Psychological portrait data generation method, psychological portrait recognition method, device and equipment

Abstract

This application proposes a method for generating psychological profile data, a method for recognizing psychological profiles, an apparatus, and a device, applicable to the field of artificial intelligence. The method for generating psychological profile data includes: constructing dialogue data in a psychological counseling scenario; labeling the dialogue data with psychological profile tags based on a multi-model voting mechanism, recalling at least one tag from the dialogue data; performing recall analysis on the at least one tag, generating recall analysis results for each tag; the recall analysis results characterize whether the tag is an incorrect or correct tag; based on the recall analysis results of at least one tag, and the thought process of performing psychological profile tag analysis on the dialogue data using a thought chain generation model, generating psychological profile data related to the thought process; the psychological profile data is used to train a psychological profile model. Thus, while providing an accurate data foundation for training the psychological profile model, it enhances the model's generalization ability.

Description

Technical Field

[0001] This application is applied to the field of artificial intelligence, and in particular relates to a method for generating psychological profile data, a method for recognizing psychological profiles, an apparatus and equipment. Background Technology

[0002] With the increasing maturity of artificial intelligence-related technologies, the application of Large Language Models (LLMs) in the field of psychological services is constantly deepening. For example, in scenarios such as psychological counseling and psychological risk early warning, the prediction of user psychological profiles by psychological profiling models is conducive to accurately matching service needs and serves as a core support for assessing users' psychological state.

[0003] The training of psychological profiling models usually adopts a large-scale supervised fine-tuning approach. The model's capabilities mainly rely on high-quality labeled data. The training method is relatively simple, and the model's generalization ability is not high. Some mislabeled labels can not only reduce the training effect of psychological profiling models, but may even mislead the direction of subsequent psychological assessment and intervention.

[0004] In view of this, how to provide a method for generating psychological profile data, a method for recognizing psychological profiles, a device and equipment that can provide an accurate data foundation for training psychological profile models while enhancing the generalization ability of the models is a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0005] To address the aforementioned issues, this application proposes a method for generating psychological profile data, a method for recognizing psychological profiles, an apparatus, and a device, which enhance the generalization ability of the model while providing an accurate data foundation for training the psychological profile model.

[0006] The first aspect of this application provides a method for generating psychological profile data, including: Constructing dialogue data in psychological counseling scenarios; The dialogue data is labeled with psychological profile tags based on a multi-model voting mechanism, and at least one tag of the dialogue data is recalled. A recall analysis is performed on the at least one tag to generate a recall analysis result for each tag; the recall analysis result indicates whether the tag is an incorrect tag or a correct tag; Based on the recall analysis results of the at least one tag, and the thinking process of the thought chain generation model performing psychological profile tag analysis on the dialogue data, psychological profile data related to the thinking process is generated; the psychological profile data is used to train the psychological profile model.

[0007] In one possible implementation, the dialogue data includes at least one of first dialogue data, second dialogue data, and third dialogue data; The first dialogue data includes real online psychological counseling dialogue data; The second dialogue data includes dialogue data constructed based on psychologically relevant search resources; The third dialogue data includes dialogue data constructed based on scene features of easily confused psychological scenarios, where multiple psychological scenarios with similar scene features are easily confused psychological scenarios with each other.

[0008] In one possible implementation, the process of constructing the second dialogue data includes: A psychological knowledge database and a clinical case database are constructed based on psychologically relevant search resources. The psychological knowledge database includes a tag description for at least one psychological profile tag. Different psychological profile tags represent different psychological scenarios, and the tag descriptions of the psychological profile tags represent the scenario characteristics of the psychological scenarios related to the psychological profile tags. By combining the label descriptions of the psychological profile tags with a review of relevant cases in the clinical case database, dialogue data in relevant psychological scenarios is constructed.

[0009] In one possible implementation, the process of constructing the third dialogue data includes: Based on the label descriptions of psychological profile tags related to easily confused psychological scenarios, the scene feature comparison information between easily confused psychological scenarios is determined; Based on the scene feature comparison information and combined with the language features of the easily confused psychological scene, dialogue data under the easily confused psychological scene is constructed.

[0010] In one possible implementation, the step of labeling the dialogue data with psychological profiles based on a multi-model voting mechanism and recalling at least one label of the dialogue data includes: The dialogue data is input into each of the at least one pre-trained label retrieval models to trigger the label retrieval model to annotate the dialogue data with psychological profile labels and retrieve the psychological profile labels of the dialogue data; the union of the psychological profile labels retrieved by the at least one label retrieval model constitutes the at least one label.

[0011] In one possible implementation, the recall analysis of the at least one tag, generating recall analysis results for each tag, includes: Perform rejection analysis on the at least one tag to generate a first analysis result for each tag; the first analysis result indicates whether the tag is a rejection tag or a non-rejection tag. The target label is analyzed by combining scene feature comparison information between easily confused psychological scenarios to generate a second analysis result of the target label; the target label is a non-rejection label representing an easily confused psychological scenario among the at least one label, and the second analysis result indicates whether the psychological scenario related to the target label matches or does not match the dialogue data; Based on the first analysis result and the second analysis result, a recall analysis result is generated for each of the at least one tag.

[0012] In one possible implementation, the recall analysis results based on the at least one tag, and the thought process of performing psychological profile tag analysis on the dialogue data using the thought chain generation model, generate psychological profile data related to the thought process, including: The dialogue data is input into the thought chain generation model to instruct the thought chain generation model to perform psychological profile label analysis on the dialogue data and output the thought chain data and label analysis results of the dialogue data. Determine whether the tag analysis result matches the correct tag indicated by the recall analysis result of the at least one tag; If the tag analysis result matches the correct tag indicated by the recall analysis result of the at least one tag, the thought chain data and the tag analysis result are determined as the psychological profile data of the dialogue data.

[0013] In one possible implementation, if the tag analysis result does not match the correct tag indicated by the recall analysis result of the at least one tag, the method further includes: Determine whether the number of times the current thought chain generation model outputs the thought chain data and tag analysis results of the dialogue data has reached the preset target number; If the number of times reaches the target number, then psychological profile data of the dialogue data is generated based on the correct label among the at least one label; If the number of attempts does not reach the target number of attempts, then the difference information of the correct label indicated by the label analysis result relative to the recall analysis result of the at least one label is determined; The difference information and the dialogue data are input into the thought chain generation model to instruct the thought chain generation model to perform psychological profile tag analysis on the dialogue data in combination with the difference information, output the thought chain data and tag analysis results of the dialogue data; and return to execute the step of "determining whether the tag analysis results match the recall analysis results of at least one tag".

[0014] The second aspect of this application provides a method for psychological profiling, including: Dialogue data from psychological counseling scenarios is input into a psychological profiling model so that the psychological profiling model can determine the psychological profile of the counseling client based on the dialogue data. The psychological profile model is trained based on psychological profile data generated by the psychological profile data generation method described in the first aspect.

[0015] A third aspect of this application provides a psychological profile data generation apparatus, comprising: The dialogue data construction unit is used to construct at least one dialogue data point in a psychological counseling scenario. The tag recall unit is used to perform psychological profile tagging on the dialogue data based on a multi-model voting mechanism, and recall at least one tag of the dialogue data. The recall analysis unit is used to perform recall analysis on the at least one tag and generate a recall analysis result for each tag; the recall analysis result indicates whether the tag is an incorrect tag or a correct tag; The psychological profile data generation unit is used to supplement the thinking process of the thought chain generation model in performing psychological profile tag analysis on the dialogue data based on the recall analysis results of the at least one tag, and generate psychological profile data related to the thinking process.

[0016] The fourth aspect of this application provides a psychological profile recognition device for inputting dialogue data in a psychological counseling scenario into a psychological profile model, so that the psychological profile model can determine the psychological profile of the counseling client based on the dialogue data; wherein, the psychological profile model is trained based on psychological profile data generated by the psychological profile data generation method described in the first aspect.

[0017] The fifth aspect of this application provides an electronic device, including a memory and a processor; The memory is connected to the processor and is used to store programs; The processor is used to implement the psychological profile data generation method as described in the first aspect of this application or any possible implementation of the first aspect of this application, or to implement the psychological profile recognition method as described in the second aspect of this application, by running the program in the memory.

[0018] The sixth aspect of this application provides a chip including a processor and a data interface. The processor reads and runs a program stored in a memory through the data interface to execute the psychological profile data generation method as described in the first aspect of this application or any possible implementation of the first aspect of this application, or to implement the psychological profile recognition method as described in the second aspect of this application.

[0019] The seventh aspect of this application provides a computer program product, including a computer program that, when executed by a processor, implements the psychological profile data generation method as described in the first aspect of this application or any possible implementation of the first aspect of this application, or implements the psychological profile recognition method as described in the second aspect of this application.

[0020] The eighth aspect of this application provides a storage medium storing a computer program, which, when executed by a processor, implements the psychological profile data generation method as described in the first aspect of this application or any possible implementation thereof, or implements the psychological profile recognition method as described in the second aspect of this application.

[0021] This application proposes a method for generating psychological profile data, a method for recognizing psychological profiles, an apparatus, and a device. After constructing dialogue data in a psychological counseling scenario, a multi-model voting mechanism is used to label the dialogue data with psychological profile tags. This enables comprehensive recall of potentially relevant tags from the dialogue data, avoiding the impact of missed tag recall on the accuracy of psychological profile data generation from the source. Furthermore, to ensure the accuracy of the recalled tags, a recall analysis is performed on all recalled tags (at least one tag). Based on the recall analysis results, it can be determined whether the recalled tag is an incorrect or correct tag, thus achieving full recall of correct tags from the dialogue data and providing an accurate data foundation for training the psychological profile model. To enhance the generalization ability of the psychological profile model, instead of directly using dialogue data carrying all correct tags as training samples, psychological profile data related to the thought process is generated based on the recall analysis results of at least one tag and the thought process of the thought chain generation model analyzing the dialogue data for psychological profile tags. This psychological profile data is used to train the psychological profile model, providing an interpretable data foundation for the training of the psychological profile model. This achieves the goal of providing an accurate data foundation for training mental profiling models while enhancing the model's generalization ability. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of this application. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0023] Figure 1 This is a schematic diagram of an implementation system architecture provided for an embodiment of this application.

[0024] Figure 2A flowchart illustrating a method for generating psychological profile data, as provided in this application embodiment.

[0025] Figure 3 This is a schematic diagram of a first task instruction template provided in an embodiment of this application.

[0026] Figure 4 This is a schematic diagram of dialogue data and visitor information provided in an embodiment of this application.

[0027] Figure 5 This application provides a flowchart of a method for performing recall analysis on at least one tag and generating recall analysis results for each tag.

[0028] Figure 6 This is a schematic diagram of the mind chain format of a mind chain generation model provided in an embodiment of this application.

[0029] Figure 7 This application provides a flowchart of a method for generating psychological profile data related to the thinking process by performing psychological profile tag analysis on dialogue data based on recall analysis results based on at least one tag, and a thinking chain generation model, as provided in the embodiments of this application.

[0030] Figure 8 This is a schematic diagram of a second task instruction template provided in an embodiment of this application.

[0031] Figure 9 This is a schematic diagram of a psychological profile data generation device provided in an embodiment of this application.

[0032] Figure 10 This is a schematic diagram of the structure of an electronic device provided according to an embodiment of this application. Detailed Implementation

[0033] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0034] As mentioned in the background section, the training of psychological profiling models currently typically employs a large-model supervised fine-tuning approach. The model's capabilities rely heavily on high-quality labeled data, resulting in a relatively singular training method and low model generalization ability. Some mislabeled data can not only reduce the training effect of psychological profiling models but may even mislead the direction of subsequent psychological assessments and interventions.

[0035] Therefore, how to provide a method for generating psychological profile data, a method for recognizing psychological profiles, an apparatus and equipment that can provide an accurate data foundation for training psychological profile models while enhancing the generalization ability of the models is a technical problem that urgently needs to be solved in this field.

[0036] Furthermore, in addition to the aforementioned technical problems, existing technologies also have significant limitations in their methods for constructing labeled data: First, there is the labeling scheme that primarily relies on professional psychological counselors to annotate dialogue data sentence by sentence, supplemented by model annotation and manual review mechanisms. This method, which mainly depends on manual data construction, is prone to label omissions, and the labor cost of manual annotation is high when the data volume is large. Second, there is the single-model automatic labeling scheme that uses a single model such as BERT or TextCNN, trained on historical labeled data, to extract surface semantic features of the text and automatically output labels. This scheme relies on a single model for labeling and does not fully explore the deep semantic relationships between words and sentences in the dialogue text, resulting in very low labeling accuracy. Third, there is the multi-model output scheme that uses multiple different types of models to annotate in parallel and take the union, followed by manual removal of meaningless labels and correction of contradictory labels, plus manual review. This scheme still requires high time and labor costs.

[0037] Therefore, if we can reduce labor costs, ensure accurate data for training psychological profiling models, and enhance the generalization ability of the models, then large language models (LLMs) can play a crucial role in the application of psychological services.

[0038] In view of this, this application proposes a method for generating psychological profile data, a method for recognizing psychological profiles, an apparatus, and a device. After constructing dialogue data in a psychological counseling scenario, a multi-model voting mechanism is used to label the dialogue data with psychological profile tags. This enables comprehensive recall of potentially relevant tags from the dialogue data, avoiding the impact of missed tag recall on the accuracy of psychological profile data generation from the source. Furthermore, to ensure the accuracy of the recalled tags, a recall analysis is performed on all recalled tags (at least one tag). Based on the recall analysis results, it can be determined whether the recalled tag is an incorrect or correct tag, thereby achieving full recall of correct tags from the dialogue data and providing an accurate data foundation for training the psychological profile model. To enhance the generalization ability of the psychological profile model, instead of directly using dialogue data carrying all correct tags as training samples, psychological profile data related to the thinking process is generated based on the recall analysis results of at least one tag and the thought process of the thought chain generation model analyzing the dialogue data for psychological profile tags. The psychological profile data is used to train the psychological profile model, providing an interpretable data foundation for the training of the psychological profile model. Therefore, the psychological profiling data not only provides an accurate data foundation for training the psychological profiling model, but also enhances the model's generalization ability. Furthermore, this application's reliance on recall analysis significantly reduces the dependence on manual annotation, effectively lowering labor costs.

[0039] Exemplary Implementation Environment It should be understood that the psychological profile data generation method provided in this application can be applied to psychological profile data generation scenarios, and the psychological profile data generated based on the psychological profile data generation method can be used to train the psychological profile model.

[0040] The psychological profile model trained on psychological profile data is applied to the psychological profile recognition method provided in this application, and the psychological profile recognition method can be applied to psychological profile recognition scenarios.

[0041] The psychological profile recognition method provided in this application inputs dialogue data from a psychological counseling scenario into a psychological profile model, so that the psychological profile model can determine the psychological profile of the counseling client based on the dialogue data.

[0042] The psychological profile data generation method / psychological profile recognition method of this application can be applied to, for example... Figure 1 The system architecture shown can include a terminal 100 and a server 200.

[0043] It is understandable that server 200 can include one or more servers. Figure 1 (This example uses a server as an illustration).

[0044] In this embodiment, server 200 can be an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms.

[0045] In this embodiment, the terminal 100 can be a mobile phone, tablet computer, learning machine, teaching large screen, wearable device, augmented reality (AR) / virtual reality (VR) device, laptop computer, ultra-mobile personal computer (UMPC), netbook, personal digital assistant (PDA), etc. This embodiment does not impose any restrictions on it.

[0046] Terminal 100 and server 200 can be directly or indirectly connected via wired or wireless communication. Terminal 100 and server 200 can be connected to form a blockchain network, and this application does not impose any restrictions on this.

[0047] Both terminal 100 and server 200 can be used independently to execute the psychological profile data generation method / psychological profile recognition method provided in the embodiments of this application.

[0048] In addition, the terminal 100 and the server 200 can also work together to execute the psychological profile data generation method / psychological profile recognition method provided in the embodiments of this application.

[0049] It is understood that the psychological profile data generation method / psychological profile recognition method provided in the embodiments of this application can be run in the above-mentioned device as a program, or as a system component in the above-mentioned device, or as a cloud service program. The specific operation mode depends on the actual scenario and is not limited here.

[0050] Exemplary methods Figure 2 A flowchart illustrating a method for generating psychological profile data, as provided in this application embodiment.

[0051] In the case of Figure 2 Before elaborating on the method for generating psychological profile data, let's first explain the labeling specifications upon which this method relies.

[0052] In this embodiment of the application, the labeling specification includes a psychological knowledge database and a differentiation rule base.

[0053] The psychological knowledge database includes a list of tags, which contains at least one psychological profile tag. That is, the at least one psychological profile tag consists of all the psychological profile tags in the tag list. Furthermore, the psychological knowledge database also includes a tag description for each of the at least one psychological profile tag.

[0054] It is understandable that different psychological profile tags represent different psychological scenarios, and the tag descriptions of psychological profile tags represent the scenario characteristics of the psychological scenarios related to the psychological profile tags.

[0055] The label description for psychological profiling includes label definition, judgment criteria, and rejection instructions.

[0056] A label definition describes the concept of a label, and can also be understood as a description of the psychological scenario associated with the label. For example, the label "emotion - anxiety" can be defined as a short-term emotional state, the feeling of tension, unease, and worry that an individual experiences when facing potential threats, stress, or uncertainty.

[0057] The tag determination criteria represent the conditions that the dialogue data must meet to recall the tag. That is, if the dialogue data meets the tag determination criteria, the tag can be recalled for that dialogue data.

[0058] The tag rejection specification includes the tag rejection features. That is, if the dialogue data includes the tag rejection features, the tag will not be recalled for that dialogue data.

[0059] In this embodiment of the application, the psychological knowledge database is constructed based on psychologically relevant search resources.

[0060] For example, by using web crawlers, academic literature searches, and psychology book compilation, resources such as psychological theories and tag explanations can be collected to build a psychological knowledge database.

[0061] It is understandable that the psychological knowledge database can be pre-constructed based on psychologically relevant search resources. This database aggregates psychological profile tags for all possible psychological scenarios a visitor might encounter from these resources. By pre-setting this psychological knowledge database, a theoretical basis and scenario reference can be provided for the psychological profile data generation method provided in this application embodiment.

[0062] Furthermore, this application can also collect clinical cases through web crawling, academic literature retrieval, and compilation of psychology books to build a clinical case database, thereby providing linguistic feature basis for dialogue data construction. For example, based on the psychological profile tag description, relevant dialogue data can be constructed using a dialogue generation model based on the summary of clinical cases.

[0063] It is understood that the clinical case database can be pre-constructed based on psychologically relevant search resources, or it can be acquired in real time in combination with actual needs during the execution of a psychological profile data generation method provided in this application embodiment. This application embodiment does not limit this.

[0064] In this embodiment, the construction process of the psychological knowledge database can be considered as the standardization process of psychological profile tags. The standardization of psychological profile tags is the core benchmark of the psychological profile data generation method provided in this embodiment.

[0065] It should be noted that at least one tag in the psychological knowledge database can be categorized based on relevant psychological expertise and specific application scenarios.

[0066] This embodiment provides a classification method for psychological profile tags: based on the severity of the psychological problems associated with the psychological scenarios related to the psychological profile tags, the psychological profile tags in the psychological knowledge data are divided into two categories: important psychological profile tags and unimportant psychological profile tags. The psychological scenarios represented by important psychological profile tags indicate more severe psychological problems in the visitors compared to those represented by unimportant psychological profile tags.

[0067] For example, the label "Family Relationships - Poor Parent-Child Communication" indicates that the visitor's psychological problems are serious and can be classified as an important psychological profile label; the label "Extroversion - Introversion" indicates that the visitor's psychological problems are not serious and can be classified as a non-important psychological profile label.

[0068] It should be noted that the severity of a client's psychological problems as represented by an individual psychological profile label is not easily assessed. That is, it can be categorized as either an important or unimportant psychological profile label. In such cases, it's advisable to classify the label into both categories simultaneously. That is, the psychological profile label belongs to both important and unimportant categories. Important psychological profile labels can be understood as key profiles, while unimportant labels can be understood as less important profiles.

[0069] Another classification method for psychological profile tags provided in this embodiment is as follows: based on four core scenarios—campus psychology, family environment, personal psychological traits, and social interaction—psychological profile tags are classified across dimensions by integrating psychological theories and practical application scenarios.

[0070] In this embodiment of the application, the psychological profile tags in the psychological knowledge database can be constructed in layers according to seven dimensions: stressful life events, learning-related dimensions, family factors, personality traits, emotional states, interpersonal relationships, and intimate relationships.

[0071] The seven dimensions of psychological profiling can be viewed as seven categories. That is, the psychological profiling tags in the psychological knowledge database are divided into seven categories: stressful life events, learning-related dimensions, family factors, personality traits, emotional states, interpersonal relationships, and intimate relationships.

[0072] Taking a psychological profile dimension as an example, this dimension encompasses at least one constituent element. Each constituent element is considered a subcategory within this dimension, and the specific information of each constituent element is considered a specific label for that subcategory. Accordingly, the psychological profile labels within a psychological profile dimension are annotated using the format "subcategory-specific label".

[0073] Another classification method for psychological profile tags provided in this application embodiment is to combine the above two classification methods. That is, after classifying the tags in the psychological knowledge database into important psychological profile tags and unimportant psychological profile tags, important psychological profile tags can be further classified according to seven psychological profile dimensions, and unimportant psychological profile tags can also be classified according to seven psychological profile dimensions.

[0074] The psychological profile tags in the psychological knowledge database provided in this application embodiment are explained below in conjunction with seven dimensions of psychological profile.

[0075] ① Stressful life events dimension: covering specific tags such as unexpected fright / accident / natural disaster, criticism or punishment, and other stressful events.

[0076] Referring to Table 1, the following explanation of psychological profile tags for the stressful life event dimension, based on the format "subcategory-specific tag", will be provided.

[0077] Table 1 - Psychological Profile Tags for Stressful Life Events This application embodiment can clarify the label descriptions for psychological profile tags in the dimension of stressful life events by combining relevant information such as event type, duration of impact, and degree of psychological impact on individuals. The specific content of the label descriptions for psychological profiles in the dimension of stressful life events can be set by those skilled in the art based on the relevant psychological scenarios of the labels; this application embodiment does not impose any limitations on this.

[0078] ②Learning-related dimensions: These include psychological profile labels such as high learning pressure, heavy learning burden, high pressure to enter higher education, learning motivation (external / internal), test-taking strategies, and learning styles (visual / auditory / kinesthetic).

[0079] Referring to Table 2, the following explanation of the psychological profile tags for learning-related dimensions will be based on the format "subcategory-specific tag".

[0080] Table 2 - Psychological Profile Tags for Learning-Related Dimensions In this embodiment, the label descriptions for the psychological profile tags in the learning-related dimensions can be clarified by combining information such as learning duration, performance fluctuations, and descriptions of learning behaviors. The specific content of the psychological profile label descriptions for the learning-related dimensions can be set by those skilled in the art based on the relevant psychological scenarios, and this embodiment does not impose any limitations.

[0081] ③ Family Factors Dimension: This includes psychological profile labels such as low family intimacy, family structure (divorced family), poor parent-child communication, excessively high parental expectations, parenting style (authoritarian / authoritative / indulgent / permissive), low quality of parent-child relationship, and negative coping styles of parents / individuals.

[0082] Referring to Table 3, the following explanation of the psychological profile labels for the family factor dimension, based on the format "subcategory-specific label", will be provided.

[0083] Table 3 - Family Factors Dimension Psychological Profile Labels This application embodiment can combine information such as family structure, interaction patterns, and parental behavior to clarify the label descriptions for psychological profile tags in the family factor dimension. The specific content of the psychological profile label descriptions in the family factor dimension can be set by those skilled in the art based on the relevant psychological scenarios, and this application embodiment does not impose any limitations.

[0084] ④ Personality trait dimensions: covering psychological profile labels such as attachment (secure / insecure), self-efficacy - low, perfectionism tendency - negative perfectionism, self-control - weak, psychological resilience (poor psychological resilience / protective factors), extraversion (extroversion / introversion), attribution style - negative attribution, and problem orientation - negative problem orientation.

[0085] Referring to Table 4, the following explanation of the psychological profile labels for personality trait dimensions will be based on the format "subcategory-specific label".

[0086] Table 4 - Personality Trait Dimensions and Psychological Profile Labels This application embodiment can combine long-term behavioral patterns, psychological tendencies, and other relevant information to clarify the label descriptions of psychological profile tags in personality trait dimensions. The specific content of the psychological profile label descriptions in personality trait dimensions can be set by those skilled in the art based on the relevant psychological scenarios, and this application embodiment does not impose any limitations.

[0087] ⑤ Emotional state dimension: includes psychological profile labels such as emotions (anxiety / loneliness / inferiority / learned helplessness), psychological feelings - frustration, and emotional regulation style - expression inhibition.

[0088] Referring to Table 5, the following explanation of the psychological profile labels for the emotional state dimension will be based on the format "subcategory-specific label".

[0089] Table 5 - Labels for Psychological Profiles of Emotional State Dimensions In this embodiment, the label descriptions for the psychological profile tags in the emotional state dimension can be clarified by combining information such as duration, intensity, and triggering scenario. The specific content of the psychological profile label descriptions in the emotional state dimension can be set by those skilled in the art based on the relevant psychological scenarios, and this embodiment does not impose any limitations.

[0090] ⑥ Interpersonal Relationship Dimension: Covers psychological profile labels such as poor peer relationships, low level of social support, low level of family support, and high level of interpersonal stress.

[0091] Referring to Table 6, the following explanation of the psychological profile labels for the interpersonal relationship dimension will be based on the format "subcategory-specific label".

[0092] Table 6 - Psychological Profile Labels for Interpersonal Relationship Dimension This application embodiment can combine information such as interaction frequency, conflict events, and others' evaluations to clarify the label descriptions for psychological profile tags in the interpersonal relationship dimension. The specific content of the psychological profile label descriptions in the interpersonal relationship dimension can be set by those skilled in the art based on the relevant psychological scenarios of the labels; this application embodiment does not impose any limitations on this.

[0093] ⑦ Intimate Relationship Related Dimensions: Covers psychological profile tags such as immature intimate relationships (affecting learning / other problems).

[0094] Referring to Table 7, the following explanation of the psychological profile labels for intimate relationship-related dimensions will be based on the format "subcategory-specific label".

[0095] Table 7 - Psychological Profile Labels for Intimate Relationship Related Dimensions In this embodiment, the label descriptions for psychological profile tags in intimate relationship-related dimensions can be clarified by combining information such as intimate relationship patterns and affected events. The specific content of the psychological profile label descriptions in intimate relationship-related dimensions can be set by those skilled in the art based on the relevant psychological scenarios, and this embodiment does not impose any limitations.

[0096] Based on the psychological profile tags involved in the seven dimensions mentioned above, this application provides a classification diagram of key profile tags and non-key profile tags, as shown in Table 8.

[0097] Table 8 - Categorization of Key and Non-Key Psychological Profiling Tags As shown in Table 8, the severity of the psychological problems represented by the "Poor Peer Relationship" psychological profile label is not easy to assess. Therefore, while classifying this label as a key psychological profile label, it is also classified as a non-key psychological profile label.

[0098] The above is a detailed description of the relevant content in the psychological knowledge database in the embodiments of this application. Specifically, the tags in the psychological knowledge database are not limited to the tags mentioned above. Those skilled in the art can set the tags of the psychological knowledge database according to their needs. This application embodiment does not limit them.

[0099] The following section, based on the above explanation of psychological knowledge databases, will explain the distinction rule base.

[0100] The distinction rule base includes each group of easily confused tags in the tag list, as well as the scene feature comparison information for each group of easily confused tags. A group of easily confused tags includes psychological profile tags with similar scene features from the psychological knowledge database. Psychological profile tags with similar scene features are easily confused tags with each other, and psychological scenes with similar scene features are easily confused psychological scenes with each other.

[0101] Taking a set of easily confused tags as an example, the scene feature comparison information of this set of easily confused tags includes: the same scene features and different scene features among the easily confused tags in this set.

[0102] It is understandable that the label descriptions of psychological profile tags represent the scene characteristics of the scenarios related to the psychological profile tags. Therefore, the scene characteristic comparison information of this group of easily confused tags specifically includes: the commonalities and differences in the label descriptions of this group of easily confused tags. The commonalities represent the same scene characteristics among the easily confused tags in this group, while the differences represent the different scene characteristics among the easily confused tags in this group.

[0103] The following section, in conjunction with the description of the psychological knowledge database and the differentiation rule base, elaborates on a psychological profile data generation method provided in this application embodiment. See details below. Figure 2 .

[0104] like Figure 2 As shown, the method for generating psychological profile data provided in this application includes the following steps: S201. Constructing dialogue data in psychological counseling scenarios; In this embodiment of the application, the dialogue data includes at least one of first dialogue data, second dialogue data, and third dialogue data.

[0105] In this embodiment of the application, the first dialogue data includes real online psychological counseling dialogue data.

[0106] The collection of first-conversation data mainly involves collecting real online psychological counseling dialogue data. First-conversation data comes from real psychological counseling scenarios and is more in line with the real environment.

[0107] Understandably, the online real-world psychological counseling dialogue data has a wide coverage and rich variety. Although casual conversations, jokes, role-playing, or fabricated stories in the psychological counseling dialogue data may introduce noise into the collection of the first dialogue data, in order to enhance the generalization ability of the psychological profile generation model, it is not necessary to remove noise from the collected online real-world psychological counseling dialogue data. Instead, the collected online real-world psychological counseling dialogue data is directly used as the first dialogue data.

[0108] In this embodiment of the application, the second dialogue data includes dialogue data constructed based on psychologically relevant search resources.

[0109] The process of constructing the second dialogue data includes: determining a psychological knowledge database and a clinical case database based on psychologically relevant retrieval resources; the psychological knowledge database includes a tag description for at least one psychological profile tag; different psychological profile tags represent different psychological scenarios, and the tag descriptions of psychological profile tags represent the scenario characteristics of the psychological scenarios related to the psychological profile tags; and constructing dialogue data under the relevant psychological scenarios by combining the tag descriptions of psychological profile tags and the reviews of relevant cases in the clinical case database.

[0110] The contents of the clinical case database and the psychological knowledge database have already been explained in detail above, and will not be repeated here.

[0111] Understandably, by combining the tag descriptions of one or more psychological profile tags in a psychological knowledge database with reviews of relevant cases in a clinical case library, dialogue data in relevant psychological scenarios can be constructed.

[0112] Taking the construction of dialogue data as an example, when constructing dialogue data, one can rely on one or more psychological profile tags in the psychological knowledge database, and combine the tag descriptions of the tags it relies on with the review of relevant cases in the clinical case library, with the aim of constructing dialogue data in the relevant psychological scenarios of the tags it relies on, and thus complete the construction of dialogue data.

[0113] In this embodiment of the application, the third dialogue data includes dialogue data constructed based on scene features of easily confused psychological scenes, and multiple psychological scenes with similar scene features are easily confused psychological scenes with each other.

[0114] The process of constructing the third dialogue data includes: determining the scene feature comparison information between easily confused psychological scenes based on the label descriptions of psychological profile labels related to easily confused psychological scenes; and constructing dialogue data under easily confused psychological scenes by referring to the scene feature comparison information and combining the language features of easily confused psychological scenes.

[0115] In this embodiment of the application, the distinction rule base includes each group of easily confused tags in the tag list, and the scene feature comparison information of each group of easily confused tags. The scene feature comparison information of a group of easily confused tags can be regarded as the scene feature comparison information between the corresponding easily confused psychological scenes.

[0116] Taking a set of easily confused tags as an example, for each tag in the set of easily confused tags, based on the tag description, the language features of the psychological scene related to the tag, and the scene feature comparison information of the easily confused tag group to which the tag belongs, the goal is to construct dialogue data under the psychological scene related to the tag, and complete the dialogue data construction.

[0117] For example, linguistic features of easily confused psychological scenarios can be considered as keywords in dialogue data within such scenarios. The above are merely preferred embodiments of linguistic features provided in this application; those skilled in the art can set the specific content of the linguistic features according to their own needs, and no limitations are imposed here.

[0118] It should be noted that the second and third dialogue data can be constructed based on the dialogue data generation model. The significance of constructing dialogue data based on the psychological knowledge database and the differentiation rule base is to make the psychological scenarios expressed by the constructed dialogue data as close as possible to the psychological scenarios related to the psychological profile labels on which the data was constructed.

[0119] S202. Based on a multi-model voting mechanism, psychological profile labels are used to label the dialogue data, and at least one label of the dialogue data is recalled. In this embodiment of the application, the dialogue data is respectively input into each of the at least one pre-trained label recall model to trigger the label recall model to label the dialogue data with psychological profiles and recall the psychological profiles of the dialogue data; the union of the psychological profiles recalled by at least one label recall model constitutes at least one label recalled by the multi-model voting mechanism.

[0120] It is understandable that by using a multi-model voting mechanism to label dialogue data with psychological profiles, we can obtain the label recall results of the dialogue data. The label recall results of the dialogue data include at least one label of the dialogue data recalled by the multi-model voting mechanism.

[0121] As a preferred embodiment of this application, at least one tag recall model includes multiple tag recall models.

[0122] Taking a tag retrieval model as an example, dialogue data is input into the tag retrieval model to trigger it to annotate the dialogue data with psychological profile tags and retrieve these tags. This includes: generating a first task instruction based on the dialogue data, visitor information, and a psychological knowledge database. This first task instruction instructs the large language model to retrieve psychological scene tags matching the dialogue data from the psychological knowledge database. The first task instruction is then input into a pre-trained tag retrieval model, enabling it to retrieve psychological scene tags matching the dialogue data by executing the first task instruction. It is understood that the psychological knowledge database involved here specifically refers to the list of tags within the psychological knowledge database, along with the tag description for each tag in the list.

[0123] Furthermore, the tag recall model, while recalling mental profile tags that match the dialogue data, also outputs a profile analysis of the recalled mental profile tags. This profile analysis includes the reasons for the recall of the mental profile tags.

[0124] In this embodiment of the application, the visitor information in the dialogue data includes the visitor's identity information. The visitor's identity information may include the visitor's educational level (primary / junior / high school / university), whether the visitor is an adult (adult / minor), etc.

[0125] The above are merely preferred contents of visitor information provided in the embodiments of this application. Those skilled in the art can set the specific contents of visitor information according to their own needs, and no limitation is made here.

[0126] In this embodiment of the application, a first task instruction is generated based on dialogue data, visitor information from the dialogue data, and a psychological knowledge database. This includes: determining a first task instruction template (prompt template), which includes a first task prompt, profile tag filling positions, tag description filling positions, visitor information filling positions, and dialogue data filling positions; and filling the psychological profile tag list from the psychological knowledge database, the tag descriptions of each psychological profile tag in the psychological profile tag list, the visitor information from the dialogue data, and the dialogue data into the corresponding filling positions of the first task instruction template to generate the first task instruction.

[0127] Figure 3 This is a schematic diagram of a first task instruction template provided in an embodiment of this application.

[0128] Figure 3 The phrase "predicting profile labels based on visitor information" recorded in the document can be considered the first task prompt. The profile scope in the first task prompt consists of the mental profile labels filled in at the profile label positions.

[0129] Figure 3 The image tag section is the image tag filling position, used to fill the list of psychological image tags in the psychological knowledge base; Figure 3 The label description section is where the label description is filled in, used to fill in the label description for each psychological profile label in the psychological profile label list; Figure 3 The "Notes" section is the area for filling in visitor information, used to populate visitor information in the dialogue data; Figure 3 The dialog data field is the location for filling in dialog data.

[0130] Figure 3 The content of the mental profile label prediction section specifies the output format for the recalled mental profile labels. (Reference) Figure 3 It is known that this output format requires outputting the mental profile label along with the mental profile label analysis (i.e., the reason for recalling the mental profile label) at the same time.

[0131] Figure 4 This is a schematic diagram illustrating dialogue data and visitor information provided in an embodiment of this application. (Combined with...) Figure 4 It is known that the visitor information includes the visitor's educational level, and the dialogue data can be real, desensitized dialogues. A real, desensitized dialogue (a piece of dialogue data) can include one round or multiple rounds of dialogue between the visitor and the counselor.

[0132] In this embodiment of the application, the training effects of the pre-trained label recall models are different. Multiple label recall models are used to recall psychological scene labels for the same dialogue data. The labels recalled by each label recall model are combined to obtain the label recall result of the multi-model voting mechanism for the dialogue data. All labels in the label recall result are called at least one label. At least one label includes all labels that may be related to the psychological scene expressed by the visitor through the dialogue data and recalled from the psychological knowledge database.

[0133] As a preferred embodiment of this application, the pre-trained multiple label recall models include multiple first label recall models and multiple second label recall models. The first label recall models are used to recall important psychological scene labels of dialogue data, and the second label recall models are used to recall non-important psychological scene labels of dialogue data.

[0134] The system retrieves important psychological scenario tags from dialogue data using a first-label recall model and non-important psychological scenario tags using a second-label recall model. By combining the tags retrieved by multiple first-label and second-label recall models, a multi-model voting mechanism is obtained to determine the tag retrieval results for the dialogue data. Further improving the accuracy of the tag retrieval results by distinguishing between key and non-key psychological profile tags is also achieved.

[0135] In this embodiment, multiple label recall models output psychological profile labels for dialogue data in parallel and independently, and the union of these outputs yields the label recall results for the dialogue data. This method of integrating the label recall results of all label recall models after multiple label recall models output psychological profile labels, through "union," forms a high-recall label dataset covering multiple dimensions such as emotion, family, personality, and interpersonal relationships, providing comprehensive candidate labels for subsequent recall analysis.

[0136] S203. Perform recall analysis on at least one label and generate recall analysis results for each label; the recall analysis results indicate whether the label is an incorrect label or a correct label; In this embodiment of the application, after generating the tag recall result (at least one tag) of the dialogue data based on the multi-model voting mechanism, recall analysis can be performed on each tag in the at least one tag to obtain the recall analysis result of the tag.

[0137] It is understandable that recall analysis includes first recall analysis and second recall analysis.

[0138] The first recall analysis is used to analyze whether the psychological profile label is a rejection label; if the psychological profile label is a rejection label, the psychological profile label fails the first recall analysis; if the psychological profile label is a non-rejection label, the psychological profile label passes the first recall analysis.

[0139] The second recall analysis is used to analyze the recall accuracy of the easily confused mental profile labels that passed the first recall analysis; if the easily confused mental profile labels are recalled accurately, then the easily confused mental profile labels pass the second recall analysis; if the easily confused mental profile labels are not recalled accurately, then the easily confused mental profile labels fail the second recall analysis.

[0140] If the psychological scenario associated with the easily confused psychological profile tag is one of the psychological scenarios expressed in the dialogue data, it indicates that the easily confused psychological profile tag recall is accurate; if the psychological scenario associated with the easily confused psychological profile tag is not one of the psychological scenarios expressed in the dialogue data, it indicates that the easily confused psychological profile tag recall is inaccurate.

[0141] Furthermore, the recall analysis may also include a third recall analysis, which is used to perform manual quality checks on non-confusing labels that passed the first recall analysis and confusing labels that passed the second recall analysis. Taking a label that undergoes manual quality check as an example, if the manual feedback indicates that the label passes the quality check, then the label is confirmed to have passed the third recall analysis; if the manual feedback indicates that the label fails the quality check, then the label is confirmed to have failed the third recall analysis.

[0142] It is understandable that manual quality inspection can not only inspect each label in the non-confusing labels that passed the first recall analysis and each label in the confusing labels that passed the second recall analysis separately, but also conduct random inspections of these labels, which is not limited in this application.

[0143] It is understood that performing a first recall analysis on a tag can generate a first analysis result for the tag, performing a second recall analysis on a tag can generate a second analysis result for the tag, and performing a third recall analysis on a tag can generate a third analysis result for the tag.

[0144] Furthermore, the first analysis results include the first analysis conclusion and the first analysis description of the label; the second analysis results include the second analysis conclusion and the second analysis description of the label; and the third analysis results include the third analysis conclusion and the third analysis description of the label.

[0145] The recall analysis results of tags indicate whether a tag is an incorrect tag or a correct tag. The tag recall analysis results include the tag itself and the analysis results of each recall analysis that the tag actually went through.

[0146] Understandably, by combining the recall analysis results of the tags, it is determined whether the tag has passed the last recall analysis it actually went through; if it has passed, the recall analysis result of the tag indicates that the tag is a correct tag; if it has not passed, the recall analysis result of the tag indicates that the tag is an incorrect tag.

[0147] According to the embodiments of this application, based on the recall analysis process, the accuracy of label recall can be effectively improved while reducing reliance on manual quality inspection.

[0148] For details on the specific implementation of step S203, please refer to [link / reference]. Figure 5 This will not be elaborated upon here.

[0149] S204. Based on the recall analysis results of at least one label, and the thinking process of the thought chain generation model to perform psychological profile label analysis on the dialogue data, psychological profile data related to the thinking process is generated; the psychological profile data is used to train the psychological profile model.

[0150] In this embodiment, a thought chain generation model is pre-set. The thought process format of the thought chain generation model is "summarizing the dialogue + main dimension evaluation + specific tag analysis + comparison analysis of easily confused tags + reflection and confirmation". Dialogue data, visitor information, and thought chains are input into the thought chain generation model to trigger the model to perform psychological profiling analysis on the dialogue data and output thought chain data and tag analysis results. The tag analysis results include psychological profile tags of the dialogue data predicted by the thought chain generation model.

[0151] Determine the correct label indicated by the recall analysis results of at least one label. The correct label can be understood as: the correct label of the dialogue data determined after performing recall analysis on at least one label of the dialogue data recalled by the multi-model voting mechanism.

[0152] For a given dialogue data set, if the label analysis result predicted by the thought chain generation model matches the correct label indicated by the recall analysis result of at least one label, then the thought chain data output by the thought chain generation model and the label analysis result are used as the mental profile data of the dialogue data. This mental profile data can be considered high-quality slow-thinking data.

[0153] If the label analysis result predicted by the thought chain generation model is the same as the correct label indicated by the recall analysis result of at least one label, then it is determined that the label analysis result predicted by the thought chain generation model matches the correct label indicated by the recall analysis result of at least one label; if the label analysis result predicted by the thought chain generation model is not the same as the correct label indicated by the recall analysis result of at least one label, then it is determined that the label analysis result predicted by the thought chain generation model does not match the correct label indicated by the recall analysis result of at least one label.

[0154] Furthermore, when it is initially determined that the label analysis results and the recall analysis results of at least one label do not match the correct label, the difference information between the label analysis results and the correct label can be compared. This difference information is then used as input to the thought chain generation model, triggering the model to initiate a new round of thinking and outputting the thought chain data and label analysis results of the new round of thinking. If the label analysis results of the new round of thinking match the correct label, the thought chain data and label analysis results of the new round of thinking are used as the psychological profile data of the dialogue data. If the label analysis results of the new round of thinking do not match the correct label, the correct label is used as the psychological profile data of the dialogue data.

[0155] Figure 6 This is a schematic diagram of the mind chain format of a mind chain generation model provided in an embodiment of this application.

[0156] like Figure 6 The "This dialogue mainly..." section shown summarizes the dialogue process; for example... Figure 6 The content shown, "Next, I will analyze all dimensions one by one...excluding related tags," outlines the process of evaluating main dimensions and analyzing specific tags; for example... Figure 6 The content shown, "Then I compared the potentially confusing labels....therefore...", describes the process of comparing and analyzing potentially confusing labels; for example... Figure 6 The "Finally...therefore does not conform to the XXX label" content shown represents a reflective determination process. This process primarily involves analyzing instances of label rejection; for example... Figure 6 The answer provided can be used to output tag analysis results.

[0157] Understandably, the label analysis results include mental profile labels for the dialogue data predicted by the thought chain generation model. The recall analysis results for each label within at least one label characterize whether the label is a correct or incorrect label. Therefore, based on the recall analysis results for each label within at least one label, the correct label within at least one label can be determined.

[0158] Compare the label analysis results with the correct labels; if they are the same, it means that the label analysis results match the correct labels indicated by the recall analysis results of at least one label; if they are different, it means that the label analysis results do not match the correct labels indicated by the recall analysis results of at least one label.

[0159] For details on the specific implementation of step S204, please refer to [link / reference]. Figure 7 This will not be elaborated upon here.

[0160] This application proposes a method for generating psychological profile data. After constructing dialogue data in a psychological counseling scenario, a multi-model voting mechanism is used to label the dialogue data with psychological profile tags. This enables comprehensive recall of potentially relevant tags from the dialogue data, avoiding the impact of missed tag recall on the accuracy of psychological profile data generation from the source. Furthermore, to ensure the accuracy of the recalled tags, a recall analysis is performed on all recalled tags (at least one tag). Based on the recall analysis results, it can be determined whether the recalled tag is an incorrect or correct tag, thus achieving full recall of correct tags from the dialogue data and providing an accurate data foundation for training the psychological profile model. To enhance the generalization ability of the psychological profile model, instead of directly using dialogue data carrying all correct tags as training samples, this application generates psychological profile data related to the thought process based on the recall analysis results of at least one tag and the thought process of the thought chain generation model analyzing the dialogue data for psychological profile tags. This psychological profile data is used to train the psychological profile model, providing an interpretable data foundation for the model training. Therefore, this method achieves the goal of providing an accurate data foundation for psychological profile model training while simultaneously enhancing the model's generalization ability.

[0161] Figure 5 This application provides a flowchart of a method for performing recall analysis on at least one tag and generating recall analysis results for each tag.

[0162] like Figure 5 As shown, the method includes: S501. Perform rejection analysis on at least one label and generate a first analysis result for each label; the first analysis result indicates whether the label is a rejection label or a non-rejection label. In this embodiment, rejection analysis can be considered as a first recall analysis. Performing a first recall analysis on a tag can generate a first analysis result for the tag. Taking a single tag as an example, the first analysis result includes a first analysis conclusion and a first analysis explanation. The first analysis conclusion indicates whether the tag is a rejected tag or a non-rejected tag, and the first analysis explanation indicates the reason for the first analysis conclusion.

[0163] In this embodiment, rejection analysis is performed on at least one tag to generate a first analysis result for each tag, including: determining a second task instruction template (prompt template), the second task instruction template including a second task prompt, a profile tag description filling position, a visitor's learning stage filling position, a dialogue data filling position, and a visitor's profile tag filling position; the tag description of at least one tag of the dialogue data recalled by the multi-model voting mechanism, the visitor's learning stage of the dialogue data, the dialogue data, and at least one tag of the dialogue data recalled by the multi-model voting mechanism are respectively filled into the corresponding filling positions of the second task instruction template to generate a second task instruction.

[0164] Figure 8 This is a schematic diagram of a second task instruction template provided in an embodiment of this application.

[0165] Figure 8 The message “Please read the following dialogue carefully...and analyze the reasons” recorded in the text can be considered as a hint for the second task.

[0166] Figure 8 The image tag description section is the image tag description filling position, used to fill in the tag description of at least one tag of the dialogue data recalled by the multi-model voting mechanism. Figure 8 The "Visitor Learning Segment" field is the location for filling in the visitor learning segment in the dialogue data. Figure 8 The dialog data field is the location for filling in dialog data; Figure 8 The visitor profile label field is the location for filling in visitor profile labels, used to fill in at least one label of the dialogue data recalled by the multi-model voting mechanism.

[0167] Figure 8 The content at the image tag judgment point specifies the output format for the first analysis result. (Reference) Figure 8 It is known that this output format requires outputting the corresponding profile analysis (first analysis description) while simultaneously outputting whether the output label meets the judgment (first analysis conclusion).

[0168] Understandably, if a label meets the "yes" condition, it means the label is not rejected; if a label meets the "no" condition, it means the label is rejected.

[0169] S502. Analyze the target label by combining the scene feature comparison information between easily confused psychological scenes, and generate a second analysis result of the target label; the target label is a non-rejection label that represents an easily confused psychological scene among at least one label, and the second analysis result represents whether the psychological scene related to the target label matches or does not match the dialogue data; In this embodiment, the analysis of the target tag in step S502 can be considered a second recall analysis. Performing this second recall analysis generates a second analysis result for the target tag. Taking a single target tag as an example, the second analysis result includes a second analysis conclusion and a second analysis explanation. The second analysis conclusion indicates whether the target tag's related psychological scenario matches or does not match the dialogue data, while the second analysis explanation explains the reasons for the second analysis conclusion.

[0170] In this embodiment of the application, the method of analyzing the target tag to generate a second analysis result of the target tag includes: determining the scene feature comparison information of the easily confused tag group to which the target tag belongs, analyzing the target tag based on the scene feature comparison information, and generating a second analysis result of the target tag.

[0171] If the psychological scenario related to the target label is one of the psychological scenarios expressed in the dialogue data, it means that the psychological scenario related to the target label matches the dialogue data; if the psychological scenario related to the target label is not one of the psychological scenarios expressed in the dialogue data, it means that the psychological scenario related to the target label does not match the dialogue data.

[0172] S503. Based on the first analysis result and the second analysis result, generate recall analysis results for each tag in at least one tag.

[0173] In this embodiment, the multi-model voting mechanism recalls at least one tag from the dialogue data. Taking one tag from the at least one tag as an example, refer to... Figure 5 The method shown performs recall analysis on this tag, and the process of generating the recall analysis results for this tag includes: Perform a first recall analysis on the tags to generate the first analysis results for the tags; The first analysis result based on the label determines whether the label is a rejection label; If the label is determined to be a rejected label, the first analysis result will be used as the recall analysis result for the label. If the label is determined to be a non-rejected label, determine whether the label is an easily confused label; If the label is determined to be a non-confusing label, the first analysis result will be used as the recall analysis result for the label. If the label is determined to be a confusing label (at which point the label can be considered the target label), a second recall analysis is performed on the target label to generate the second analysis result of the target label, and the first and second analysis results of the target label are used as the recall analysis result of the target label.

[0174] Furthermore, the recall analysis may also include a third recall analysis, which is used for manual quality inspection of non-confusing labels that passed the first recall analysis and confusing labels that passed the second recall analysis.

[0175] Taking a label undergoing manual quality inspection as an example, the manual inspection of this label can receive the third analysis results input by the quality inspector. The third analysis results include the third analysis conclusion and the third analysis explanation. The third analysis conclusion indicates whether the label passed the manual quality inspection; the third analysis explanation explains the reasons for the third analysis conclusion.

[0176] Furthermore, for labels that have undergone manual quality inspection, the third-party analysis results of the labels can be added to the labels that have undergone similar quality inspection. Figure 5 The method shown generates recall analysis results for this tag, in order to obtain richer recall analysis results for this tag.

[0177] Furthermore, for each of the at least one tags, the profile analysis of the tag output by the multi-model voting mechanism when recalling the tag can be added to the recall analysis results of the tag to obtain richer recall analysis results for the tag.

[0178] In this embodiment, after recalling at least one label of the dialogue data based on the multi-model voting mechanism, a three-level process of initial screening (first recall analysis), fine-grained quality inspection (second recall analysis), and sampling quality inspection (third recall analysis) is adopted. Compared with the existing technology of directly inspecting the recalled labels manually, this not only effectively reduces labor costs, but also realizes closed-loop control of label quality and enhances the model's rejection capability.

[0179] Figure 7 This application provides a flowchart of a method for generating psychological profile data related to the thinking process by performing psychological profile tag analysis on dialogue data based on recall analysis results based on at least one tag, and a thinking chain generation model, as provided in the embodiments of this application.

[0180] like Figure 7 As shown, the method includes: S701. Input the dialogue data into the thought chain generation model to instruct the thought chain generation model to perform psychological profile label analysis on the dialogue data and output the thought chain data and label analysis results of the dialogue data. In this embodiment of the application, dialogue data is input into the thought chain generation model to instruct the thought chain generation model to perform psychological profile label analysis on the dialogue data based on the thought chain, and output the thought chain data and label analysis results of the current round of thinking process.

[0181] The format of the thought chain generation model is detailed above. Figure 6 This will not be elaborated upon here.

[0182] S702. Determine whether the tag analysis result matches the correct tag indicated by the recall analysis result of at least one tag; if the tag analysis result matches the correct tag indicated by the recall analysis result of at least one tag, then proceed to step S703; if the tag analysis result does not match the correct tag indicated by the recall analysis result of at least one tag, then proceed to step S704. In this embodiment of the application, it is determined whether the tag analysis result output by the thinking chain generation model in the current round matches the correct tag indicated by the recall analysis result of at least one tag; if the tag analysis result output by the thinking chain generation model in the current round matches the correct tag indicated by the recall analysis result of at least one tag, then step S703 is executed; if the tag analysis result output by the thinking chain generation model in the current round does not match the correct tag indicated by the recall analysis result of at least one tag, then step S704 is executed.

[0183] S703. The mind chain data and tag analysis results are identified as psychological profile data for dialogue data; In this embodiment, if the tag analysis result output by the thought chain generation model in the current round matches the correct tag indicated by the recall analysis result of at least one tag, the thought chain data and tag analysis result output by the thought chain generation model in the current round are determined as the mental profile data of the dialogue data. The mental profile data is used to train the mental profile model. At this time, the tag analysis result is the tag annotation result of the dialogue data. When training the mental profile model based on the dialogue data carrying the tag analysis result, the thought chain data provides an interpretable data foundation for the training of the mental profile model, effectively enhancing the model's generalization ability.

[0184] S704. Determine whether the number of times the current thought chain generation model outputs the thought chain data and tag analysis results of the dialogue data has reached the preset target number; if the number has reached the target number, proceed to step S705; if the number has not reached the target number, proceed to step S706. In this embodiment of the application, the total number of rounds of thought chain data and tag analysis results of dialogue data output by the current thought chain generation model up to the present time can be considered as the number of times the current thought chain generation model outputs thought chain data and tag analysis results of dialogue data.

[0185] In this embodiment of the application, the target number of times can preferably be set to 1 or 2 times. The above are merely preferred values ​​of the target number of times provided by this embodiment of the application. Those skilled in the art can set the specific value of the target number of times according to their own needs, and it is not limited here.

[0186] Understandably, this applies when the target number of attempts is set to 1. Figure 7Steps S706-S707 shown can be omitted. If the label analysis result output by the first round of the thought chain generation model does not match the correct label indicated by the recall analysis result of at least one label, the correct label indicated by the recall analysis result of at least one label is directly used as the mental profile data of the dialogue data.

[0187] Understandably, when the target number of attempts is set to 2, if the label analysis result output by the thought chain generation model in the first round does not match the correct label indicated by the recall analysis result of at least one label, it does not directly use the correct label indicated by the recall analysis result of at least one label as the mental profile data of the dialogue data. Instead, it determines the difference information between the label analysis result output by the thought chain generation model in the first round and the correct label indicated by the recall analysis result of at least one label. Then, based on the difference information, it supplements the thought chain generation model's thought process of performing mental profile label analysis on the dialogue data, so as to trigger the thought chain generation model to output the second round of thought chain data and label analysis results.

[0188] S705. Generate psychological profile data of dialogue data based on the correct label in at least one of the labels; In this embodiment, if the number of times the current thought chain generation model outputs thought chain data and tag analysis results of the dialogue data has reached a preset target number, then the correct tag from at least one tag is determined as the mental profile data of the dialogue data, and the mental profile data is used to train the mental profile model. At this time, the correct tag is the tag annotation result of the dialogue data. Training the mental profile model based on the dialogue data carrying the correct tag can ensure the training effect of the mental profile generation model.

[0189] S706. Determine the difference information between the label analysis results and the recall analysis results of at least one label indicating the correct label; In this embodiment of the application, the difference information between the tag analysis result output by the mind chain generation model in the current round and the correct tag indicated by the recall analysis result of at least one tag is determined.

[0190] It is understood that determining the difference information between the label analysis results and the correct labels indicated by the recall analysis results of at least one label includes: determining the difference labels between the label analysis results and the correct labels indicated by the recall analysis results of at least one label, and obtaining the recall analysis results of the difference labels from the recall analysis results of at least one label; the difference labels and the recall analysis results of the difference labels constitute the difference information.

[0191] S707. Input the difference information and dialogue data into the thought chain generation model to instruct the thought chain generation model to perform psychological profile label analysis on the dialogue data in combination with the difference information, output the thought chain data and label analysis results of the dialogue data, and return to execute S702.

[0192] In this embodiment, difference information and dialogue data are input into the thought chain generation model to instruct the thought chain generation model to perform psychological profile label analysis on the dialogue data based on the thought chain and difference information, and output the thought chain data and label analysis results of the current round of thinking process.

[0193] Understandable, Figure 7 In the process of performing psychological profiling and labeling analysis on dialogue data, the thought chain generation model in China usually incorporates visitor information from the dialogue data. In other words, visitor information from the dialogue data is typically used as input data for the thought chain generation model along with the dialogue data itself; this will not be elaborated upon in the embodiments of this application.

[0194] This application provides a method for generating psychological profile data, which can construct multiple dialogue data sets in psychological counseling scenarios. The method generates psychological profile data for each dialogue data set separately, yielding individual psychological profile data for each set. A portion of the dialogue data consists of correct labels, while another portion consists of label analysis results and thought chain data. The psychological profile model is trained using the psychological profile data from all dialogue data sets. Compared to existing technologies that rely solely on high-quality labeled data, this method offers a more comprehensive training approach. Furthermore, the thought chain data provides an interpretable data foundation for training the psychological profile model, effectively enhancing its generalization ability.

[0195] Furthermore, the psychological profile tag data generation method provided in this application can also combine the thought process of tag analysis of dialogue data with thought chain data to further adjust the tag content in the psychological knowledge database and the differentiation rule base, so as to make the tag content in the psychological knowledge database and the differentiation rule base more accurate, thereby improving the accuracy of the entire psychological profile data generation method. Fundamentally, by combining thought chain data to adjust the psychological knowledge database and the differentiation rule base, the tag-related content in the various models involved in this application adaptively adjusts after these two databases are adjusted.

[0196] It is understandable that the process of adjusting the label content in the psychological knowledge database and the differentiation rule base based on the thinking chain data can be regarded as the label standard maintenance link. This link is to synchronize information such as high-frequency deviation types, deviation reasons, and text feature recognition points in the thinking chain data (slow thinking data) to the model involved in the embodiments of this application, thereby optimizing the model and further improving the accuracy of psychological profile data generation.

[0197] Furthermore, this application also provides a psychological profile recognition method, which includes: inputting dialogue data from a psychological counseling scenario into a psychological profile model, so that the psychological profile model determines the psychological profile of the client based on the dialogue data; wherein, the psychological profile model is trained on psychological profile data generated by a psychological profile data generation method provided in this application. Therefore, by using psychological profile data to train and generate the psychological profile model, the model has strong generalization ability. Thus, after receiving dialogue data from a psychological counseling scenario, the psychological profile model can output an accurate psychological profile of the client, providing accurate information support for assessing the client's psychological state.

[0198] Exemplary device Accordingly, this application also provides a psychological profile data generation device.

[0199] Please see Figure 9 In one exemplary embodiment, a psychological profile data generation apparatus 900 is provided, the psychological profile data generation apparatus 900 including: Dialogue data construction unit 901 is used to construct at least one dialogue data in a psychological counseling scenario; Tag recall unit 902 is used to perform psychological profile tagging on dialogue data based on a multi-model voting mechanism and recall at least one tag of the dialogue data. The recall analysis unit 903 is used to perform recall analysis on at least one tag and generate recall analysis results for each tag; the recall analysis results indicate whether the tag is an incorrect tag or a correct tag; The psychological profile data generation unit 904 is used to supplement the thinking process of the thought chain generation model in performing psychological profile tag analysis on dialogue data based on the recall analysis results of at least one label, and generate psychological profile data related to the thinking process.

[0200] In one possible implementation, the dialogue data includes at least one of a first dialogue data, a second dialogue data, and a third dialogue data; the first dialogue data includes real online psychological counseling dialogue data; the second dialogue data includes dialogue data constructed based on psychologically relevant search resources; and the third dialogue data includes dialogue data constructed based on the scene features of easily confused psychological scenarios, wherein multiple psychological scenarios with similar scene features are easily confused psychological scenarios with each other.

[0201] In one possible implementation, the process of constructing the second dialogue data includes: determining a psychological knowledge database and a clinical case database based on psychologically relevant retrieval resources, wherein the psychological knowledge database includes a tag description for at least one psychological profile tag; different psychological profile tags represent different psychological scenarios, and the tag descriptions of the psychological profile tags represent the scenario characteristics of the psychological scenarios related to the psychological profile tags; and constructing dialogue data under the relevant psychological scenarios by combining the tag descriptions of the psychological profile tags and the reviews of relevant cases in the clinical case database.

[0202] In one possible implementation, the process of constructing the third dialogue data includes: determining scene feature comparison information between easily confused psychological scenes based on the label descriptions of psychological profile labels related to easily confused psychological scenes; and constructing dialogue data under easily confused psychological scenes by referring to the scene feature comparison information and combining the language features of easily confused psychological scenes.

[0203] In one possible implementation, the tag recall unit is specifically used to input dialogue data into each of the at least one pre-trained tag recall models to trigger the tag recall models to annotate the dialogue data with mental profile tags and recall the mental profile tags of the dialogue data; the union of the mental profile tags recalled by at least one tag recall model constitutes at least one tag.

[0204] In one possible implementation, the recall analysis unit includes: The first analysis unit is used to perform rejection analysis on at least one tag and generate a first analysis result for each tag; the first analysis result indicates whether the tag is a rejection tag or a non-rejection tag. The second analysis unit is used to analyze the target label by combining the scene feature comparison information between easily confused psychological scenes, and generate a second analysis result of the target label; the target label is a non-rejection label that represents an easily confused psychological scene among at least one label, and the second analysis result indicates whether the psychological scene related to the target label matches or does not match the dialogue data; The result generation unit is used to generate recall analysis results for each of at least one tag based on the first analysis results and the second analysis results.

[0205] In one possible implementation, the psychological profile data generation unit includes: The first tag analysis unit is used to input dialogue data into the thought chain generation model, so as to instruct the thought chain generation model to perform psychological profile tag analysis on the dialogue data and output the thought chain data and tag analysis results of the dialogue data. The first judgment unit is used to determine whether the tag analysis result matches the correct tag indicated by the recall analysis result of at least one tag; The first data generation unit is used to determine the mind chain data and the tag analysis results as psychological profile data of the dialogue data if the tag analysis results match the correct tag indicated by the recall analysis results of at least one tag.

[0206] Furthermore, the psychological profiling data generation unit also includes: The second judgment unit is used to determine whether the number of times the current thought chain generation model outputs the thought chain data and tag analysis results of the dialogue data has reached the preset target number. The second data generation unit is used to generate psychological profile data of the dialogue data based on the correct label in at least one of the labels if the number of attempts reaches the target number of attempts. The difference information determination unit is used to determine the difference information of the correct label indicated by the label analysis result relative to the recall analysis result of at least one label if the number of attempts does not reach the target number of attempts. The second tag analysis unit is used to input the difference information and dialogue data into the thought chain generation model, so as to instruct the thought chain generation model to perform psychological profile tag analysis on the dialogue data in combination with the difference information, output the thought chain data and tag analysis results of the dialogue data, and return to execute the step of "determining whether the tag analysis results match the recall analysis results of at least one tag".

[0207] The psychological profile data generation device 900 provided in this embodiment belongs to the same application concept as the psychological profile data generation method provided in the above embodiments of this application. It can execute the psychological profile data generation method provided in any of the above embodiments of this application and has the corresponding functional modules and beneficial effects for executing the psychological profile data generation method. Technical details not described in detail in this embodiment can be found in the corresponding method embodiments of this application, and will not be repeated here.

[0208] Accordingly, this application also provides a psychological profile recognition device.

[0209] A psychological profile recognition device is used to input dialogue data in a psychological counseling scenario into a psychological profile model, so that the psychological profile model can determine the psychological profile of the counseling client based on the dialogue data; wherein, the psychological profile model is trained based on psychological profile data generated by the psychological profile data generation method provided in the embodiments of this application.

[0210] The psychological profile recognition device provided in this embodiment belongs to the same concept as the psychological profile recognition method provided in the above embodiments of this application. It can execute the psychological profile recognition method provided in any of the above embodiments of this application and has the corresponding functional modules and beneficial effects for executing the psychological profile recognition method. Technical details not described in detail in this embodiment can be found in the corresponding method embodiments of this application, and will not be repeated here.

[0211] The functions implemented by each unit in the above device can be implemented by the same or different processors, and this application embodiment does not limit this.

[0212] It should be understood that the units in the above device can be implemented by a processor calling software. For example, the device includes a processor connected to a memory containing instructions. The processor calls the instructions stored in the memory to implement any of the above methods or to implement the functions of each unit in the device. The processor can be a general-purpose processor, such as a CPU or microprocessor, and the memory can be internal or external to the device. Alternatively, the units in the device can be implemented as hardware circuits. By designing the hardware circuits, some or all of the unit functions can be implemented. The hardware circuits can be understood as one or more processors. For example, in one implementation, the hardware circuit is an ASIC, and the functions of some or all of the above units are implemented by designing the logical relationships between the components within the circuit. In another implementation, the hardware circuit can be implemented using a PLD, such as an FPGA, which can include a large number of logic gates. The connection relationships between the logic gates are configured through configuration files to implement the functions of some or all of the above units. All units in the above device can be implemented entirely by a processor calling software, entirely by hardware circuits, or partially by a processor calling software with the remaining parts implemented by hardware circuits.

[0213] In this application embodiment, a processor is a circuit with signal processing capabilities. In one implementation, the processor can be a circuit with instruction reading and execution capabilities, such as a CPU, microprocessor, GPU, or DSP. In another implementation, the processor can implement certain functions through the logical relationships of hardware circuits. These logical relationships are fixed or reconfigurable. For example, the processor may be a hardware circuit implemented as an ASIC or PLD, such as an FPGA. In a reconfigurable hardware circuit, the process of the processor loading a configuration document and configuring the hardware circuit can be understood as the processor loading instructions to implement the functions of some or all of the above units. Furthermore, it can also be a hardware circuit designed for artificial intelligence, which can be understood as an ASIC, such as an NPU, TPU, or DPU.

[0214] As can be seen, each unit in the above device can be one or more processors (or processing circuits) configured to implement the above methods, such as: CPU, GPU, NPU, TPU, DPU, microprocessor, DSP, ASIC, FPGA, or a combination of at least two of these processor forms.

[0215] Furthermore, the units in the above devices can be integrated in whole or in part, or they can be implemented independently. In one implementation, these units are integrated together and implemented in the form of a System-on-Chip (SoC). The SoC may include at least one processor for implementing any of the above methods or implementing the functions of the units in the device. The at least one processor may be of different types, such as CPU and FPGA, CPU and artificial intelligence processor, CPU and GPU, etc.

[0216] Exemplary electronic devices Another embodiment of this application also proposes an electronic device. See [link to relevant documentation]. Figure 10 As shown, the electronic device may include: a memory 1000 and a processor 1010; wherein, the memory 1000 is connected to the processor 1010 and is used to store programs; the processor 1010 is used to implement the psychological profile data generation method / psychological profile recognition method disclosed in any of the above embodiments by running the programs stored in the memory 1000.

[0217] Specifically, the aforementioned electronic device may also include: a bus, a communication interface 1020, an input device 1030, and an output device 1040.

[0218] The processor 1010, memory 1000, communication interface 1020, input device 1030, and output device 1040 are interconnected via a bus. Among them: A bus can include a pathway for transmitting information between various components of a computer system.

[0219] The processor 1010 can be a general-purpose processor, such as a general-purpose central processing unit (CPU), a microprocessor, etc., or an application-specific integrated circuit (ASIC), or one or more integrated circuits used to control the execution of the program of the present application. It can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), an off-the-shelf programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0220] The processor 1010 may include a main processor, as well as a baseband chip, modem, etc.

[0221] The memory 1000 stores a program for executing the technical solution of this application, and may also store an operating system and other critical business functions. Specifically, the program may include program code, which includes computer operation instructions. More specifically, the memory 1000 may include read-only memory (ROM), other types of static storage devices capable of storing static information and instructions, random access memory (RAM), other types of dynamic storage devices capable of storing information and instructions, disk storage, flash memory, etc.

[0222] Input device 1030 may include a device for receiving user input data and information, such as a keyboard, mouse, camera, scanner, light pen, voice input device, touch screen, pedometer, or gravity sensor.

[0223] Output device 1040 may include devices that allow information to be output to a user, such as a display screen, printer, speaker, etc.

[0224] The communication interface 1020 may include a device that uses any transceiver to communicate with other devices or communication networks, such as Ethernet, Radio Access Network (RAN), Wireless Local Area Network (WLAN), etc.

[0225] The processor 1010 executes the program stored in the memory 1000 and calls other devices, which can be used to implement each step of any of the psychological profile data generation methods / psychological profile recognition methods provided in the above embodiments of this application.

[0226] This application also proposes a chip, which includes a processor and a data interface. The processor reads and runs a program stored in the memory through the data interface to execute any of the psychological profile data generation methods / psychological profile recognition methods provided in the above embodiments. For the specific processing procedures and their beneficial effects, please refer to the description of the corresponding embodiments of the above methods.

[0227] Exemplary computer program products and storage media In addition to the methods and devices described above, embodiments of this application may also be computer program products, which include computer program instructions that, when executed by a processor, cause the processor to perform the steps in the psychological profile data generation method / psychological profile recognition method according to various embodiments of this application as described in any of the above embodiments of this specification.

[0228] The computer program product can be written in any combination of one or more programming languages ​​to perform the operations of the embodiments of this application. The programming languages ​​include object-oriented programming languages ​​such as Java and C++, as well as conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on the user's computing device, partially on the user's computing device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0229] Furthermore, embodiments of this application may also be storage media storing a computer program, which is executed by a processor in the steps of the psychological profile data generation method / psychological profile recognition method according to various embodiments of this application described in any of the above embodiments of this specification.

[0230] For the foregoing method embodiments, in order to simplify the description, they are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, because according to this application, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0231] It should be noted that the various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For apparatus embodiments, since they are basically similar to method embodiments, the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

[0232] The steps in the methods of the various embodiments of this application can be adjusted, merged, or deleted in order according to actual needs, and the technical features described in each embodiment can be replaced or combined.

[0233] The modules and sub-modules in the apparatus and terminal in the various embodiments of this application can be merged, divided, and deleted according to actual needs.

[0234] It should be understood that the disclosed terminals, devices, and methods can be implemented in other ways, given the several embodiments provided in this application. For example, the terminal embodiments described above are merely illustrative. For instance, the division of modules or sub-modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple sub-modules or modules may be combined or integrated into another module, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or modules, and may be electrical, mechanical, or other forms.

[0235] The modules or submodules described as separate components may or may not be physically separate. The components that constitute a module or submodule may or may not be physical modules or submodules; that is, they may be located in one place or distributed across multiple network modules or submodules. Some or all of the modules or submodules can be selected to achieve the purpose of this embodiment's solution, depending on actual needs.

[0236] Furthermore, the functional modules or sub-modules in the various embodiments of this application can be integrated into one processing module, or each module or sub-module can exist physically separately, or two or more modules or sub-modules can be integrated into one module. The integrated modules or sub-modules described above can be implemented in hardware or in the form of software functional modules or sub-modules.

[0237] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0238] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly by hardware, a software unit executed by a processor, or a combination of both. The software unit can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

[0239] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0240] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A psychological portrait data generation method characterized by comprising: include: Constructing dialogue data in psychological counseling scenarios; The dialogue data is labeled with psychological profile tags based on a multi-model voting mechanism, and at least one tag of the dialogue data is recalled. Recall analysis is performed on the at least one tag, and recall analysis results are generated for each tag. The recall analysis results indicate whether the label is an incorrect label or a correct label; Based on the recall analysis results of the at least one tag, and the thinking process of the thought chain generation model to perform psychological profile tag analysis on the dialogue data, psychological profile data related to the thinking process is generated. The psychological profile data is used to train the psychological profile model.

2. The method of claim 1, wherein, The dialogue data includes at least one of the first dialogue data, the second dialogue data, and the third dialogue data; The first dialogue data includes real online psychological counseling dialogue data; The second dialogue data includes dialogue data constructed based on psychologically relevant search resources; The third dialogue data includes dialogue data constructed based on scene features of easily confused psychological scenarios, where multiple psychological scenarios with similar scene features are easily confused psychological scenarios with each other.

3. The method of claim 2, wherein, The process of constructing the second dialogue data includes: A psychological knowledge database and a clinical case database are constructed based on psychologically relevant search resources. The psychological knowledge database includes a tag description for at least one psychological profile tag. Different psychological profile tags represent different psychological scenarios, and the tag descriptions of the psychological profile tags represent the scenario characteristics of the psychological scenarios related to the psychological profile tags. By combining the label descriptions of the psychological profile tags with a review of relevant cases in the clinical case database, dialogue data in relevant psychological scenarios is constructed.

4. The method according to claim 2, characterized in that, The process of constructing the third dialogue data includes: Based on the label descriptions of psychological profile tags related to easily confused psychological scenarios, the scene feature comparison information between easily confused psychological scenarios is determined; Based on the scene feature comparison information and combined with the language features of the easily confused psychological scene, dialogue data under the easily confused psychological scene is constructed.

5. The method according to claim 1, characterized in that, The step of using a multi-model voting mechanism to label the dialogue data with psychological profiles and recall at least one label from the dialogue data includes: The dialogue data is input into each of the at least one pre-trained label retrieval models to trigger the label retrieval model to annotate the dialogue data with psychological profile labels and retrieve the psychological profile labels of the dialogue data; the union of the psychological profile labels retrieved by the at least one label retrieval model constitutes the at least one label.

6. The method according to claim 1, characterized in that, The recall analysis of the at least one tag, generating recall analysis results for each tag, includes: Perform rejection analysis on the at least one tag to generate a first analysis result for each tag; the first analysis result indicates whether the tag is a rejection tag or a non-rejection tag. The target label is analyzed by combining scene feature comparison information between easily confused psychological scenarios to generate a second analysis result of the target label; the target label is a non-rejection label representing an easily confused psychological scenario among the at least one label, and the second analysis result indicates whether the psychological scenario related to the target label matches or does not match the dialogue data; Based on the first analysis result and the second analysis result, a recall analysis result is generated for each of the at least one tag.

7. The method according to claim 1, characterized in that, The recall analysis results based on the at least one tag, and the thought process of performing psychological profile tag analysis on the dialogue data by the thought chain generation model, generate psychological profile data related to the thought process, including: The dialogue data is input into the thought chain generation model to instruct the thought chain generation model to perform psychological profile label analysis on the dialogue data and output the thought chain data and label analysis results of the dialogue data. Determine whether the tag analysis result matches the correct tag indicated by the recall analysis result of the at least one tag; If the tag analysis result matches the correct tag indicated by the recall analysis result of the at least one tag, the thought chain data and the tag analysis result are determined as the psychological profile data of the dialogue data.

8. The method according to claim 7, characterized in that, If the tag analysis result does not match the correct tag indicated by the recall analysis result of the at least one tag, the method further includes: Determine whether the number of times the current thought chain generation model outputs the thought chain data and tag analysis results of the dialogue data has reached the preset target number; If the number of times reaches the target number, then psychological profile data of the dialogue data is generated based on the correct label among the at least one label; If the number of attempts does not reach the target number of attempts, then the difference information of the correct label indicated by the label analysis result relative to the recall analysis result of the at least one label is determined; The difference information and the dialogue data are input into the thought chain generation model to instruct the thought chain generation model to perform psychological profile tag analysis on the dialogue data in combination with the difference information, output the thought chain data and tag analysis results of the dialogue data; and return to execute the step of "determining whether the tag analysis results match the recall analysis results of at least one tag".

9. A method for identifying psychological profiles, characterized in that, include: Dialogue data from psychological counseling scenarios is input into a psychological profiling model so that the psychological profiling model can determine the psychological profile of the counseling client based on the dialogue data. The psychological profile model is trained based on psychological profile data generated by the method described in any one of claims 1 to 8.

10. A psychological profiling data generation device, characterized in that, include: The dialogue data construction unit is used to construct at least one dialogue data point in a psychological counseling scenario. The tag recall unit is used to perform psychological profile tagging on the dialogue data based on a multi-model voting mechanism, and recall at least one tag of the dialogue data. The recall analysis unit is used to perform recall analysis on the at least one tag and generate recall analysis results for each tag. The recall analysis results indicate whether the label is an incorrect label or a correct label; The psychological profile data generation unit is used to supplement the thinking process of the thought chain generation model in performing psychological profile tag analysis on the dialogue data based on the recall analysis results of the at least one tag, and generate psychological profile data related to the thinking process.

11. An electronic device, characterized in that, Including memory and processor; The memory is connected to the processor and is used to store programs; The processor is configured to implement the method as described in any one of claims 1 to 9 by running a program in the memory.

Images