A consultation management method and system based on language analysis

Abstract

The present application relates to the field of consulting management technology, and in particular to a consulting management method and system based on language analysis. The method includes: obtaining consulting information provided by a user, preprocessing the consulting information, and determining the preprocessing results; analyzing the preprocessing results, and determining a number of user suspicion evaluation factors; determining the user's suspicion level based on the number of user suspicion evaluation factors; and determining and outputting corresponding explanation information based on the suspicion level and consulting information. The present application determines and outputs explanation information corresponding to the suspicion level based on the user's suspicion level. By identifying and processing suspicions in the user's consulting process, it helps to establish a sense of trust in the user, and improve the user's understanding and acceptance of feedback information. In the long run, it can effectively improve user loyalty and return rate, and improve the user's experience.

Description

Technical Field

[0001] The present application relates to the field of consulting management technology, and in particular to a consulting management method and system based on language analysis. Background Art

[0002] With the development of artificial intelligence technology, intelligent consulting management systems based on language analysis technology will become an important development direction of consulting management systems. Intelligent consulting management systems can provide efficient consulting services for institutions or enterprises, quickly locate problems and provide corresponding solutions, and have huge advantages in supporting multilingual situations.

[0003] However, the existing intelligent consulting management systems have not yet gained a high degree of trust from the majority of users. When using the intelligent consulting management systems, users usually find it difficult to fully trust the intelligent consulting management systems due to various considerations. The existing intelligent consulting management systems are unable to keenly capture users' doubts, which results in the existing intelligent consulting management systems providing responses to information that users have doubts about, which cannot satisfy and convince users, and may even further deepen users' suspicion of the intelligent consulting management systems. Summary of the invention

[0004] The present application provides a consultation management method and system based on language analysis to solve the above technical problems.

[0005] In a first aspect, the present application provides a consultation management method based on language analysis, the method comprising:

[0006] Obtaining consulting information provided by a user, preprocessing the consulting information, and determining a preprocessing result;

[0007] Analyzing the preprocessing results to determine a number of user suspicion assessment factors;

[0008] Determining the user's suspicion level according to the plurality of user suspicion assessment factors;

[0009] According to the suspicion level and the consultation information, corresponding explanation information is determined and output.

[0010] The method provided by the present application pre-processes the consulting information provided by the user to determine the pre-processing results that can reflect the key information in the consulting information, improve the accuracy of understanding and response speed of the consulting information, and then analyze the pre-processing results to determine the user suspicion evaluation factors that can reflect the user's suspicion attitude, and then determine the user's suspicion level based on the user suspicion evaluation factors, and determine and output the explanation information corresponding to the suspicion level based on the user's suspicion level. By identifying and processing the suspicions in the user consultation process, it is helpful to establish the user's sense of trust and improve the user's understanding and acceptance of feedback information. In the long run, it can effectively improve the user's loyalty and return rate and improve the user's usage experience.

[0011] Optionally, preprocessing the consultation information and determining a preprocessing result includes:

[0012] Perform word segmentation processing on the consulting information to extract several consulting keywords;

[0013] Analyzing the consultation information and the plurality of consultation keywords to determine the emotional characteristics of the user;

[0014] Analyze the consultation information and extract the operation characteristics of the user;

[0015] The several consulting keywords, the emotional features and the operating features are used as the preprocessing results.

[0016] The above technical solution performs word segmentation on the consulting information, extracts consulting keywords, determines the consulting intention directly expressed in the user's consulting information and the suspicion information directly expressed by the user, and indirectly reflects the current user's skeptical attitude by determining the emotional characteristics and operational characteristics, providing an important data basis for subsequently deriving the user's degree of suspicion based on the user's emotional characteristics and operational characteristics, and optimizing the user experience and improving user satisfaction through comprehensive preprocessing and analysis methods.

[0017] Optionally, analyzing the consultation information and the plurality of consultation keywords to determine the emotional characteristics of the user includes:

[0018] Analyzing the consulting information and the several consulting keywords to determine the sentiment polarity of the user;

[0019] If the sentiment polarity is negative polarity, obtaining context information according to the consultation information;

[0020] Screening the context information to determine context emotion change information;

[0021] The emotion polarity and the context emotion change information are used as the emotion features.

[0022] The above technical solution analyzes the consulting information and several consulting keywords to determine the user's emotional polarity. When the emotional polarity is negative, the contextual information is further extracted and analyzed to obtain the user's contextual emotional change information. The emotional polarity and contextual emotional change information are used as emotional features to comprehensively and accurately analyze the user's current emotional state, providing an important data basis for the subsequent analysis of the user's skeptical attitude based on the user's emotional characteristics.

[0023] Optionally, analyzing the consultation information and extracting the operation characteristics of the user includes:

[0024] Determine a response interval time value according to the consultation information;

[0025] Determining a response time ratio according to the consultation information and the response interval time value;

[0026] Comparing the response time ratio with a preset response time ratio range to determine whether the response time ratio is greater than the preset response time ratio range;

[0027] If the response time ratio is greater than a preset response time ratio range, obtaining a user operation record corresponding to the consultation information;

[0028] The user operation record is analyzed to determine a number of pause nodes, and the number of pause nodes are used as operation features.

[0029] The above technical solution determines the response interval time value based on the consultation information, and then determines the response time ratio. By comparing the response time ratio with the preset response time ratio range, it is concluded whether the current user is skeptical. The user operation records are further analyzed, and the pause nodes in the operation records are used as operation features. This more carefully reflects the relationship between the user operation features and the user's skeptical attitude, providing a highly representative data basis for the subsequent analysis of the user's skeptical attitude based on the operation features.

[0030] Optionally, the analyzing the preprocessing result to determine a number of user suspicion assessment factors includes:

[0031] Based on the several consulting keywords, analyzing the context emotion change information, and determining the directed emotion corresponding to the emotion polarity;

[0032] Based on the directed emotion, analyzing context emotion change information, and extracting a number of tone feature words of the user;

[0033] The directed emotion and the plurality of tone feature words are used as the user suspicion assessment factors.

[0034] The above technical solution analyzes the contextual emotion change information, determines the specific emotion pointed to by the current user's emotional polarity, that is, the pointed emotion, and determines a number of tone feature words in the contextual emotion change information that conform to the pointed emotion based on the pointed emotion, and uses the pointed emotion and a number of tone feature words as user suspicion evaluation factors to comprehensively reflect the relationship between user emotions and skeptical attitudes, and further improve the accuracy of subsequent analysis of the user's degree of suspicion.

[0035] Optionally, the analyzing the preprocessing result to determine a number of user suspicion assessment factors includes:

[0036] Determine the number of pause nodes and the average pause duration according to the plurality of pause nodes;

[0037] Based on the plurality of pause nodes, analyzing the consultation information, determining a plurality of precipitation information and a plurality of correction information before and after each pause node;

[0038] The average pause duration, the number of pause nodes, the plurality of precipitation information and the plurality of correction information are used as the user suspicion assessment factors.

[0039] The above technical scheme analyzes several pause nodes to determine the number of pause nodes and the average pause duration. Through the number of pause nodes and the average pause duration, the number of pause nodes and the average pause duration are used as user suspicion evaluation factors to accurately reflect the user's suspicion attitude. Further analysis is performed to determine the precipitation information and correction information before and after the pause node, and the precipitation information and correction information are used as user suspicion evaluation factors, which is helpful for subsequent analysis of the detailed operations of the user during the pause process, and further improves the accuracy of the judgment of the degree of suspicion.

[0040] Optionally, determining the user's suspicion level according to the plurality of user suspicion assessment factors includes:

[0041] Determining the emotion intensity of the user according to the directed emotion;

[0042] Determining the frequency of modal words according to the modal feature words;

[0043] Determining an average length of the precipitation information according to the plurality of precipitation information;

[0044] Determining the number of corrections according to the correction information;

[0045] Determining an emotion influencing factor according to the emotion intensity and the tone word frequency;

[0046] Determining an operation influencing factor according to the average pause duration, the number of pause nodes, the average length of the sedimentation information, and the number of corrections;

[0047] Determining a suspicion index according to the emotion influencing factor and the operation influencing factor;

[0048] According to the suspicion index, a suspicion level is determined.

[0049] The above technical scheme combines the emotion intensity and the frequency of modal words to obtain the emotion influence factor reflecting the relationship between the user emotion and the degree of suspicion, and combines the average pause duration, the number of pause nodes, the average length of sedimented information and the number of corrections to obtain the operation influence factor reflecting the relationship between the user operation and the degree of suspicion. Then, a suspicion index quantifying the degree of user suspicion is obtained. According to the suspicion index, the user's suspicion level is accurately divided, and various factors reflecting the user's suspicion level are fully considered to ensure the comprehensiveness and accuracy of the suspicion level.

[0050] Optionally, determining the emotion influencing factor according to the emotion intensity and the tone word frequency includes:

[0051] ;

[0052] in, is the emotion influencing factor, is the emotion intercept term, is the regression coefficient of emotion intensity, For the intensity of the emotion, is the modal particle frequency regression coefficient, is the frequency of the modal particle;

[0053] The determining of the operation influencing factor according to the average pause duration, the number of pause nodes, the average length of the precipitation information and the number of corrections includes:

[0054] ;

[0055] in, is the operation influencing factor, is the intercept term, is the pause duration regression coefficient, is the average pause time, is the regression coefficient of the number of pauses, is the number of paused nodes, is the information length regression coefficient, is the average length of the precipitation information, is the corrected regression coefficient, is the number of corrections.

[0056] The above technical solution designs formulas to accurately quantify the emotion influence factors and the operation influence factors respectively, and controls the specific values ​​of the emotion influence factors and the operation influence factors within a limited range, accurately and intuitively reflecting the impact of user emotions and operations on the reflected degree of suspicion, and providing an important data basis for the subsequent quantification of the suspicion index according to the emotion influence factors and the operation influence factors.

[0057] Optionally, determining a suspicion index according to the emotion influencing factor and the operation influencing factor includes:

[0058] ;

[0059] in, is the suspicion index, is the emotional impact weight, is the emotion influencing factor, is the operation impact weight, is the operation influencing factor.

[0060] The above technical solution designs a formula to perform weighted calculation on the emotional influence factors and the operational influence factors, and obtains the suspicion index by comprehensively calculating the weighted values ​​corresponding to the emotional influence factors and the operational influence factors. By comparing the method of summing up the emotional influence factors and the operational influence factors, the different effects of the emotional influence factors and the operational influence factors on the suspicion index are more comprehensively considered, making the obtained suspicion index more accurate.

[0061] In a second aspect, the present application provides a consultation management system based on language analysis, the system comprising:

[0062] A preprocessing module, which obtains the consulting information provided by the user, preprocesses the consulting information, and determines the preprocessing result;

[0063] A factor determination module, used to analyze the preprocessing results and determine a number of user suspicion assessment factors;

[0064] A level analysis module, used to determine the user's suspicion level based on the plurality of user suspicion assessment factors;

[0065] The output module is used to determine and output corresponding explanation information according to the suspicion level and the consultation information.

[0066] Optionally, the preprocessing module is specifically used to:

[0067] Perform word segmentation processing on the consulting information to extract several consulting keywords;

[0068] Analyzing the consultation information and the plurality of consultation keywords to determine the emotional characteristics of the user;

[0069] Analyze the consultation information and extract the operation characteristics of the user;

[0070] The several consulting keywords, the emotional features and the operating features are used as the preprocessing results.

[0071] Optionally, the preprocessing module is specifically used to:

[0072] Analyzing the consulting information and the several consulting keywords to determine the sentiment polarity of the user;

[0073] If the sentiment polarity is negative polarity, obtaining context information according to the consultation information;

[0074] Screening the context information to determine context emotion change information;

[0075] The emotion polarity and the context emotion change information are used as the emotion features.

[0076] Optionally, the preprocessing module is specifically used to:

[0077] Determine a response interval time value according to the consultation information;

[0078] Determining a response time ratio according to the consultation information and the response interval time value;

[0079] Comparing the response time ratio with a preset response time ratio range to determine whether the response time ratio is greater than the preset response time ratio range;

[0080] If the response time ratio is greater than a preset response time ratio range, obtaining a user operation record corresponding to the consultation information;

[0081] The user operation record is analyzed to determine a number of pause nodes, and the number of pause nodes are used as operation features.

[0082] Optionally, the factor determination module is specifically used to:

[0083] Based on the several consulting keywords, analyzing the context emotion change information, and determining the directed emotion corresponding to the emotion polarity;

[0084] Based on the directed emotion, analyzing context emotion change information, and extracting a number of tone feature words of the user;

[0085] The directed emotion and the plurality of tone feature words are used as the user suspicion assessment factors.

[0086] Optionally, the factor determination module is specifically used to:

[0087] Determine the number of pause nodes and the average pause duration according to the plurality of pause nodes;

[0088] Based on the plurality of pause nodes, analyzing the consultation information, determining a plurality of precipitation information and a plurality of correction information before and after each pause node;

[0089] The average pause duration, the number of pause nodes, the plurality of precipitation information and the plurality of correction information are used as the user suspicion assessment factors.

[0090] Optionally, the grade analysis module is specifically used to:

[0091] Determining the emotion intensity of the user according to the directed emotion;

[0092] Determining the frequency of modal words according to the modal feature words;

[0093] Determining an average length of the precipitation information according to the plurality of precipitation information;

[0094] Determining the number of corrections according to the correction information;

[0095] Determining an emotion influencing factor according to the emotion intensity and the tone word frequency;

[0096] Determining an operation influencing factor according to the average pause duration, the number of pause nodes, the average length of the precipitation information, and the number of corrections;

[0097] Determining a suspicion index according to the emotion influencing factor and the operation influencing factor;

[0098] According to the suspicion index, a suspicion level is determined.

[0099] Optionally, the grade analysis module is specifically used to:

[0100] ;

[0101] in, is the emotion influencing factor, is the emotion intercept term, is the regression coefficient of emotion intensity, For the intensity of the emotion, is the modal particle frequency regression coefficient, is the frequency of the modal particle;

[0102] ;

[0103] in, is the operation influencing factor, is the intercept term, is the pause duration regression coefficient, is the average pause time, is the regression coefficient of the number of pauses, is the number of paused nodes, is the information length regression coefficient, is the average length of the precipitation information, is the corrected regression coefficient, is the number of corrections.

[0104] Optionally, the grade analysis module is specifically used to:

[0105] ;

[0106] in, is the suspicion index, is the emotional impact weight, is the emotion influencing factor, is the operation impact weight, is the operation influencing factor. BRIEF DESCRIPTION OF THE DRAWINGS

[0107] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0108] Figure 1 A schematic diagram of an application scenario provided for an embodiment of the present application;

[0109] Figure 2 A flowchart of a consultation management method based on language analysis provided in one embodiment of the present application;

[0110] Figure 3 A schematic diagram of the structure of a consultation management system based on language analysis provided in one embodiment of the present application. DETAILED DESCRIPTION

[0111] In order to make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0112] In addition, the term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article, unless otherwise specified, generally means that the associated objects before and after are in an "or" relationship.

[0113] The embodiments of the present application are further described in detail below in conjunction with the drawings in the specification.

[0114] The existing intelligent consulting management systems have not yet gained a high degree of trust from the majority of users. When using the intelligent consulting management systems, users usually find it difficult to fully trust the intelligent consulting management systems due to various considerations. The existing intelligent consulting management systems are unable to keenly capture users' doubts, which results in the existing intelligent consulting management systems providing responses to information that users have doubts about, which cannot satisfy and convince users, and may even further deepen users' suspicion of the intelligent consulting management systems.

[0115] Based on this, the present application provides a consulting management method and system based on language analysis, which pre-processes the consulting information provided by the user to determine the pre-processing results that can reflect the key information in the consulting information, improve the accuracy of understanding and response speed of the consulting information, and then analyze the pre-processing results to determine the user suspicion evaluation factors that can reflect the user's suspicion attitude, and then determine the user's suspicion level based on the user suspicion evaluation factors, and determine and output the explanation information corresponding to the suspicion level based on the user's suspicion level. By identifying and processing the suspicions in the user consultation process, the user feels valued, which helps to establish a sense of trust between the consulting management system and the user, and improve the user's understanding and acceptance of the information provided by the consulting management system. In the long run, it can effectively improve user loyalty and return rate, and improve the user's usage experience.

[0116] Figure 1 A schematic diagram of an application scenario provided for the present application. When a user needs consulting services, the method provided by the present application is applied to analyze the consulting information provided by the user, and to provide the user with explanatory information that can dispel the user's doubts.

[0117] Specifically, the method provided in the present application is applied to any server. The server analyzes the consulting information provided by the user, determines the preprocessing results that can reflect the key information in the consulting information, improves the accuracy of understanding and response speed of the consulting information, and then determines the user suspicion evaluation factors that reflect the user's skeptical attitude. Then, based on the user suspicion evaluation factors, the user's suspicion level is determined, and based on the user's suspicion level, the explanation information corresponding to the suspicion level is determined and output. By identifying and processing the suspicions in the user consultation process, the user feels valued, which helps to establish a sense of trust between the consulting management system and the user, and improves the user's understanding and acceptance of the information provided by the consulting management system. In the long run, it can effectively improve user loyalty and return rate, and improve the user experience.

[0118] For specific implementation methods, please refer to the following embodiments.

[0119] Figure 2 This is a flowchart of a consultation management method based on language analysis provided in an embodiment of the present application. The method of this embodiment can be applied to the server in the above scenario. Figure 2 As shown, the method includes:

[0120] S201. Obtain consulting information provided by a user, pre-process the consulting information, and determine a pre-processing result.

[0121] The consultation information can be considered as text information to be consulted input by the user.

[0122] The preprocessing result can be considered as a series of key information in the consulting information obtained after analyzing the consulting information through text analysis tools.

[0123] Specifically, during the user consultation process, the user will provide a series of consultation information and obtain reply information corresponding to the consultation information, so as to solve the problem the user wants to consult through the reply information. However, the consultation information provided by the user usually contains redundant information that is irrelevant to the user's core demands. The appearance of this redundant information is due to the characteristics of natural language and the user's language expression habits. In the process of interpersonal communication, this redundant information helps the parties communicating with each other to deepen their understanding. However, in the process of user interaction with the system, the redundant information will affect the accuracy of the consultation information analysis, and at the same time prolong the response time, bringing a negative experience to the user.

[0124] Therefore, after obtaining the consulting information provided by the user, it is necessary to analyze the consulting information through text analysis tools to extract key information and improve the accuracy of understanding and response speed of the user's consulting information.

[0125] In some other implementations, the consulting information provided by the user may be voice information, which is converted into text information through a voice recognition tool, and then pre-processed through a text analysis tool to improve user experience and enhance adaptability.

[0126] S202: Analyze the preprocessing results and determine several user suspicion assessment factors.

[0127] The user skepticism evaluation factor can be considered as a key factor that can reflect the user's skeptical attitude.

[0128] Specifically, when users consult, they may doubt the feedback information. If the user's doubts are not responded to in a timely and accurate manner, the user's trust will continue to decline, which will have a negative impact on the user's experience and even cause the user to have negative emotions.

[0129] Therefore, in the user consultation process, it is necessary to analyze and determine in real time the user suspicion evaluation factors that can reflect the user's skeptical attitude. The user suspicion evaluation factors are usually related to the detailed content in the user consultation information. The user will directly or indirectly express suspicion in the consultation information. The preprocessing results are analyzed from the two aspects of emotion and operation through sentiment analysis tools and log management technology, and the user suspicion evaluation factors are accurately and comprehensively derived.

[0130] S203: Determine the user's suspicion level based on a number of user suspicion evaluation factors.

[0131] The suspicion level can be considered as information used to express the user's current suspicion level.

[0132] The suspicion level can be divided into three levels: a mild suspicion level, a moderate suspicion level, and a severe suspicion level. Different levels of explanation can be given for different suspicion levels. For example, a brief explanation can be given for the mild suspicion level, auxiliary data can be used to explain the moderate suspicion level, and detailed evidence can be provided for the severe suspicion level. The level can also be divided according to actual conditions, and this embodiment does not make any specific limitations.

[0133] Specifically, when users are skeptical, their degree of suspicion will vary. By comprehensively analyzing several user suspicion assessment factors, quantifying the user's degree of suspicion, and then grading the user's degree of suspicion, the user's suspicion level is determined, providing an important data basis for subsequently determining corresponding explanatory information based on the user's suspicion level.

[0134] S204: Determine and output corresponding explanation information according to the suspicion level and the consultation information.

[0135] Explanation information can be considered as a series of information, such as text information and data reports, that is derived based on the user's suspicion level and the consultation information and can eliminate the doubts generated during the user's consultation process.

[0136] Specifically, providing corresponding explanatory information for different levels of user suspicion can make users feel the pertinence and professionalism of the service, thereby improving the overall user experience. For example, detailed explanations and evidence support are provided to users with severe suspicion levels to reduce or eliminate the user's suspicion, while concise answers are provided to users with mild suspicion to ensure response speed, thereby eliminating the user's suspicion and improving the user's experience.

[0137] Through the method provided in this embodiment, the consulting information provided by the user is preprocessed to determine the preprocessing results that can reflect the key information in the consulting information, improve the accuracy of understanding and response speed of the consulting information, and then analyze the preprocessing results to determine the user suspicion evaluation factors that can reflect the user's suspicion attitude, and then determine the user's suspicion level based on the user suspicion evaluation factors, and determine and output the explanation information corresponding to the suspicion level based on the user's suspicion level. By identifying and processing the suspicions in the user consultation process, it is helpful to establish the user's sense of trust and improve the user's understanding and acceptance of feedback information. In the long run, it can effectively improve the user's loyalty and return rate and improve the user's usage experience.

[0138] In some embodiments, the consulting information is segmented to extract a number of consulting keywords; the consulting information and the consulting keywords are analyzed to determine the user's emotional characteristics; the consulting information is analyzed to extract the user's operational characteristics; and the consulting keywords, emotional characteristics, and operational characteristics are used as preprocessing results.

[0139] Consulting keywords can be considered as key words that can accurately express the consulting subject or content of consulting information.

[0140] Repeated word frequency can be considered as the number of times each consulting keyword appears in the consulting information.

[0141] Emotional features can be considered as relevant information that can reflect the current user's emotional state and the process of emotional change.

[0142] Operational features can be considered as relevant information that can reflect the current user interaction status, such as typing smoothness.

[0143] In a specific implementation, natural language analysis tools such as LTP (Language Technology Platform) and HanLP (Han Language Processing) are used to perform word segmentation on the consulting information, and keyword extraction algorithms such as TF-IDF are used to identify the word segments in the consulting information, and then determine a number of consulting keywords. By extracting keywords, the consulting intentions directly expressed in the user's consulting information and the doubts directly expressed by the user can be determined.

[0144] Among them, the TF-IDF algorithm can be trained using some statistical data to achieve the recognition of consulting information word segmentation. The above statistical data can use historical consulting data.

[0145] Since users may be skeptical during the consultation process, and this skepticism may not be directly reflected through the corresponding consultation keywords, users' skepticism is usually reflected through the negative emotions revealed in the user's consultation information and consultation keywords. For example, if the user expresses dissatisfaction, worry, anger and communication, they can all be regarded as users being skeptical. Therefore, by comprehensively analyzing the consultation information and consultation keywords, we can determine the user's current emotional characteristics, thereby indirectly reflecting the current user's skepticism, and providing an important data basis for subsequently deriving the user's degree of skepticism based on the user's emotional characteristics.

[0146] In addition to the above-mentioned emotional characteristics, the operational characteristics that reflect the user's interaction status can also indirectly reflect the current user's skeptical attitude. For example, after receiving feedback information, the user takes a long time to raise follow-up questions, or in the process of raising follow-up questions, the typing fluency is low, reflecting that the user is repeating the wording, which can also be regarded as the current user's lack of trust. Therefore, by extracting the user's operational characteristics, the current user's skeptical attitude can also be indirectly reflected, providing an important data basis for subsequently deriving the user's degree of suspicion based on the user's operational characteristics.

[0147] Through the method provided in this embodiment, the consulting information is segmented, consulting keywords are extracted, the consulting intention directly expressed in the user's consulting information and the suspicion information directly expressed by the user are determined, and by determining the emotional characteristics and operational characteristics, the current user's skeptical attitude is indirectly reflected, which provides an important data basis for subsequently deriving the user's degree of suspicion based on the user's emotional characteristics and operational characteristics, and optimizes the user experience and improves user satisfaction through comprehensive preprocessing and analysis methods.

[0148] In some embodiments, consulting information and several consulting keywords are analyzed to determine the user's emotional polarity; if the emotional polarity is negative, context information is obtained based on the consulting information; the context information is screened to determine context emotion change information; and the emotional polarity and context emotion change information are used as emotion features.

[0149] The sentiment polarity may be considered as the emotional tendency of the user, including but not limited to negative polarity, neutral polarity and positive polarity.

[0150] Context information can be considered as a series of interactive information between the user and the consulting information management system before and after the consulting information.

[0151] The context emotion change information may be considered as relevant information in the context information that may reflect the emotion change of the user.

[0152] Specifically, during the interaction process, if the user has doubts, the user's emotions will change, and the change pattern of the user's emotions is usually reflected in the change of the user's emotional polarity, such as from positive polarity or neutral polarity to negative polarity. This emotional change is gradual. After the user has doubts, he may gradually become uneasy, anxious, or even angry. These emotional changes will be reflected in the consulting information and corresponding contextual information during the user interaction process.

[0153] In the process of analyzing consulting information and several consulting keywords and determining the user's emotional polarity, a sentiment dictionary can be constructed through a large amount of historical interaction data, and the consulting keywords can be labeled with sentiment polarity through sentiment dictionary technology. Then, the current user's emotional polarity can be determined through the labeled consulting keywords.

[0154] If the user's emotional polarity is negative, it may indicate that the current user is skeptical. At this time, through session management technology, historical conversation information is obtained based on consultation information to determine context information. Then, through sentiment analysis technology, such as VADER (Valence Aware Dictionary and sEntiment Reasoner), the context information is analyzed based on the above sentiment dictionary to obtain contextual emotional change information.

[0155] In some other embodiments, a sentiment analysis model, such as BERT (Bidirectional Encoder Representations from Transformers, a Transformer-based encoder-decoder attention model), can be trained using a large amount of historical conversation data, and contextual information can be analyzed using the sentiment analysis model to obtain contextual emotional change information.

[0156] Through the method provided in this embodiment, the consulting information and several consulting keywords are analyzed to determine the user's emotional polarity. When the emotional polarity is negative, the contextual information is further extracted and analyzed to obtain the user's contextual emotional change information. The emotional polarity and contextual emotional change information are used as emotional features to comprehensively and accurately analyze the user's current emotional state, providing an important data basis for the subsequent analysis of the user's skeptical attitude based on the user's emotional features.

[0157] In some embodiments, a response interval time value is determined based on the consultation information; a response time ratio is determined based on the consultation information and the response interval time value; the response time ratio is compared with a preset response time ratio range to determine whether the response time ratio is greater than the preset response time ratio range; if the response time ratio is greater than the preset response time ratio range, the user operation record corresponding to the consultation information is obtained; the user operation record is analyzed to determine a number of pause nodes, and the number of pause nodes are used as operation features.

[0158] The response interval value can be considered as the time difference between the user receiving the feedback information and the user providing the next round of consultation information.

[0159] The response time ratio may be considered as the ratio between the response interval time value and the corresponding consultation information length, and the consultation information length may be considered as the character length in the consultation information.

[0160] The preset response time ratio range can be considered as the ratio range between the response interval time value and the corresponding consultation information length when the user is in a normal state. The normal state can be considered as a state where the user is not skeptical. The preset response time ratio range can be obtained by analyzing a large amount of historical conversation data.

[0161] User operation records can be considered as relevant operation records during the user's editing of consulting information.

[0162] The pause node can be considered as the time node where a pause occurs during the process of the user editing the consulting information.

[0163] Specifically, when a user is skeptical, due to the user's doubts, the user will hesitate and delay in providing consulting information. The user's hesitation and delay can be directly reflected in the response time ratio. According to the consulting information, the user's response interval time value is obtained through log management technology, and then the length of the consulting information is statistically analyzed through natural language processing tools to obtain the ratio between the response interval time value and the corresponding consulting information length, that is, the response time ratio. The larger the response time ratio, the greater the difference between the time spent by the user to edit the consulting information of the current length and the normal time. When the response time ratio is greater than the preset response time ratio range, it can be considered that the user currently has doubts.

[0164] At this time, the user operation records are obtained through log management technology, and the user operation records are further analyzed to determine the pause nodes in the process of users editing consulting information. This is because the pauses when users edit consulting information may be due to doubts and thinking about feedback information, or repeatedly modifying the wording during the doubt process in the hope that the next round of interaction can obtain reliable information. Therefore, using pause nodes as operation features can more carefully reflect the relationship between user operation features and user skepticism.

[0165] Through the method provided in this embodiment, the response interval time value is determined according to the consultation information, and then the response time ratio is determined. By comparing the response time ratio with the preset response time ratio range, it is concluded whether the current user is skeptical. The user operation records are further analyzed, and the pause nodes in the operation records are used as operation features. This more carefully reflects the relationship between the user operation features and the user's skeptical attitude, providing a highly representative data basis for the subsequent analysis of the user's skeptical attitude based on the operation features.

[0166] In some embodiments, based on several consulting keywords, the contextual emotion change information is analyzed to determine the directed emotion corresponding to the emotional polarity; based on the directed emotion, the contextual emotion change information is analyzed to extract several tone feature words of the user; and the directed emotion and several tone feature words are used as user suspicion assessment factors.

[0167] The pointed emotion can be considered as the specific emotion information pointed to by the contextual emotion change information after the user's emotion changes during the consultation process.

[0168] Tone feature words can be considered as tone words in the context emotion change information that are consistent with the current user's emotional polarity.

[0169] Specifically, it can be seen from the aforementioned embodiments that when the user's emotional polarity is negative, it is highly likely that the user is skeptical, but negative polarity includes a variety of specific emotions, and different specific emotions can reflect differences in the degree of user suspicion. The specific emotion is the directional emotion of the user's emotional polarity. Sentiment analysis technology, such as VADER, can be used to analyze contextual emotional change information based on several consulting keywords to derive a directional emotion that can reflect the user's degree of suspicion. For example, when the directional emotion is anger, it may mean that the user still has not obtained the key information he needs after a long period of interaction, thereby generating a strong sense of distrust. Therefore, using the directional emotion as a user suspicion assessment factor can help improve the accuracy of subsequent analysis of the user's degree of suspicion.

[0170] Furthermore, after determining the directed emotion, since the same directed emotion also has different emotional intensity, and the difference in emotional intensity can also indirectly reflect the user's degree of suspicion, the emotional intensity of the directed emotion can be reflected by the tone feature words used by the user in the contextual emotional change information. Therefore, through the analysis of sentiment dictionary technology, the contextual emotional change information is analyzed, and then a number of tone feature words in the contextual emotional change information that are consistent with the directed emotion are determined, and a number of tone feature words are used as user suspicion assessment factors to further improve the accuracy of subsequent analysis of the user's degree of suspicion.

[0171] Through the method provided in this embodiment, the contextual emotion change information is analyzed to determine the specific emotion pointed to by the current user's emotional polarity, that is, the pointed emotion, and based on the pointed emotion, several tone feature words in the contextual emotion change information that meet the pointed emotion are determined. The pointed emotion and several tone feature words are used as user suspicion evaluation factors to comprehensively reflect the relationship between user emotion and suspicion attitude, thereby further improving the accuracy of subsequent analysis of the user's degree of suspicion.

[0172] In some embodiments, based on a number of pause nodes, the number of pause nodes and the average pause duration are determined; based on a number of pause nodes, the consulting information is analyzed to determine a number of precipitation information and a number of correction information before and after each pause node; the average pause duration, the number of pause nodes, a number of precipitation information and a number of correction information are used as user suspicion assessment factors.

[0173] The number of pause nodes can be considered as the number of times the user pauses when editing the consultation information.

[0174] The average pause time can be considered as the average pause time of users when editing consulting information.

[0175] Sedimentary information can be considered as the information entered by the user after a pause when editing the consultation information.

[0176] The revised information can be considered as the information modified by the user after pausing when editing the consulting information.

[0177] Specifically, when users are skeptical, they will be obviously hesitant in the process of editing consulting information, such as frequent pauses, repeated wording, and modification of information.

[0178] Taking the pause node as the starting point, the number of pause nodes, that is, the number of pauses, is counted. The more pause nodes there are, the more obvious doubts the user has at this time.

[0179] By counting the total pause time, we can obtain the ratio of the total pause time to the number of pause nodes, that is, the average pause time. The longer the average pause time is when the user is editing the consulting information, the more doubts the user has in the process of editing the consulting information, which causes the user to need time to stop and think in the process of editing the consulting information.

[0180] After determining the pause node, the consulting information is analyzed through log management technology to determine the sedimentation information and correction information provided by the user before and after the pause node. The sedimentation information can be regarded as the detailed wording information that the user considers appropriate and input after pausing and thinking, while the correction information can be regarded as part of the consulting information that the user considers ambiguous in the current consulting information during the pause process and then modified by the user. Using the sedimentation information and correction information as user suspicion assessment factors will help to analyze the detailed operations of the user during the pause process and further improve the accuracy of the judgment of the degree of suspicion.

[0181] Through the method provided in this embodiment, several pause nodes are analyzed to determine the number of pause nodes and the average pause duration. The number of pause nodes and the average pause duration are used as user suspicion evaluation factors to accurately reflect the user's suspicion attitude. Further analysis is performed to determine the precipitation information and correction information before and after the pause node, and the precipitation information and correction information are used as user suspicion evaluation factors, which is helpful for subsequent analysis of the detailed operations of the user during the pause process, and further improves the accuracy of the judgment of the degree of suspicion.

[0182] In some embodiments, the user's emotion intensity is determined based on the pointed emotion; the tone word frequency is determined based on a number of tone feature words; the average length of the precipitation information is determined based on a number of precipitation information; the number of corrections is determined based on a number of correction information; the emotion influencing factor is determined based on the emotion intensity and the tone word frequency; the operation influencing factor is determined based on the average pause duration, the number of pause nodes, the average length of the precipitation information and the number of corrections; the suspicion index is determined based on the emotion influencing factor and the operation influencing factor; and the suspicion level is determined based on the suspicion index.

[0183] Emotion intensity can be considered as the intensity of the user's current emotion in the negative polarity emotions.

[0184] The frequency of modal particles can be considered as the frequency of occurrence of modal particles in the contextual emotional change information.

[0185] The average length of the deposited information can be considered as the average character length of all the deposited information entered by the user.

[0186] The number of revisions can be considered as the number of times the user modifies the consulting information during the process of editing the consulting information.

[0187] The emotion impact factor can be considered as the quantitative impact data of user emotions on the degree of user suspicion.

[0188] The operation impact factor can be considered as the quantitative impact data of user operations on the user's degree of suspicion.

[0189] The suspicion index can be considered as quantitative data used to reflect the degree of user suspicion.

[0190] Specifically, the user's degree of suspicion can be reflected in several user suspicion evaluation factors, mainly in two aspects, namely, emotion and operation. The emotion aspect mainly includes emotion intensity and tone word frequency. The emotion intensity reflects the intensity of the current user's directed emotion in the negative polarity emotion. For example, the intensity of anger is stronger than that of anxiety, and the intensity of anxiety is stronger than that of worry. By analyzing a large amount of historical conversation data, the intensity of each emotion under negative polarity can be set, so as to obtain the user's emotion intensity according to the current user's directed emotion. The tone word frequency mainly reflects the intensity of the user's current directed emotion itself. For example, if the user shows worry emotion, and the frequency of tone words is at a high level, it means that the user shows strong worry emotion at this time. By counting the frequency of tone words in the context emotion change information, the tone word frequency can be obtained. By analyzing the emotion intensity and tone word frequency, the emotion intensity corresponding to the user's current directed emotion in the negative polarity emotion can be reflected, and at the same time, the degree difference of the user's directed emotion itself can be reflected. By comprehensively considering the emotion intensity and tone word frequency, the emotion influencing factor used to reflect the relationship between the user's emotion and the degree of suspicion can be obtained.

[0191] The operational aspect mainly includes the average pause duration, the number of pause nodes, the average length of precipitation information and the number of corrections. The reflection of the average pause duration and the number of pause nodes on the user's skeptical attitude has been explained in the aforementioned embodiment. As for the precipitation information and correction information, the average length of the precipitation information and the number of corrections are mainly counted. The longer the average length of the precipitation information, the more detailed the wording of the user in the consulting information, which can reflect the user's distrust and worry that his or her ideas will be misunderstood. The more corrections, the anxiety and worry of the user, which also shows the user's distrust. By comprehensively considering the average pause duration, the number of pause nodes, the average length of precipitation information and the number of corrections, the operational influencing factor reflecting the relationship between user operation and the degree of suspicion can be obtained.

[0192] After obtaining the emotional influencing factor and the operational influencing factor, the two influencing factors are combined to obtain a suspicion index that quantifies the degree of user suspicion. According to the specific numerical value of the suspicion index, the user's suspicion level can be accurately divided.

[0193] Through the method provided in this embodiment, the emotion intensity and the frequency of modal words are combined to obtain an emotion influence factor reflecting the relationship between the user emotion and the degree of suspicion. The average pause duration, the number of pause nodes, the average length of sedimentation information and the number of corrections are combined to obtain an operation influence factor reflecting the relationship between the user operation and the degree of suspicion. Then, a suspicion index quantifying the degree of user suspicion is obtained. According to the suspicion index, the user's suspicion level is accurately divided, and various factors reflecting the user's suspicion level are fully considered to ensure the comprehensiveness and accuracy of the suspicion level.

[0194] In some embodiments, the emotion influencing factor is determined according to the emotion intensity and the tone word frequency, referring to formula (1):

[0195] (1)

[0196] in, Emotional influence factors, is the emotion intercept term, is the regression coefficient of emotion intensity, For emotional intensity, is the modal particle frequency regression coefficient, is the frequency of modal particles;

[0197] According to the average pause time, the number of pause nodes, the average length of the sedimentation information and the number of corrections, the operation impact factor is determined, referring to formula (2):

[0198] (2)

[0199] in, is the operational impact factor, is the intercept term, is the pause duration regression coefficient, is the average pause time, is the regression coefficient of the number of pauses, is the number of paused nodes, is the information length regression coefficient, is the average length of precipitation information, is the corrected regression coefficient, is the number of corrections.

[0200] Specifically, through formula (1) and formula (2), the values ​​of the emotion influence factor and the operation influence factor can be controlled in the interval of [0, 1] respectively. The closer the value is to 1, the greater the influence on the final reflected degree of suspicion is. Conversely, the smaller the influence on the final reflected degree of suspicion is. Formula (1) and formula (2) can accurately and intuitively reflect the influence of user emotions and operations on the reflected degree of suspicion.

[0201] The above-mentioned regression coefficients of emotion intensity, modal word frequency, pause duration, pause number, information length and correction number can all be obtained by fitting a large amount of corresponding historical data.

[0202] The above-mentioned emotional intercept term and operational intercept term are both constant terms that reduce the error of the final result of their respective formulas. The emotional intercept term and operational intercept term can be obtained by analyzing historical data.

[0203] Through the method provided in this embodiment, formulas are designed to accurately quantify the emotion influence factor and the operation influence factor, respectively, and the specific values ​​of the emotion influence factor and the operation influence factor are controlled within a limited range, accurately and intuitively reflecting the influence of user emotions and operations on the reflected degree of suspicion, and providing an important data basis for the subsequent quantification of the suspicion index according to the emotion influence factor and the operation influence factor.

[0204] In some embodiments, the suspicion index is determined based on the emotion influence factor and the operation influence factor, referring to formula (3):

[0205] (3)

[0206] in, For the suspicion index, is the emotional impact weight, Emotional influence factors, is the operation impact weight, is the operational influencing factor.

[0207] The above-mentioned emotional impact weight represents the degree of influence of the emotional impact factor on the suspicion index, and the operational impact weight represents the degree of influence of the operational impact factor on the suspicion index. Both the emotional impact weight and the operational impact weight can be obtained by analyzing a large amount of historical data.

[0208] Specifically, after obtaining the quantitative data corresponding to the emotional influence factor and the operational influence factor, since there are differences in the degree of influence of emotions and operations on the suspicion index, the weighted values ​​of the emotional influence factor and the operational influence factor are calculated by designing formula (3), and the weighted values ​​corresponding to the emotional influence factor and the operational influence factor are combined to obtain the suspicion index, making the suspicion index more accurate.

[0209] Through the method provided in this embodiment, a formula is designed to perform weighted calculation on the emotional influence factor and the operational influence factor, and the weighted values ​​corresponding to the emotional influence factor and the operational influence factor are comprehensively considered to obtain the suspicion index. Compared with the method of summing the emotional influence factor and the operational influence factor, the different effects of the emotional influence factor and the operational influence factor on the suspicion index are more comprehensively considered, making the resulting suspicion index more accurate.

[0210] Figure 3 A structural diagram of a consultation management system based on language analysis provided in one embodiment of the present application, such as Figure 3 As shown, the consultation management system 300 based on language analysis of this embodiment includes: a pre-processing module 301 , a factor determination module 302 , a level analysis module 303 and an output module 304 .

[0211] The preprocessing module 301 obtains the consulting information provided by the user, preprocesses the consulting information, and determines the preprocessing result;

[0212] A factor determination module 302 is used to analyze the preprocessing results and determine a number of user suspicion assessment factors;

[0213] A level analysis module 303, used to determine the user's suspicion level according to the plurality of user suspicion evaluation factors;

[0214] The output module 304 is used to determine and output corresponding explanation information according to the suspicion level and the consultation information.

[0215] Optionally, the preprocessing module 301 is specifically used for:

[0216] Perform word segmentation processing on the consulting information to extract several consulting keywords;

[0217] Analyzing the consultation information and the plurality of consultation keywords to determine the emotional characteristics of the user;

[0218] Analyze the consultation information and extract the operation characteristics of the user;

[0219] The several consulting keywords, the emotional features and the operating features are used as the preprocessing results.

[0220] Optionally, the preprocessing module 301 is specifically used for:

[0221] Analyzing the consulting information and the several consulting keywords to determine the sentiment polarity of the user;

[0222] If the sentiment polarity is negative polarity, obtaining context information according to the consultation information;

[0223] Screening the context information to determine context emotion change information;

[0224] The emotion polarity and the context emotion change information are used as the emotion features.

[0225] Optionally, the preprocessing module 301 is specifically used for:

[0226] Determine a response interval time value according to the consultation information;

[0227] Determining a response time ratio according to the consultation information and the response interval time value;

[0228] Comparing the response time ratio with a preset response time ratio range to determine whether the response time ratio is greater than the preset response time ratio range;

[0229] If the response time ratio is greater than a preset response time ratio range, obtaining a user operation record corresponding to the consultation information;

[0230] The user operation record is analyzed to determine a number of pause nodes, and the number of pause nodes are used as operation features.

[0231] Optionally, the factor determination module 302 is specifically configured to:

[0232] Based on the several consulting keywords, analyzing the context emotion change information, and determining the directed emotion corresponding to the emotion polarity;

[0233] Based on the directed emotion, analyzing context emotion change information, and extracting a number of tone feature words of the user;

[0234] The directed emotion and the plurality of tone feature words are used as the user suspicion assessment factors.

[0235] Optionally, the factor determination module 302 is specifically configured to:

[0236] Determine the number of pause nodes and the average pause duration according to the plurality of pause nodes;

[0237] Based on the plurality of pause nodes, analyzing the consultation information, determining a plurality of precipitation information and a plurality of correction information before and after each pause node;

[0238] The average pause duration, the number of pause nodes, the plurality of precipitation information and the plurality of correction information are used as the user suspicion assessment factors.

[0239] Optionally, the level analysis module 303 is specifically used to:

[0240] Determining the emotion intensity of the user according to the directed emotion;

[0241] Determining the frequency of modal words according to the modal feature words;

[0242] Determining an average length of the precipitation information according to the plurality of precipitation information;

[0243] Determining the number of corrections according to the correction information;

[0244] Determining an emotion influencing factor according to the emotion intensity and the tone word frequency;

[0245] Determining an operation influencing factor according to the average pause duration, the number of pause nodes, the average length of the precipitation information, and the number of corrections;

[0246] Determining a suspicion index according to the emotion influencing factor and the operation influencing factor;

[0247] According to the suspicion index, a suspicion level is determined.

[0248] Optionally, the level analysis module 303 is specifically used to:

[0249] ;

[0250] in, is the emotion influencing factor, is the emotion intercept term, is the regression coefficient of emotion intensity, For the intensity of the emotion, is the modal particle frequency regression coefficient, is the frequency of the modal particle;

[0251] ;

[0252] in, is the operation influencing factor, is the intercept term, is the pause duration regression coefficient, is the average pause time, is the regression coefficient of the number of pauses, is the number of paused nodes, is the information length regression coefficient, is the average length of the precipitation information, is the corrected regression coefficient, is the number of corrections.

[0253] Optionally, the level analysis module 303 is specifically used to:

[0254] ;

[0255] in, is the suspicion index, is the emotional impact weight, is the emotion influencing factor, is the operation impact weight, is the operation influencing factor.

[0256] The system of this embodiment can be used to execute the method of any of the above embodiments. The implementation principles and technical effects are similar and will not be described in detail here.

Claims

1. A consulting management method based on language analysis, characterized in that: include: Obtaining consulting information provided by a user, preprocessing the consulting information, and determining a preprocessing result; Analyzing the preprocessing results to determine a number of user suspicion assessment factors; Determining the user's suspicion level according to the plurality of user suspicion assessment factors; Determine and output corresponding explanation information according to the suspicion level and the consultation information; The analyzing the preprocessing results to determine a number of user suspicion assessment factors include: According to a number of pause nodes, determine the number of pause nodes and the average pause duration; Based on the plurality of pause nodes, analyzing the consultation information, determining a plurality of precipitation information and a plurality of correction information before and after each pause node; Using the average pause duration, the number of pause nodes, the plurality of precipitation information, and the plurality of correction information as the user suspicion assessment factors; The preprocessing of the consultation information and determining the preprocessing result includes: Perform word segmentation processing on the consulting information to extract several consulting keywords; Analyzing the consultation information and the plurality of consultation keywords to determine the emotional characteristics of the user; Analyze the consultation information and extract the operation characteristics of the user; taking the several consulting keywords, the emotional features and the operational features as the preprocessing results; The analyzing the consulting information and the plurality of consulting keywords to determine the emotional characteristics of the user includes: Analyzing the consulting information and the several consulting keywords to determine the sentiment polarity of the user; If the sentiment polarity is negative polarity, obtaining context information according to the consultation information; Screening the context information to determine context emotion change information; Using the emotion polarity and the context emotion change information as the emotion feature; The analyzing the preprocessing results to determine a number of user suspicion assessment factors include: Based on the several consulting keywords, analyzing the context emotion change information, and determining the directed emotion corresponding to the emotion polarity; Based on the directed emotion, analyzing context emotion change information, and extracting a number of tone feature words of the user; Using the directed emotion and the plurality of tone feature words as the user suspicion assessment factors; The determining the user's suspicion level according to the plurality of user suspicion assessment factors includes: Determining the emotion intensity of the user according to the directed emotion; Determining the frequency of modal words according to the modal feature words; Determining an average length of the precipitation information according to the plurality of precipitation information; Determining the number of corrections according to the correction information; Determining an emotion influencing factor according to the emotion intensity and the tone word frequency; Determining an operation influencing factor according to the average pause duration, the number of pause nodes, the average length of the precipitation information, and the number of corrections; Determining a suspicion index according to the emotion influencing factor and the operation influencing factor; According to the suspicion index, a suspicion level is determined.

2. The method according to claim 1, characterized in that The analyzing the consultation information and extracting the operation characteristics of the user includes: Determine a response interval time value according to the consultation information; Determining a response time ratio according to the consultation information and the response interval time value; Comparing the response time ratio with a preset response time ratio range to determine whether the response time ratio is greater than the preset response time ratio range; If the response time ratio is greater than a preset response time ratio range, obtaining a user operation record corresponding to the consultation information; The user operation record is analyzed to determine the plurality of pause nodes, and the plurality of pause nodes are used as operation features.

3. The method according to claim 1, characterized in that Determining the emotion influencing factor according to the emotion intensity and the tone word frequency includes: ; in, is the emotion influencing factor, is the emotion intercept term, is the regression coefficient of emotion intensity, For the intensity of the emotion, is the modal particle frequency regression coefficient, is the frequency of the modal particle; The determining of the operation influencing factor according to the average pause duration, the number of pause nodes, the average length of the precipitation information and the number of corrections includes: ; in, is the operation influencing factor, is the intercept term, is the pause duration regression coefficient, is the average pause time, is the regression coefficient of the number of pauses, is the number of paused nodes, is the information length regression coefficient, is the average length of the precipitation information, is the corrected regression coefficient, is the number of corrections.

4. The method according to claim 3, characterized in that Determining the suspicion index according to the emotion influencing factor and the operation influencing factor includes: ; in, is the suspicion index, is the emotional impact weight, is the emotion influencing factor, is the operation impact weight, is the operation influencing factor.

5. A consultation management system based on language analysis, characterized in that: The method as claimed in any one of claims 1 to 4 comprises: A preprocessing module, which obtains the consulting information provided by the user, preprocesses the consulting information, and determines the preprocessing result; A factor determination module, used to analyze the preprocessing results and determine a number of user suspicion assessment factors; A level analysis module, used to determine the user's suspicion level based on the plurality of user suspicion assessment factors; An output module, used to determine and output corresponding explanation information according to the suspicion level and the consultation information; The factor determination module is specifically used for: According to a number of pause nodes, determine the number of pause nodes and the average pause duration; Based on the plurality of pause nodes, analyzing the consultation information, determining a plurality of precipitation information and a plurality of correction information before and after each pause node; The average pause duration, the number of pause nodes, the plurality of precipitation information and the plurality of correction information are used as the user suspicion assessment factors.

Images