the user based on the first information; determining that the user is suitable for psychological counseling services based on the psychological counseling suitability index, then obtaining second information of the user, and matching a first psychological counseling method in a preset psychological counseling method library based on the second information; obtaining third information of the user, and matching a first psychological counselor in a preset psychological counselor library based on the third information.

[0004] However, when selecting a consultant based on the corresponding indicators, it is also necessary to determine the changes in the user's intentions and emotional tendencies during the consultation, and to adjust the settings according to each person's habits to improve the accuracy of the consultation. Summary of the invention

[0005] In order to solve the above technical problems, the technical solution adopted by the present invention is: an interactive psychological consultation conversation system, including: a dialogue management module, which is used to receive the consultation information of the current user and determine the context flow and sequence of the user's consultation.

[0006] The language processing module is used to analyze the consulting information and understand the emotional tendencies expressed by the current user.

[0007] The consultation knowledge module is used to obtain historical data of user consultations and determine the emotional pattern and consultation degree information of user consultations in combination with the emotional tendency of the consultation information.

[0008] The consultation recommendation module is used to determine user preferences based on the user's consultation emotional patterns and consultation level information; and recommend consultation plans based on user preferences.

[0009] The consultation feedback module is used to obtain consultation feedback after the implementation of the consultation plan, and extract the consultation stage of the current user from the consultation plan, and combine the consultation plan and consultation feedback of each consultation stage to obtain the feedback evaluation plan of the current user.

[0010] The beneficial effects of the present invention are: 1. The present invention ensures that the context perception capability of the consultation session system is stronger by real-time monitoring of the user's consultation intention changes, emotional state, time and other factors, thereby providing more personalized and appropriate consultation services.

[0011] 2. Through in-depth analysis of user consultation information, the system of the present invention can accurately identify the user's emotional tendencies and recommend more appropriate consultation plans accordingly.

[0012] 3. The present invention utilizes the user's historical consultation data, and the system can identify the user's emotional pattern and consultation level information, and then provide consultation suggestions that better meet the user's needs.

[0013] 4. By comprehensively considering the user's emotional pattern and consultation level information, the system of the present invention can determine the user's preference and recommend the most suitable consultation plan for the current user based on it.

[0014] 5. The present invention not only provides consulting services, but also can evaluate the effectiveness of consulting solutions by collecting consulting feedback, and continuously optimize consulting services accordingly. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0016] Figure 1 It is a system framework diagram based on an interactive psychological counseling conversation system.

[0017] Figure 2 It is a system diagram based on an interactive psychological counseling conversation system.

[0018] Figure 3 It is a flow chart of a consultation feedback module based on an interactive psychological consultation conversation system. DETAILED DESCRIPTION

[0019] The embodiments of the present invention are described in detail below. The embodiments described below are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention. If no specific techniques or conditions are specified in the embodiments, the techniques or conditions described in the literature in the art or the product specifications are used.

[0020] See also Figure 1 , Figure 2 , an interactive psychological counseling conversation system, including: dialogue management module, language processing module, counseling knowledge module, counseling recommendation module, and counseling feedback module.

[0021] The dialogue management module receives the user's consultation information; the dialogue management module passes the user's consultation information and its context to the language processing module for further analysis; the language processing module analyzes the user's consultation information, understands the emotional tendency expressed by the user, and passes the emotional tendency analysis results to the consultation knowledge module to further determine the emotional pattern; the consultation knowledge module obtains the historical data of the user's consultation, and passes the emotional pattern and consultation degree information to the consultation recommendation module to recommend consultation plans based on user preferences; the consultation recommendation module determines the recommended consultation plan, and passes the recommended consultation plan to the consultation feedback module to collect feedback after implementation; the consultation feedback module obtains the consultation feedback after the consultation plan is implemented. The feedback evaluation plan is passed back to the dialogue management module or other related modules to further optimize the consultation process.

[0022] The dialogue management module is used to receive the current user's consultation information and determine the context flow and sequence of the user's consultation.

[0023] The language processing module is used to analyze the consulting information and understand the emotional tendencies expressed by the current user.

[0024] The consultation knowledge module is used to obtain historical data of user consultations and determine the emotional pattern and consultation degree information of user consultations in combination with the emotional tendency of the consultation information.

[0025] The consultation recommendation module is used to determine user preferences based on the user's consultation emotional patterns and consultation level information; and recommend consultation plans based on user preferences.

[0026] The consultation feedback module is used to obtain consultation feedback after the implementation of the consultation plan, and extract the consultation stage of the current user from the consultation plan, and combine the consultation plan and consultation feedback of each consultation stage to obtain the feedback evaluation plan of the current user.

[0027] In normal psychological counseling, since the issues that need to be consulted will contain various complex elements, and whether the counselor can accurately express the current psychological problems, and after determining the psychological problems expressed by the counselor, how to characterize the current psychological problems, and judge what intervention plans should be adopted for the current psychological problems, it is necessary to characterize the counselor's plan according to the corresponding situation, and analyze each problem separately, so as to determine whether the psychological problems of the counselor and the psychological problems expressed by the counselor are the same problem, how the related problems arise, and how to reduce the impact of psychological problems on the counselor, so as to select a suitable plan to guide the counselor.

[0028] At this time, when conducting a psychological counseling session with the user, the user will first be asked to take a specific simulation test to verify whether the user has mental illnesses such as mania and autism. The corresponding scale used will be used for initial testing, such as the Minnesota Multiphasic Personality Questionnaire, Beck Depression Inventory, etc., and then the counseling information sent by the user will be received, and the progress of the conversation will be gradually tracked, and this part of the tracking information will be input into the language processing module.

[0029] That is, the implementation method of determining the context flow and sequence of user consultation also includes: monitoring the changes in the intention of each round of dialogue during user consultation, and recording the scene of the dialogue, the user's emotional state and time.

[0030] The changes in the intention of each round of dialogue can be obtained through the recognition content of the consulting knowledge module, and the user's emotional state is obtained by analyzing the consulting information through the language processing module, so that the user's session records can be continuously recorded and tracked.

[0031] The following are some scales used in psychological counseling. When using these scales, part of them will be selected for testing based on the current diagnostic method, and combined with the form of communication to determine the psychological problems that the current user may have.

[0032] Minnesota Multiphasic Personality Inventory (MMPI): This is a widely used adult personality test that assesses a variety of psychological disorders and personality traits.

[0033] Beck Depression Inventory (BDI): Used to assess the degree of depression in patients with depression.

[0034] Hamilton Anxiety Rating Scale (HAM-A): used to assess the severity of anxiety symptoms.

[0035] Generalized Anxiety Disorder Scale (GAD-7): This is a brief self-report scale used to screen for generalized anxiety disorder.

[0036] Patient Health Questionnaire (PHQ-9): This is a self-report questionnaire that assesses symptoms of depression.

[0037] Young Mania Rating Scale (YMRS): Used to assess the severity of manic symptoms in patients with bipolar disorder.

[0038] Brief Psychiatric Rating Scale (BPRS): Used to assess the severity of symptoms in patients with psychosis, particularly schizophrenia.

[0039] Child Behavior Checklist (CBCL): This is a tool used to assess behavioral, emotional, and social problems in children and adolescents.

[0040] Eysenck Personality Questionnaire (EPQ): Used to assess personality traits such as neuroticism, introversion, and psychoticism.

[0041] Stanford-Binet Intelligence Scale (SBIS): This is an intelligence test used to assess an individual's cognitive abilities and IQ level.

[0042] The language processing module analyzes the consulting information and understands the emotional tendency expressed by the current user in the following manner: obtain the user's preliminary verification information, and determine the user's psychological counseling index based on the preliminary verification information; extract the first key information from the consulting information based on the obtained preliminary verification information, and determine the user's corresponding confession preference index based on the first key information; determine the user's related symptom matching index based on the confession preference index, and the symptom matching index is used to indicate the degree of the current user's symptom and the specific manifestation type of the symptom; translate the current user's consulting information based on the obtained symptom matching index to obtain the emotional tendency expressed by the user; the purpose of translation here is to verify the illogical behavior of the current user before and after the consultation, as well as the problems in the preliminary verification, so as to select the processing method that best suits the current user.

[0043] Initial verification information usually refers to the basic information provided by the user when they first contact the psychological counseling service. This information may include the user's age, gender, occupation, educational background, family status, personal medical history, current major problems or concerns, etc. This information helps the psychological counselor to form a preliminary understanding and assessment of the user.

[0044] The first key information is the core points extracted and analyzed from the preliminary verification information provided by the user. This information may include the user's most prominent psychological problems, emotional state, behavioral performance, thinking patterns, etc. The first key information is more focused and specific, and can directly reflect the user's current psychological state and need to confide.

[0045] The final emotional tendency is the result of the consultant's comprehensive use of professional knowledge, skills and experience to transform the user's original expression (which may be vague, confusing or emotional) into a clearer, more specific, easier to understand and operate form. It may include the description of the nature, cause and development process of the user's psychological problems, the interpretation of the user's emotions, behaviors and cognitive patterns, and the identification of the user's potential needs and expectations.

[0046] When psychological counseling indicators are expressed as the current preliminary verification information, the basic indicators corresponding to the current user, such as the psychological state classification, emotion classification, behavior classification, environment classification, etc. of the current user, will be extracted from the preliminary verification information to obtain the required indicator values ​​in these classifications. This indicator value is generally expressed as a value from 0 to 1 to quantify whether the current user has a corresponding condition. If the performance is normal, the corresponding value is regarded as 0.

[0047] Psychological state classification: Depression: determine whether the user shows depressive tendencies; Anxiety: assess the user's anxiety level; Stress: measure the degree of stress the user feels; Self-efficacy: evaluate the user's confidence level in dealing with challenges; Stress response: identify the user's response pattern when facing stress.

[0048] Emotion classification: Emotional stability: evaluate the degree of user emotional fluctuations; Positive emotions: evaluate the positive emotions expressed by users; Negative emotions: identify the negative emotions expressed by users; Emotional expression ability: whether users can clearly express their emotions.

[0049] Behavior classification: Coping mechanism: the coping strategies adopted by users when facing difficulties; Social interaction: the way and quality of users' interactions with others; Living habits: eating, sleeping and other living habits; Behavioral patterns: users' behavioral habits and their changing trends. Occupational status: whether employed or not, job satisfaction.

[0050] Environmental classification: Living environment: living conditions and their impact on emotions; Economic situation: financial situation and the stress it brings; Major life events: major changes or events experienced recently; Traumatic experience: whether there are signs of post-traumatic stress disorder (PTSD); Coping ability: performance in the face of recent challenges.

[0051] According to the preliminary verification information obtained, the method of extracting the first key information from the consulting information is as follows: according to the preliminary verification information obtained, the information in the consulting information corresponding to the preliminary verification information is obtained as the first key information; at this time, verify whether the classification of the consulting information is the same as that of the corresponding information in the preliminary verification information, and if they are the same, verify the frequency of occurrence of the corresponding words in the consulting information, and sort the recognized text according to the frequency of occurrence of the words to form the first key information.

[0052] Based on the first key information, the implementation method of determining the user's corresponding confession bias index is to set the emotional tendency score, the help-seeking bias score, the information sharing tendency score, and the problem-oriented score according to the user's consulting information, and take the weighted average of the emotional tendency score, the help-seeking bias score, the information sharing tendency score, and the problem-oriented score as the output confession bias index.

[0053] Sentiment tendency score: indicates that users are more likely to express emotions during communication, whether positive or negative. Sentiment analysis technology can be used to score each message of the user, and then the overall sentiment tendency score can be summarized; the score range is: -1 to +1; -1 indicates extremely negative emotions, and +1 indicates extremely positive emotions.

[0054] Help-seeking tendency score: indicates that the user may be inclined to seek help, whether it is seeking solutions, emotional support or information query. Intent recognition technology can be used to determine how much help-seeking components are contained in the user's information; score range: 0 to 1; 0 means no intention to seek help at all, and 1 means completely seeking help.

[0055] Information Sharing Propensity Score: Indicates that users may be more willing to share their experiences, feelings, or ideas; content analysis can be used to determine the amount of information shared by users; score range: 0 to 1; 0 means almost no information sharing, and 1 means very willing to share.

[0056] Problem-oriented score: indicates that the user may be more inclined to talk about specific problems or concerns rather than general feelings or experiences; question type identification technology can be used to assess whether the user's information is focused on problem solving.

[0057] As for the method of obtaining the symptom matching index, the obtained confession bias index is matched with the preset symptom, and the index with the highest similarity after matching is used as the output symptom matching index. At this time, the matching method is to use the confession bias index as input and compare it with the corresponding data in the symptom, calculate the Pearson correlation coefficient, and use the symptom matching index corresponding to the maximum value of the Pearson correlation coefficient as the index with the highest similarity, thereby selecting the corresponding symptom matching index.

[0058] The symptom matching index can reflect the severity of the user's symptoms. For example, a higher index value may mean that the symptoms exhibited by the user are more consistent with the diagnostic criteria of a certain mental illness.

[0059] Based on the obtained disease matching index, the consultation information of the current user is translated to obtain the emotional tendency expressed by the user. In this case, the meaning conveyed to the user itself needs to be based on the user's emotional state and potential psychological condition; if the user shows a tendency to depression, then the user may have more negative emotional weights in certain language expressions. At this time, for the recognition of the language, it is necessary to pay more attention to the user's expression in the relevant language.

[0060] That is, the current user's consultation information is identified, the words with the highest confidence in the consultation information are processed as the keywords for current identification, the emotional tendencies of the identified keywords in multiple scenarios are determined, and a group of emotional tendencies with the highest similarity to the disease matching indicators are used as the emotional tendencies expressed by the current user.

[0061] At this time, the similarity between the selected emotional tendency and the symptom matching index is obtained by comparing the emotional score in the emotional tendency with the emotional score value corresponding to the symptom matching index in the corresponding situation. The one with the smallest difference in the emotional scores between the two is regarded as the one with the highest similarity. At the same time, the emotional score is obtained by comparing the keywords through a pre-set emotional dictionary. Since the symptom matching index represents the score of the current user under the corresponding symptoms, the main emotional preference of the user in the corresponding symptom stage can be obtained. At this time, the user-related indicators can be obtained based on the symptom matching index, and part of them can be quantified into emotional scores, so as to monitor the problems of the user in the corresponding consultation situation.

[0062] After obtaining the emotional tendency of the consulting information in the consulting knowledge module, combined with the historical data of the current user's previous consultations, we can know the emotional pattern of the user's consultation at this time, and identify this obtained emotional pattern to know the user's current state. At the same time, other information about the consultation can be obtained, such as the complexity, detail or urgency of the user's questions, to judge the expected effect of answering the user's consultation at this time, so as to quantify the corresponding consultation level information.

[0063] The method for determining the emotional pattern of user consultation is to compare the emotional tendency of the consultation information with historical data, determine the correlation between the emotional tendencies, and determine the changing trend of the emotional tendencies over time; based on the correlation between the emotional tendencies and the changing trend of the emotional tendencies, identify the emotional pattern corresponding to the current emotional tendency; and use the identified emotional pattern as the output emotional pattern.

[0064] The method of determining the correlation between emotional tendencies requires converting the currently identified emotional tendencies into emotional scores, calculating the correlation coefficient corresponding to the current emotional tendencies based on the current emotional scores and the emotional scores in the historical data, and using the calculated correlation coefficient as the correlation between the emotional tendencies.

[0065] Among them, r xy represents the correlation coefficient corresponding to the current emotional tendency, n represents the number of current emotional scores, i = 1, 2,,, n. Since there are fluctuations in emotional tendencies during normal consultation, the value of each emotional score when the emotional tendency fluctuates is recorded at this time to determine the correlation coefficient corresponding to the corresponding score of the current emotional tendency, x i represents the current i-th sentiment score, y i represents the i-th sentiment score in the historical data, represents the average value of the current sentiment score, Represents the average of sentiment scores in historical data.

[0066] For the average value of the current emotion score selected at this time, the multiple emotion scores after the emotional tendency is converted are identified, and then the corresponding average value is obtained; for the average value of the emotion score selected by the historical data, it is determined based on the user's historical consultation situation, and the average value of the emotion score set for the user in the history is calculated. If the user has not participated in psychological counseling, the average value of the emotion score in the historical data selected at this time is the average value of the emotion score obtained in the overall historical data under the symptoms corresponding to the current user; to quantify the user's display under different symptoms.

[0067] At this time, the correlation coefficient can be used to determine the change in the user's emotional tendency during the current interactive psychological consultation, as well as the corresponding correlation under the change in emotional tendency, to determine whether the current user's status is the same as the historical data that has been identified, to assist in selecting the consultation method for the current user.

[0068] When identifying the changing trend of emotional tendencies, the main thing to identify is the change in the emotional score corresponding to the emotional tendency, so as to determine the change in the emotional tendency; first, identify the turning point when the emotional tendency changes, as well as the difference between the maximum and minimum values ​​of the emotional score corresponding to the emotional tendency within a fixed time. The difference is regarded as the basic difference, and the basic difference needs to be smaller than the preset difference. The purpose of setting the basic difference is to judge the fluctuation of the emotional tendency within a fixed time to determine the emotional stability of the current consulting user; the fixed time setting range is 10 minutes, 30 minutes, and 1 hour, and the specific selection range can also be adjusted according to the needs of the current user to quantify the accuracy of the current user's processing.

[0069] Determine the change value of the emotion score of the emotion tendency between the identified turning points, determine the emotion pattern corresponding to the turning point according to the change value of the emotion score, and use the emotion pattern with the highest probability of appearing in the emotion pattern as the emotion pattern output at this time.

[0070] The emotional pattern with the highest probability of occurrence is usually more representative and can better reflect the mainstream trend of the user's emotional state; the emotional pattern with the highest probability of occurrence is usually more representative and can better reflect the mainstream trend of the user's emotional state; at this time, the focus is on processing the user's emotional changes in the short term, so as to carry out timely processing according to the user's emotional changes in the short term.

[0071] Assume that the changing trend of the user's emotional tendency is as follows: Turning point 1: emotional score is -0.5; Turning point 2: emotional score is -0.4; Turning point 3: emotional score is -0.7 (lowest point); Turning point 4: emotional score is -0.6; Turning point 5: emotional score is -0.3 (highest point); Turning point 6: emotional score is -0.5.

[0072] Based on the above sentiment scores, the following sentiment patterns can be identified: Declining pattern: From turning point 1 to turning point 3, the sentiment score drops from -0.5 to -0.7.

[0073] Rising pattern: From turning point 3 to turning point 5, the sentiment score rises from -0.7 to -0.3.

[0074] Fluctuation pattern: From turning point 3 to turning point 6, the sentiment score fluctuates between -0.7 and -0.5.

[0075] The consultation degree information is realized by extracting the consultation questions of the current consultation information, and taking the probability distribution of the user's consultation questions, the consultation time and the level of the consultation questions as the consultation degree information.

[0076] The probability distribution of user consulting questions is obtained by calculating the proportion of the type corresponding to the current consulting question to the total consulting questions; the consultation time records the duration of the current user's consultation to determine how long the current user needs to communicate to complete a psychological consultation process; the level of the consulting question indicates the level to which the current consulting question can be decomposed. This level is used to sufficiently refine the current consulting question to determine the essence of the consulting question, such as breaking down the consulting question into three levels, or more levels, where the first level is the original question, the second level represents the main aspects of the consulting question, and the third level is the details and action steps corresponding to the main aspects. If the decomposition exceeds three levels, the first two levels are the same, and the third level is converted into a component of the main aspect, and the fourth level can be the details of the component; if it is decomposed into two levels, the second level is the details of the original question, and so on. The number of levels currently selected will be adjusted according to the complexity of the consulting question.

[0077] The method of obtaining consulting solutions is to determine user preferences and classify user preferences based on the emotional patterns and consultation degree information of the user's consultation; set priorities for each type of user preferences, and select the user's consulting solution based on the priorities within the user preferences.

[0078] Determine user preferences, and classify user preferences in the following way: use the probability distribution of user consultation questions in the consultation degree information, the consultation time, and the level of the consultation questions as the classification basis to obtain a first classification list, and use the emotional pattern as the classification basis to obtain a second classification list; compare the first classification list and the second classification list, and compare the data in the first classification list with the data in the second classification list. At this time, what is calculated is the value of the data after the emotional pattern and the consultation degree information are classified according to different situations. The comparison method is to calculate the Pearson correlation coefficient of these data, and use the classification list with the largest Pearson correlation coefficient after comparison as the output classification list, and set the priority according to the compared classification list. The set priority is set according to the weighted average of the corresponding weights of the probability distribution of user consultation questions, consultation time, the level of consultation questions, and the emotional pattern in the classification list. At this time, a classification list will be obtained, and the consultation plan corresponding to the output classification list will be obtained, and the consultation plan with the largest priority will be selected as the output consultation plan.

[0079] like Figure 3 As shown, when obtaining the feedback evaluation plan, the cosine characteristics of the consulting plans in all consulting stages are used as the basic evaluation conditions for the feedback evaluation at this time; when the basic evaluation conditions meet the preset conditions, the key factors that have changed in the consulting stage are recorded, and the key factors are subjected to component analysis to obtain three groups of consulting plans, and the consulting plan after splicing the three groups of consulting plans is obtained to obtain the current feedback evaluation plan.

[0080] In this plan, consulting feedback is expressed in the form of user consulting goal achievement, user satisfaction, behavior changes, etc. When processing the user in combination with the consulting feedback, the consulting plan with the changed consulting feedback will be used as the processing plan at this time to determine what plan should be used for feedback after the user undergoes corresponding changes, thereby improving the targeted processing effect for different users.

[0081] At this time, the key factors that change during the consultation stage mainly include the changes in the user's emotional patterns at each consultation stage and the changes in the content of the questions the user wants to consult. At this time, the changes in the user's emotions during each consultation will be recorded, and the content most relevant to the emotional changes will be used as the key factor for identification at this time; that is, the key factor is the corresponding changes in the following content.

[0082] Changes in counseling content: shifts in counseling topics; Impact of specific events: major events or experiences in life; Changes in emotional patterns: shifts from one emotional pattern to another; Adjustments in counseling strategies: counselors adjust counseling methods based on user situations; Use these contents as analytical data for feedback and evaluation of counseling programs implemented in all counseling stages.

[0083] The first group of consulting solutions is reconstructed according to the dimensions of each consulting solution, and the first group of consulting solutions is obtained recursively.

[0084] At this time, define the initial consulting solution set, and subdivide each dimension, and generate new consulting solutions based on the subdivided dimensions until the required consulting solution is reached.

[0085] Assume that the initial consultation plan set is: short-term consultation, high-frequency problems, medium-level problems, stable mode; medium-term consultation, medium-frequency problems, medium-level problems, descending mode; long-term consultation, low-frequency problems, medium-level problems, ascending mode.

[0086] Refine each dimension: Consultation time: short-term -> short-term 1, short-term 2; medium-term -> medium-term 1, medium-term 2; long-term -> long-term 1, long-term 2.

[0087] Probability distribution of consulting questions: high frequency -> high frequency 1, high frequency 2; medium frequency -> medium frequency 1, medium frequency 2; low frequency -> low frequency 1, low frequency 2.

[0088] The levels of consulting questions: Intermediate -> Intermediate 1, Intermediate 2; Basic -> Basic 1, Basic 2; Advanced -> Advanced 1, Advanced 2.

[0089] Emotional pattern: stable->stable1, stable2; declining->declining1, declining2; rising->rising1, rising2; fluctuating->fluctuating1, fluctuating2.

[0090] Combine to generate new solutions: short range 1, high frequency 1, intermediate level 1, stable 1; short range 2, high frequency 2, intermediate level 2, stable 2; medium range 1, medium frequency 1, intermediate level 1, descending 1; medium range 2, medium frequency 2, intermediate level 2, descending 2; long range 1, low frequency 1, intermediate level 1, ascending 1; long range 2, low frequency 2, intermediate level 2, ascending 2.

[0091] Repeat the above process until each dimension is fully subdivided to form the first set of consulting solutions.

[0092] The second group of consulting solutions is reconstructed recursively according to the probability of the key factors appearing in each consulting solution to obtain the second group of consulting solutions.

[0093] At this time, the initial consulting solution set is defined, and then the probability of occurrence of each key factor is counted, and screening is performed based on the obtained key factors until the calculated indicators of each key factor reach the target value.

[0094] Assume that the initial consultation plan set is: short-term consultation, high-frequency issues, medium-level issues, and stable mode.

[0095] Mid-range consultation, mid-frequency issues, intermediate issues, descending mode.

[0096] Long-term consultation, low-frequency issues, intermediate issues, rising patterns.

[0097] Calculate the probability of key factors: Specific events: family conflict 0.3, work pressure 0.5, interpersonal relationship 0.2.

[0098] Consultation topics: mental health 0.6, academic counseling 0.3, career planning 0.1.

[0099] Emotional mode: stable mode 0.4, falling mode 0.3, rising mode 0.2, fluctuating mode 0.1.

[0100] According to the probability, the most relevant consulting solutions for key factors are selected, such as: work pressure related: medium-term consulting, high-frequency problems, intermediate problems, and declining patterns.

[0101] Mental health related: long-term counseling, low-frequency problems, intermediate problems, stable patterns.

[0102] Repeat the above process until each key factor has been fully screened and reconstructed to form the second set of consulting solutions.

[0103] At this time, the method to achieve the target value for the calculated indicators of each key factor is to calculate the screened consulting plan and the standard plan, determine whether the Pearson correlation coefficient between the indicator value corresponding to the component of the selected plan and the indicator value of the standard plan can take the maximum value, and iterate in sequence until all consulting plans are reconstructed.

[0104] The third group of consulting solutions extracts key components from the first and second groups of consulting solutions, maps the obtained key components, and determines the impact range corresponding to each mapped key component; and sets the consulting solutions whose impact range meets the initially set range as the third group of consulting solutions.

[0105] At this time, an initial set of consulting solutions is defined, key components are extracted from the first and second sets of consulting solutions, the scope of influence is determined based on the key components, and consulting solutions whose scope of influence meets the initially set scope are screened out.

[0106] The key components are assumed to be: Duration of consultation: short, medium, long.

[0107] Probability distribution of consulting questions: high frequency, medium frequency, and low frequency.

[0108] Levels of consulting questions: basic, intermediate, advanced.

[0109] Emotional patterns: stable, falling, rising, fluctuating.

[0110] Determine the scope of impact: Consultation duration: short-term (≤1 month), medium-term (1-6 months), and long-term (>6 months).

[0111] Probability distribution of consulting questions: high frequency (≥50%), medium frequency (20%-50%), and low frequency (<20%).

[0112] Levels of consulting questions: basic (simple), intermediate (medium), advanced (complex).

[0113] Emotional patterns: stable (fluctuation ≤ ±0.2), decreasing (fluctuation ≤ -0.5), increasing (fluctuation ≥ +0.5), and fluctuating (fluctuation ≥ ±0.5).

[0114] According to the scope of influence, qualified consulting plans are screened out, such as: short-term consultation (≤1 month), high-frequency problems (≥50%), intermediate problems (medium), and stable mode (fluctuation ≤±0.2); finally, the third group of consulting plans are formed.

[0115] The first, second and third consulting solutions are converted into feature vectors, and the solution with the shortest feature vector distance among the three consulting solutions is used as the final output feedback evaluation solution.

[0116] At this time, the mutual distances between the transformed feature vectors are calculated, and the distance d1 between the feature vectors of the first group of consulting solutions and the second group of consulting solutions, the distance d2 between the feature vectors of the second group of consulting solutions and the third group of consulting solutions, and the distance d3 between the feature vectors of the first group of consulting solutions and the third group of consulting solutions are output. The distances d1, d2, and d3 are summed up to take the minimum value, and the consulting solution corresponding to the corresponding feature vector is output to obtain the final feedback evaluation solution.

[0117] This final plan can combine the changes of users at different consultation stages to generate a more comprehensive consultation plan that can be more suitable for users to adjust, so as to improve the effect of user consultation, guide users in time during the interaction process, and adjust the user's psychological state.

[0118] The sum of cosine features of consulting plans in all consulting stages is to convert the consulting plans set in all consulting stages into vectors and calculate the cosine similarity of each vector. At this time, each consulting plan will convert the consulting features existing in the consulting plan into multiple vectors, and calculate the cosine similarity of the vectors of the current consulting plan and the standard consulting plan one by one. After summing up the calculated cosine similarities, the sum of cosine features of the consulting plan can be obtained.

[0119] The preset conditions set for the basic evaluation conditions require that the cosine characteristics of the consultation plan and the value set in the preset conditions be greater than the value set in the preset conditions, for example, the vector V of the consultation time is obtained. t , the probability distribution vector V of the consulting question p , the vector V of the level of the consulting question c , the vector V of the emotional pattern e ; Calculate the cosine similarities corresponding to these four vectors in turn.

[0120] Among them, Similarity(t) represents the cosine similarity of consultation time, U t A standard vector representing the consultation time.

[0121] The calculation method for the probability distribution of consulting questions, the level of consulting questions, and the cosine similarity of emotional patterns is the same as the above formula, and they are not listed here one by one; after calculating these four cosine similarities, they are summed up to get the corresponding cosine feature sum. At the same time, the preset conditions set at this time require that each cosine similarity calculated must be greater than 0.6, and the cosine feature sum must be greater than 2.8, to verify whether the consulting plan used in each stage meets the requirements.

[0122] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations on the present invention. A person skilled in the art may make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention and they are still covered by the protection scope of the present invention.

Claims

1. An interactive psychological consultation conversation system, characterized in that: include: The dialogue management module is used to receive the current user's consultation information and determine the context flow and sequence of the user's consultation; The language processing module is used to analyze the consultation information and understand the emotional tendency expressed by the current user; The consultation knowledge module is used to obtain the historical data of user consultations and determine the emotional pattern and consultation degree information of the user consultations in combination with the emotional tendency of the consultation information; A consultation recommendation module is used to determine user preferences based on the user's consultation sentiment pattern and consultation degree information; Recommend consulting solutions based on user preferences; The consultation feedback module is used to obtain consultation feedback after the implementation of the consultation plan, and extract the consultation stage of the current user from the consultation plan, and combine the consultation plan and consultation feedback of each consultation stage to obtain the feedback evaluation plan of the current user.

2. The interactive psychological consultation conversation system according to claim 1, characterized in that: The implementation method of determining the context flow and sequence of user consultation also includes: monitoring the changes in the intention of each round of dialogue during user consultation, and recording the scene of the conversation, the user's emotional state and time.

3. The interactive psychological consultation conversation system according to claim 1, characterized in that: The language processing module analyzes the consulting information and understands the emotional tendency expressed by the current user in the following manner: obtaining the user's preliminary verification information, and determining the user's psychological counseling index based on the preliminary verification information; extracting the first key information from the consulting information based on the obtained preliminary verification information, and determining the user's corresponding confession bias index based on the first key information; determining the user's related symptom matching index based on the confession bias index; and translating the current user's consulting information based on the obtained symptom matching index to obtain the emotional tendency expressed by the user.

4. The interactive psychological consultation conversation system according to claim 3 is characterized in that: The method of extracting the first key information from the consulting information according to the obtained preliminary verification information is to obtain information in the consulting information corresponding to the preliminary verification information as the first key information according to the obtained preliminary verification information; Verify whether the classification of the consulting information is the same as that of the corresponding information in the preliminary verification information. If they are the same, verify the frequency of occurrence of the corresponding words in the consulting information, and sort the recognized words according to the frequency of occurrence of the words to form the first key information; Based on the first key information, the implementation method of determining the user's corresponding confession bias index is to set the emotional tendency score, the help-seeking bias score, the information sharing tendency score, and the problem-oriented score according to the user's consulting information, and take the weighted average of the emotional tendency score, the help-seeking bias score, the information sharing tendency score, and the problem-oriented score as the output confession bias index.

5. The interactive psychological consultation conversation system according to claim 1, characterized in that: The method of determining the emotional pattern of the user's consultation is to compare the emotional tendency of the consultation information with the historical data, determine the correlation between the emotional tendencies, and determine the changing trend of the emotional tendencies over time; according to the correlation between the emotional tendencies and the changing trend of the emotional tendencies, identify the emotional pattern corresponding to the current emotional tendency; The identified emotion pattern is used as the output emotion pattern.

6. The interactive psychological consultation conversation system according to claim 5, characterized in that: The method of determining the correlation between sentiment tendencies requires converting the currently identified sentiment tendencies into sentiment scores, calculating the correlation coefficient corresponding to the current sentiment tendencies based on the current sentiment scores and the sentiment scores in the historical data, and using the calculated correlation coefficient as the correlation between sentiment tendencies: Among them, r xy represents the correlation coefficient corresponding to the current sentiment tendency, n represents the number of current sentiment scores, i = 1, 2,,, n, x i represents the current i-th sentiment score, y i represents the i-th sentiment score in the historical data, represents the average value of the current sentiment score, Represents the average of sentiment scores in historical data.

7. The interactive psychological consultation conversation system according to claim 5, characterized in that: When identifying the changing trend of the emotional tendency, first identify the turning point of the emotional tendency when it changes, and the difference between the maximum and minimum emotional scores corresponding to the emotional tendency within a fixed time. The difference is regarded as the basic difference, and the basic difference is less than the preset difference; Determine the change value of the emotion score of the emotion tendency between the identified turning points, determine the emotion pattern corresponding to the turning point according to the change value of the emotion score, and use the emotion pattern with the highest probability of appearing in the emotion pattern as the emotion pattern output at this time.

8. The interactive psychological consultation conversation system according to claim 1, characterized in that: The consultation degree information is realized by extracting the consultation questions of the current consultation information, and taking the probability distribution of the user's consultation questions, the consultation time and the level of the consultation questions as the consultation degree information.

9. The interactive psychological consultation conversation system according to claim 8, characterized in that: The method of obtaining consulting solutions is to determine user preferences based on the emotional patterns and consultation degree information of users, and classify user preferences; set priorities for each type of user preferences, and select the user's consulting solution based on the priorities within the user preferences; Determine user preferences, and classify user preferences by taking the probability distribution of user consultation questions, consultation time, and the level of consultation questions in the consultation degree information as the classification basis to obtain a first classification list, and taking the emotional pattern as the classification basis to obtain a second classification list; Compare the first classification list with the second classification list, compare the data in the first classification list with the data in the second classification list, take the classification list with the largest Pearson correlation coefficient after comparison as the output classification list, and set the priority according to the compared classification list. The set priority is set according to the weighted average of the corresponding weights of the probability distribution of the user's consultation questions, the consultation time, the level of the consultation questions and the emotional pattern in the classification list, obtain the consultation plan corresponding to the output classification list, and select the consultation plan with the largest priority as the output consultation plan.

10. The interactive psychological consultation conversation system according to claim 1, characterized in that: Feedback evaluation plans can be obtained by: The sum of the cosine characteristics of the consulting solutions in all consulting stages is used as the basic evaluation condition for the feedback evaluation at this time; when the basic evaluation condition meets the preset conditions, the key factors that have changed in the consulting stage are recorded, and the key factors are analyzed for components to obtain three groups of consulting solutions. The consulting solution after splicing the three groups of consulting solutions is obtained to obtain the current feedback evaluation solution; The first set of consulting solutions is reconstructed according to the dimensions of each consulting solution, and the first set of consulting solutions is obtained recursively; The second group of consulting solutions is reconstructed recursively according to the probability of the key factors appearing in each consulting solution, and the second group of consulting solutions is obtained; The third group of consulting solutions extracts key components from the first and second groups of consulting solutions, maps the obtained key components, and determines the impact range corresponding to each mapped key component; and sets the consulting solutions whose impact ranges meet the initially set range as the third group of consulting solutions; The first, second and third consulting solutions are converted into feature vectors, and the solution with the shortest feature vector distance among the three consulting solutions is used as the final output feedback evaluation solution.

Images