A method, device, and medium for predicting the timeliness requirements of complaint event handling

Abstract

The present invention relates to a method, device and medium for predicting the timeliness requirement of complaint event processing. The method includes the following steps: collecting complaint event text data, performing word segmentation, and extracting emotional feature words and situational features therein; constructing an emotional feature word library and setting three-dimensional scores for each emotional feature word in the emotional feature word library; constructing corresponding three-dimensional emotional coordinates and emotional interaction matrices for each emotional feature word in the complaint event text data based on the emotional feature word library; calculating the overall emotional state trend of the complaint event based on the three-dimensional emotional coordinates and the emotional interaction matrix; calculating the narrative tension of the complaint event according to the overall emotional state trend of the complaint event and the situational features; constructing a user emotion prediction model to predict the future trend of the emotion of the user corresponding to the complaint event; and calculating the processing timeliness requirement of the current complaint event based on the narrative tension of the complaint event and the future trend of the user's emotion.

Description

Technical Field

[0001] The present invention relates to a method, device and medium for predicting the timeliness requirements of complaint event handling, and belongs to the technical field of complaint management. Background Art

[0002] With the rapid development of Internet technology and customer service business, users' expectations for service quality are constantly increasing. How to quickly and accurately judge the urgency and handling priority of complaint events has become one of the important challenges for enterprises to improve customer satisfaction and service level.

[0003] Traditional methods for predicting the timeliness of complaint event handling generally rely on manual experience or simple rule judgments, lacking objectivity and accuracy. At the same time, although some existing technologies attempt to use simple sentiment analysis models, they usually ignore the multi-dimensionality of complaint text sentiment, the dynamic characteristics of emotional changes, and the interactive effects between emotions, resulting in the prediction results being difficult to comprehensively reflect the true emotional state of users and its trend changes. Summary of the Invention

[0004] In order to solve the problems existing in the above-mentioned prior art, the present invention proposes a method, device and medium for predicting the timeliness requirements of complaint event handling.

[0005] The technical solution of the present invention is as follows:

[0006] On the one hand, the present invention provides a method for predicting the timeliness requirements of complaint event handling, including the following steps:

[0007] Collect complaint event text data, and perform word segmentation on the complaint event text data through a pre-trained large model and extract the emotional feature words and situational features therein;

[0008] Construct an emotional feature word library and set three-dimensional scores for each emotional feature word in the emotional feature word library, including emotional intensity score, emotional valence score, and emotional control sense score;

[0009] Based on the emotional feature word library, construct corresponding three-dimensional emotional coordinates and emotional interaction matrices for each emotional feature word in the complaint event text data;

[0010] Based on the corresponding three-dimensional emotional coordinates and emotional interaction matrices constructed for each emotional feature word, calculate the overall emotional state trend of the complaint event;

[0011] Calculate the narrative tension of the complaint event according to the overall emotional state trend of the complaint event and the situational features;

[0012] Construct a user emotion prediction model based on an improved bidirectional long short-term memory network, and predict the future trend of the emotion of the user corresponding to the complaint event through the overall emotion state trend of the complaint event. The improved bidirectional long short-term memory network includes a forward and backward long short-term memory network layer and an attention fusion layer;

[0013] Calculate the processing time limit requirement of the current complaint event based on the narrative tension of the complaint event and the future trend of the user's emotion.

[0014] Preferably, construct a corresponding three-dimensional emotion coordinate for each emotion feature word in the complaint event text data based on the emotion feature word library as shown in the following formula:

[0015] ;

[0016] Where: Represents the three-dimensional emotion coordinate corresponding to the th emotion feature word in the complaint event text data; Represents the emotion intensity score of the th emotion feature word; Represents the emotion valence score of the th emotion feature word; Represents the emotion control sense score of the th emotion feature word;

[0017] Construct a corresponding emotion interaction matrix for each emotion feature word in the complaint event text data based on the emotion feature word library as shown in the following formula:

[0018] ;

[0019] Where: Represents the emotion interaction matrix of the th emotion feature word; Represents the natural attenuation coefficient of emotion intensity; Represents the natural attenuation coefficient of emotion valence; Represents the natural attenuation coefficient of emotion control sense; Represents the th emotion feature word dimensional score and dimensional score interaction parameter, , , .

[0020] Preferably, based on the corresponding three-dimensional emotion coordinate and emotion interaction matrix constructed for each emotion feature word, the specific steps for calculating the emotion intensity curve, emotion valence curve and emotion control sense curve of the complaint event are:

[0021] Construct an emotional state vector based on the three-dimensional emotional coordinates of each emotional feature word according to the appearance order of the emotional feature words in the complaint event text data , where represents the three-dimensional emotional coordinates of the th emotional feature word that appears in the complaint event text data, represents the three-dimensional emotional coordinates of the last emotional feature word that appears in the complaint event text data;

[0022] At the same time, according to the appearance order of the emotional feature words in the complaint event text data, construct an emotional interaction matrix vector based on the emotional interaction matrix of each emotional feature word , where represents the emotional interaction matrix of the th emotional feature word that appears in the complaint event text data, represents the emotional interaction matrix of the last emotional feature word that appears in the complaint event text data;

[0023] Construct an emotional dynamics equation based on the emotional state vector and the emotional interaction matrix vector, as shown in the following formula:

[0024] ;

[0025] ;

[0026] where: represents the trend of the emotional state vector of the th emotional feature word that appears within the corresponding time interval ; represents the starting moment of the th emotional feature word, that is, the position of this emotional feature word in the complaint event text data; represents the ending moment of the th emotional feature word, that is, the position of the th emotional feature word in the complaint event text data; represents the external emotional influence of the th emotional feature word that appears; represents the weight of the th emotional feature word that appears; represents the natural constant;

[0027] For the time interval corresponding to each emotional feature word that appears, use the end state of the previous time interval as the initial state of the next interval, and use the RK4 method to independently solve the above formula to obtain the trend of the emotional state vector for each time interval, and then splice the trends of the emotional state vectors for each time interval to obtain the overall emotional state trend.

[0028] Preferably, the calculation formula for the narrative tension of a complaint event is specifically shown as follows:

[0029] ;

[0030] ;

[0031] ;

[0032] where: represents the narrative tension of the complaint event; represents the weighted probability distribution of the emotional state of the th emotional feature word that appears in the complaint event text data within the corresponding time interval ; represents the emotional entropy of the th emotional feature word that appears in the complaint event text data within the corresponding time interval ; represents the smoothing coefficient; represents the emotional self-sensitivity coefficient; represents the sensitivity of the situation characteristics of the complaint event; represents the trend of the emotional state vector of the th emotional feature word that appears within the corresponding time interval ; , , respectively represent the trends of the emotional intensity score, emotional valence score, and emotional control sense score of the th emotional feature word that appears within the corresponding time interval .

[0033] Preferably, the forward and backward long short-term memory network layer is specifically shown as follows:

[0034] ;

[0035] ;

[0036] ;

[0037] where: represents the output result of the forward and backward long short-term memory network layer at the th moment; represents the hidden state of the forward long short-term memory network at the th moment; represents the hidden state of the backward long short-term memory network at the th moment; represents the forward and backward long short-term memory network layer at the Input at a moment; Indicates the hidden state of the forward long short-term memory network at a moment; Indicates the hidden state of the backward long short-term memory network at a moment; Indicates the long short-term memory network; Indicates concatenation of and ;

[0038] Each unit of the forward and backward long short-term memory networks is provided with a dynamic emotion enhancement gate, as specifically shown in the following formula:

[0039] ;

[0040] ;

[0041] ;

[0042] ;

[0043] ;

[0044] ;

[0045] ;

[0046] Where: , , respectively represent the output of the forget gate, input gate, and output gate of the forward and backward long short-term memory network units at a moment; , , respectively represent the weight matrices of the forget gate, input gate, and output gate; represents the output result of the forward and backward long short-term memory network layer at a moment; , , respectively represent the bias vectors of the forget gate, input gate, and output gate; represents the output of the dynamic emotion enhancement gate of the forward and backward long short-term memory network units at a moment; represents the weight matrix of the dynamic emotion enhancement gate; represents the multi-scale attention fusion feature output by the attention fusion layer at a moment; represents the bias vector of the dynamic emotion enhancement gate; represents the forward and backward long short-term memory network units at Enhanced emotional candidate state at a moment; Indicates the cell state of the forward and backward long short-term memory network units at a moment; Indicates the cell state of the forward and backward long short-term memory network units at a moment; Weight matrix representing the cell states of the forward and backward long short-term memory network units; Bias vector representing the cell states of the forward and backward long short-term memory network units; Indicates the hidden state of the forward and backward long short-term memory networks at a moment; Indicates the Hadamard product operation; Indicates the sigmoid function.

[0047] Preferably, the attention fusion layer includes a multi-scale convolution extraction unit, a scale interaction unit, a scale selection unit, and a multi-scale feature fusion unit;

[0048] The multi-scale convolution extraction unit performs feature extraction of different scales on the outputs of the forward and backward long short-term memory network layers through convolutional layers with different sizes of convolutional kernels, as shown in the following formula:

[0049] ;

[0050] Where: Indicates the output feature of the forward and backward long short-term memory network layer at a moment with a scale of ; Indicates the convolution operation with a scale of ; Indicates the output sequence of the forward and backward long short-term memory network layer from a moment to a moment;

[0051] Define the interaction weight matrix and set the size of the interaction weight matrix according to the number of scales. The interaction weight matrix is specifically shown in the following formula:

[0052] ;

[0053] ;

[0054] ;

[0055] Where: Indicates the interaction weight matrix at a moment; Indicates function; Indicates Query matrix at a moment; represent; Key matrix at a moment; Learnable parameter matrix representing the query matrix; Learnable parameter matrix representing the key matrix; Represent a preset feature dimension; Represent a transpose operation;

[0056] The scale interaction unit updates the output features of the forward and backward long short-term memory network layers at each scale based on the interaction weight matrix, as shown in the following formula:

[0057] ;

[0058] Where: Represent the updated Output features of the forward and backward long short-term memory network layers at the scale of at the moment; represent; Output features of the forward and backward long short-term memory network layers at the scale of at the moment;

[0059] The scale selection unit is used to calculate the contribution degrees of different scales, as shown in the following formula:

[0060] ;

[0061] Where: Represent the contribution degree of the scale selection unit at the scale of at the moment; ; Represent the learnable parameter matrix of the scale selection unit; Represent the bias vector of the scale selection unit;

[0062] Calculate the final output features of the forward and backward long short-term memory network layers at different scales based on the contribution degrees of different scales, as shown in the following formula:

[0063] ;

[0064] The multi-scale feature fusion unit is used to fuse the final output features of the forward and backward long short-term memory network layers at different scales calculated by the contribution degrees of different scales, as described in the following formula:

[0065] ;

[0066] ;

[0067] Where: Represent the scale The position adaptive weight at moment; represents the attention query vector; is a learnable parameter matrix representing the position adaptive weight; is the bias vector representing the position adaptive weight.

[0068] Preferably, the processing time requirement of the current complaint event is calculated based on the narrative tension of the complaint event and the future trend of the user's emotion, as shown in the following formula:

[0069] ;

[0070] Where: represents the processing time requirement of the current complaint event; represents the sensitivity adjustment coefficient of the time requirement; represents the future trend of the user's emotion corresponding to the current complaint event predicted by the user emotion prediction model at represents the critical threshold of the situation feature.

[0071] On the other hand, the present invention also provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, the method described in the present invention is implemented.

[0072] On yet another aspect, the present invention also provides a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, the method described in the present invention is implemented.

[0073] The present invention has the following beneficial effects:

[0074] 1. The present invention constructs an emotional feature word library and systematically scores three dimensions of emotional intensity, emotional valence, and emotional control sense, effectively reflecting the multi-dimensional features of the emotional state and improving the comprehensiveness and meticulousness of emotional analysis.

[0075] 2. The present invention accurately captures the dynamic change law of emotions and the interactive influence between emotional dimensions in the complaint text by constructing a three-dimensional emotional coordinate and an emotion interaction matrix, thereby realizing an accurate prediction of the trend of the emotional state.

[0076] 3. The present invention proposes a model based on the fusion of an improved bidirectional long short-term memory network and an attention mechanism, enhancing the model's ability to capture the trend of emotional changes, realizing an accurate prediction of the future trend of the user's emotion, and thus more effectively inferring the actual processing requirements of the complaint event. BRIEF DESCRIPTION OF THE DRAWINGS

[0077] Figure 1This is the flowchart of the method of the present invention. Detailed implementation manners

[0078] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0079] It should be understood that the step numbers used in the text are only for convenient description and do not limit the execution order of the steps.

[0080] It should be understood that the terms used in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the specification of the present invention and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include the plural forms.

[0081] The terms "comprising" and "including" indicate the presence of the described features, wholes, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or their combinations.

[0082] The term "and / or" refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0083] Embodiment 1:

[0084] Refer to Figure 1 , a prediction method for the timeliness requirement of complaint event processing, including the following steps:

[0085] Collect the text data of complaint events, tokenize the text data of complaint events through a pre-trained large model, and extract the emotional feature words and situational features therein;

[0086] Construct an emotional feature word library and set three-dimensional scores for each emotional feature word in the emotional feature word library, including emotional intensity score (from calm to extremely intense), emotional valence score (from very negative to very positive), and emotional control sense score (from completely helpless to highly autonomous);

[0087] Based on the emotional feature word library, construct corresponding three-dimensional emotional coordinates and emotional interaction matrices for each emotional feature word in the text data of complaint events;

[0088] Based on the corresponding three-dimensional emotional coordinates and emotional interaction matrices constructed for each emotional feature word, calculate the overall emotional state trend of the complaint event;

[0089] Calculate the narrative tension of the complaint event according to the overall emotional state trend and situational characteristics of the complaint event;

[0090] Construct a user emotion prediction model based on an improved bidirectional long short-term memory network, and predict the future emotional trend of the user corresponding to the complaint event through the overall emotional state trend of the complaint event. The improved bidirectional long short-term memory network includes forward and backward long short-term memory network layers and an attention fusion layer;

[0091] Calculate the processing time limit requirement of the current complaint event based on the narrative tension of the complaint event and the future emotional trend of the user.

[0092] As a preferred implementation manner of this embodiment, construct a corresponding three-dimensional emotion coordinate for each emotion feature word in the complaint event text data based on the emotion feature word library as shown in the following formula:

[0093] ;

[0094] Where: Represents the three-dimensional emotion coordinate corresponding to the th emotion feature word in the complaint event text data; Represents the emotion intensity score of the th emotion feature word; Represents the emotion valence score of the th emotion feature word; Represents the emotion control sense score of the th emotion feature word;

[0095] Construct a corresponding emotion interaction matrix for each emotion feature word in the complaint event text data based on the emotion feature word library as shown in the following formula:

[0096] ;

[0097] Where: Represents the emotion interaction matrix of the th emotion feature word; Represents the natural attenuation coefficient of emotion intensity; Represents the natural attenuation coefficient of emotion valence; Represents the natural attenuation coefficient of emotion control sense; Represents the th emotion feature word dimensional score and dimensional score interaction parameter between, , , .

[0098] As a preferred implementation of this embodiment, the specific steps for constructing a corresponding three-dimensional emotion coordinate and an emotion interaction matrix based on each emotion feature word, and calculating the emotion intensity curve, emotion valence curve, and emotion control sense curve of the complaint event are as follows:

[0099] Based on the appearance order of the emotion feature words in the complaint event text data, construct an emotion state vector based on the three-dimensional emotion coordinates of each emotion feature word , where represents the three-dimensional emotion coordinate of the th emotion feature word that appears in the complaint event text data, represents the three-dimensional emotion coordinate of the last emotion feature word that appears in the complaint event text data;

[0100] At the same time, based on the appearance order of the emotion feature words in the complaint event text data, construct an emotion interaction matrix vector based on the emotion interaction matrix of each emotion feature word , where represents the emotion interaction matrix of the th emotion feature word that appears in the complaint event text data, represents the emotion interaction matrix of the last emotion feature word that appears in the complaint event text data;

[0101] Construct an emotion dynamic equation based on the emotion state vector and the emotion interaction matrix vector, as shown in the following formula:

[0102] ;

[0103] ;

[0104] Where: represents the trend of the emotion state vector of the th emotion feature word within the corresponding time interval ; represents the starting moment of the th emotion feature word, that is, the position of this emotion feature word in the complaint event text data; represents the ending moment of the th emotion feature word, that is, the position of the th emotion feature word in the complaint event text data; represents the external emotion influence of the th emotion feature word; represents the weight of the th emotion feature word; represents the natural constant;

[0105] For each time interval corresponding to the emotional feature words that appear, taking the state at the end of the previous time interval as the initial state of the next interval, the RK4 method is used to independently solve the above formula to obtain the trend of the emotional state vector for each time interval, and then the trends of the emotional state vectors for each time interval are concatenated to obtain the overall emotional state trend.

[0106] As a preferred implementation manner of this embodiment, the calculation formula for the narrative tension of the complaint event is specifically shown as follows:

[0107] ;

[0108] ;

[0109] ;

[0110] Where: represents the narrative tension of the complaint event; represents the weighted probability distribution of the emotional state of the th emotional feature word that appears in the complaint event text data within the corresponding time interval ; represents the emotional entropy of the th emotional feature word that appears in the complaint event text data within the corresponding time interval ; represents the smoothing coefficient; represents the emotional self-sensitivity coefficient; represents the sensitivity of the situation characteristics of the complaint event; represents the trend of the emotional state vector of the th emotional feature word that appears in the corresponding time interval ; , , respectively represent the trends of the emotional intensity score, emotional valence score, and emotional control sense score of the th emotional feature word that appears in the corresponding time interval ;

[0111] The sensitivity of the situation characteristics of the complaint event is determined based on a pre-designed scoring system for the complaint field. Taking the catering field as an example, it is specifically shown in the following table:

[0112]

[0113] According to the situation characteristics and their corresponding sensitivities that appear in the complaint event text data, a weighted sum is performed to obtain the sensitivity of the situation characteristics of the complaint event.

[0114] As a preferred implementation manner of this embodiment, the forward and backward long short-term memory network layer is specifically shown as the following formula:

[0115] ;

[0116] ;

[0117] ;

[0118] Where: represents the output result of the forward and backward long short-term memory network layer at moment, that is, the future trend of the emotion of the user corresponding to the future moment of the current complaint event; represents the hidden state of the forward long short-term memory network at moment; represents the hidden state of the backward long short-term memory network at moment; represents the input of the forward and backward long short-term memory network layer at moment; represents the hidden state of the forward long short-term memory network at moment; represents the hidden state of the backward long short-term memory network at moment; represents the long short-term memory network; represents the and are concatenated;

[0119] Each unit of the forward and backward long short-term memory network is provided with a dynamic emotion enhancement gate, specifically shown as the following formula:

[0120] ;

[0121] ;

[0122] ;

[0123] ;

[0124] ;

[0125] ;

[0126] ;

[0127] Where: , , respectively represent the outputs of the forget gate, input gate, and output gate of the forward and backward long short-term memory network units at moment; , , respectively represent the weight matrices of the forget gate, input gate, and output gate; represents the output result of the forward and backward long short-term memory network layer at moment; , , respectively represent the bias vectors of the forget gate, input gate, and output gate; represents the output of the dynamic emotion enhancement gate of the forward and backward long short-term memory network units at moment; represents the weight matrix of the dynamic emotion enhancement gate; represents the multi-scale attention fusion feature output by the attention fusion layer at moment; represents the bias vector of the dynamic emotion enhancement gate; represents the enhanced emotion candidate state of the forward and backward long short-term memory network units at moment; represents the cell state of the forward and backward long short-term memory network units at moment; represents the cell state of the forward and backward long short-term memory network units at moment; represents the weight matrix of the cell state of the forward and backward long short-term memory network units; represents the bias vector of the cell state of the forward and backward long short-term memory network units; represents the hidden state of the forward and backward long short-term memory network at moment; represents the Hadamard product operation; represents the sigmoid function.

[0128] As a preferred implementation manner of this embodiment, the attention fusion layer includes a multi-scale convolution extraction unit, a scale interaction unit, a scale selection unit, and a multi-scale feature fusion unit;

[0129] In this embodiment, the multi-scale convolution extraction unit respectively performs feature extraction of different scales on the output of the forward and backward long short-term memory network layers through convolution layers with three different sizes of convolution kernels, as shown in the following formula:

[0130] , ;

[0131] Among them: represents The output features of the forward and backward long short-term memory network layer with a time scale of ; denotes the convolution operation with a scale of ; denotes the output sequence of the forward and backward long short-term memory network layer from time to time

[0132] Define the interaction weight matrix and set the size of the interaction weight matrix according to the number of scales. In this embodiment, it is . The specific form of the interaction weight matrix is as follows:

[0133] ;

[0134] ;

[0135] ;

[0136] where: denotes the interaction weight matrix at time ; denotes function; denotes the query matrix at time denotes the key matrix at time denotes the learnable parameter matrix of the query matrix; denotes the learnable parameter matrix of the key matrix; denotes the preset feature dimension; denotes the transpose operation;

[0137] The scale interaction unit updates the output features of the forward and backward long short-term memory network layer at each scale based on the interaction weight matrix. The specific form is as follows:

[0138] ;

[0139] where: denotes the updated output features of the forward and backward long short-term memory network layer with a time scale of at time ; denotes the output features of the forward and backward long short-term memory network layer with a time scale of at time

[0140] The scale selection unit is used to calculate the contribution degrees of different scales. The specific form is as follows:

[0141] ;

[0142] in: Indicates that the scale selection unit is Time output scale Contribution of represents the learnable parameter matrix of the scale selection unit; represents the bias vector of the scale selection unit;

[0143] The final output features of the forward and reverse long short-term memory network layers at different scales are calculated based on the contribution of different scales, as shown in the following formula:

[0144] ;

[0145] The multi-scale feature fusion unit is used to calculate the final output features of the forward and reverse long short-term memory network layers at different scales by the contribution of different scales, as described in the following formula:

[0146] ;

[0147] ;

[0148] in: Representation scale exist The position adaptive weight at the moment; Represents the attention query vector, which can be regarded as a "template" or "standard feature" of emotional features. The features extracted at each scale will be compared with the "standard". The more similar the features are, the more they will be noticed by the model, thus obtaining a higher weight. The features with large differences will be assigned a lower attention weight. a learnable parameter matrix representing the position adaptation weights; Bias vector representing the position adaptation weights.

[0149] As a preferred implementation of this embodiment, the processing time requirement of the current complaint event is calculated based on the narrative tension of the complaint event and the future trend of the user's emotions, as shown in the following formula:

[0150] ;

[0151] in: Indicates the time limit required for handling the current complaint event; Indicates the timeliness demand sensitivity adjustment coefficient (determined based on historical complaint processing timeliness data); Represents the future prediction of the current complaint event by the user sentiment prediction model The future trend of the user's emotions at the moment, ; It represents the situational feature critical threshold, which is set based on the average situational sensitivity.

[0152] Embodiment 2:

[0153] An intelligent judgment system for complaint events includes a presentation layer, an application layer, a support layer, and a data layer;

[0154] The presentation layer includes a PC web end and a front-end architecture. The PC web end is used for the user operation interface, facilitating the public to submit complaints and view the processing progress. The front-end architecture uses JS / CSS, vue.js, and element-plus to ensure a beautiful interface, friendly interaction, and smooth operation, enhancing the user experience.

[0155] The application layer is mainly composed of python and server services, redis services, and message queue services. Python provides the basis for the development language. The server service supports server-side customized development to ensure the system service process. The redis service performs structured caching, and the message queue service constructs a queue producer and consumer mode to improve the system efficiency.

[0156] The support layer includes the Doubao large model, which has functions such as intelligent agent question answering, proprietary domain knowledge base, proprietary domain workflow, large model fine-tuning, and customized development and deployment. At the same time, it includes a method for predicting the processing timeliness requirements of complaint events described in Embodiment 1, which is used to realize the sentiment analysis, narrative tension evaluation, and processing timeliness prediction of complaint events. These functions help to accurately judge complaints from the public intelligently and provide reasonable handling suggestions and legal bases.

[0157] The data layer is divided into a public cloud and a private cloud. The public cloud stores the desensitized large model question-and-answer data and desensitized knowledge base data, and the private cloud stores user data and sensitive data. Through this data storage method, both data security and effective utilization of data can be ensured.

[0158] Embodiment 3:

[0159] This embodiment proposes an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, it implements the method described in any embodiment of the present invention.

[0160] Embodiment 4:

[0161] This embodiment proposes a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, it implements the method described in any embodiment of the present invention.

[0162] In the embodiments of the present application, "at least one" means one or more, and "a plurality" means two or more. "And / or" describes the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent the cases of A existing alone, A and B existing simultaneously, and B existing alone. Wherein A and B can be singular or plural. The character " / " generally represents an "or" relationship between the associated objects before and after. "At least one of the following" and its similar expressions refer to any combination of these items, including any combination of single items or plural items. For example, at least one of a, b, and c can represent: a, b, c, a and b, a and c, b and c, or a and b and c, where a, b, and c can be single or multiple.

[0163] Those of ordinary skill in the art can realize that the various units and algorithm steps described in the embodiments disclosed herein can be implemented by a combination of electronic hardware, computer software, and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods for each specific application to implement the described functions, but such implementation should not be considered to exceed the scope of the present application.

[0164] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0165] In several embodiments provided by the present application, if any function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (hereinafter referred to as ROMs), random access memories (hereinafter referred to as RAMs), magnetic disks, or optical discs that can store program codes.

[0166] The above are only the embodiments of the present invention and do not limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made by using the content of the specification and drawings of the present invention, or directly or indirectly applied in other related technical fields, shall be equally included in the patent protection scope of the present invention.

Claims

1. A method for predicting the timeliness demand for complaint event processing, characterized in that: The following steps are involved: Collect complaint event text data, segment the complaint event text data through a pre-trained large model, and extract sentiment feature words and situational features; Construct an emotional feature word library and set a three-dimensional score for each emotional feature word in the emotional feature word library, including an emotional intensity score, an emotional valence score, and an emotional control score; Based on the emotional feature word library, the corresponding three-dimensional emotional coordinates and emotional interaction matrix are constructed for each emotional feature word in the complaint event text data; The three-dimensional emotion coordinates are specifically shown in the following formula: ； in: Indicates the first The three-dimensional emotional coordinates corresponding to the emotional feature words; Indicates The emotional intensity score of each emotional feature word; Indicates The emotional valence score of the emotional feature words; Indicates The emotional control score of the emotional feature words; The emotion interaction matrix is ​​specifically shown in the following formula: ； in: Indicates The emotional interaction matrix of emotional feature words; Indicates the natural attenuation coefficient of emotion intensity; represents the natural attenuation coefficient of emotional valence; It represents the natural attenuation coefficient of emotional control; Indicates sentiment feature words Dimension ratings and The interaction parameters between dimension ratings, , , , Indicates the emotional intensity dimension, Represents the emotional valence dimension, It represents the dimension of emotional control; Based on each emotional feature word, the corresponding three-dimensional emotional coordinates and emotional interaction matrix are constructed to calculate the overall emotional state trend of the complaint event; Calculate the narrative tension of the complaint event based on the overall emotional state trend and situational characteristics of the complaint event; A user emotion prediction model is constructed based on an improved bidirectional long short-term memory network, and the future trend of the emotion of the user corresponding to the complaint event is predicted through the overall emotional state trend of the complaint event. The improved bidirectional long short-term memory network includes a positive and negative long short-term memory network layer and an attention fusion layer; The time efficiency requirement for handling the current complaint event is calculated based on the narrative tension of the complaint event and the future trend of the user's emotions.

2. A method for predicting the timeliness demand for handling complaint events according to claim 1, characterized in that: Based on each emotional feature word, the corresponding three-dimensional emotional coordinates and emotional interaction matrix are constructed, and the specific steps for calculating the emotional intensity curve, emotional valence curve, and emotional control curve of the complaint event are as follows: Construct the emotional state vector based on the order of occurrence of emotional feature words in the complaint event text data and the three-dimensional emotional coordinates of each emotional feature word ,in, Indicates the first The three-dimensional emotional coordinates of the emotional feature words that appear, The three-dimensional sentiment coordinates representing the last sentiment feature word that appears in the complaint event text data; At the same time, the emotional interaction matrix vector is constructed based on the appearance order of emotional feature words in the complaint event text data and the emotional interaction matrix of each emotional feature word. ,in, Indicates the first The emotional interaction matrix of the emotional feature words that appear, The sentiment interaction matrix representing the last sentiment feature word that appears in the complaint event text data; The emotional dynamics equation is constructed based on the emotional state vector and the emotional interaction matrix vector, as shown in the following formula: ； ； in: Indicates The emotional feature words that appear in the corresponding time interval The emotional state vector trend within; Indicates The starting time of the occurrence of the sentiment feature word, that is, the position of the sentiment feature word in the complaint event text data; Indicates The ending moment of the first emotional feature word, that is, The position of the emotional feature words in the complaint event text data; Indicates The external emotional impact of the emotional feature words that appear; Indicates The weight of the sentiment feature words that appear; represents a natural constant; For each time interval corresponding to the occurrence of the emotional feature word, the final state of the previous time interval is used as the initial state of the next interval. The above formula is independently solved using the RK4 method to obtain the emotional state vector trend of each time interval. The emotional state vector trends of each time interval are then spliced ​​together to obtain the overall emotional state trend.

3. A method for predicting the timeliness demand for handling complaint events according to claim 2, characterized in that: The calculation formula of the narrative tension of the complaint event is as follows: ； ； ； in: Indicates the narrative tension of the complaint incident; Indicates the first The emotional feature words that appear in the corresponding time interval Weighted probability distribution of emotional states within; Indicates the first The emotional feature words that appear in the corresponding time interval The emotional entropy within; represents the smoothing coefficient; represents the emotional self-sensitivity coefficient; Indicates the sensitivity of the situational features of the complaint event; Indicates The emotional feature words that appear in the corresponding time interval The emotional state vector trend within; , , Respectively represent The emotional feature words that appear in the corresponding time interval The trends of emotion intensity scores, emotion valence scores, and emotion control scores.

4. A method for predicting the timeliness demand for handling complaint events according to claim 1, characterized in that: The forward and reverse long short-term memory network layer is specifically shown in the following formula: ； ； ； in: Indicates the forward and reverse long short-term memory network layers in Output results at the moment; Indicates that the forward long short-term memory network is The hidden state of the moment; Reverse long short-term memory network The hidden state of the moment; Indicates the forward and reverse long short-term memory network layers in Input of time; Indicates that the positive long short-term memory network is The hidden state of the moment; Reverse long short-term memory network The hidden state of the moment; represents the long short-term memory network; Express and To splice; Each unit of the forward and reverse long short-term memory networks is provided with a dynamic emotion enhancement gate, as shown in the following formula: ； ； ； ； ； ； ； in: , , Represents the forget gate, input gate, and output gate of the forward and reverse long short-term memory network units respectively. Output at the moment; , , Represent the weight matrices of the forget gate, input gate, and output gate respectively; Indicates the forward and reverse long short-term memory network layers in Output results at the moment; , , Represent the bias vectors of the forget gate, input gate, and output gate respectively; Dynamic emotion enhancement gates representing forward and reverse LSTM network units Output at the moment; represents the weight matrix of the dynamic emotion enhancement gate; Represents the attention fusion layer Multi-scale attention fusion features output at every moment; represents the bias vector of the dynamic emotion enhancement gate; Represents the forward and reverse long short-term memory network units in moment-to-moment enhanced emotional candidate states; Represents the forward and reverse long short-term memory network units in The cell state at a given moment; Represents the forward and reverse long short-term memory network units in The cell state at a given moment; The weight matrix representing the cell states of the forward and reverse LSTM networks; Bias vectors representing the cell states of forward and reverse LSTM networks; Represents the forward and reverse long short-term memory networks in The hidden state of the moment; represents the Hadamard product operation; Represents the sigmoid function.

5. A method for predicting the timeliness demand for handling complaint events according to claim 4, characterized in that: The attention fusion layer includes a multi-scale convolution extraction unit, a scale interaction unit, a scale selection unit and a multi-scale feature fusion unit; The multi-scale convolution extraction unit extracts features of different scales from the outputs of the forward and reverse long short-term memory network layers through convolution layers with convolution kernels of different sizes, as shown in the following formula: ； in: express The time scale is The output features of the forward and reverse long short-term memory network layers; The scale is Convolution operation; express Time to The output sequence of the forward and reverse long short-term memory network layer at the moment; Define the interaction weight matrix, and set the size of the interaction weight matrix according to the number of scales. The interaction weight matrix is ​​specifically shown in the following formula: ； ； ； in: express The interaction weight matrix at the moment; express function; express The query matrix at the moment; express The key matrix at the moment; a matrix of learnable parameters representing the query matrix; a learnable parameter matrix representing the bond matrix; Indicates the preset feature dimension; Represents a transpose operation; The scale interaction unit updates the output features of the forward and reverse long short-term memory network layers at each scale based on the interaction weight matrix, as shown in the following formula: ； in: Indicates updated The time scale is The output features of the forward and reverse long short-term memory network layers; express The time scale is The output features of the forward and reverse long short-term memory network layers; The scale selection unit is used to calculate the contribution of different scales, as shown in the following formula: ； in: Indicates that the scale selection unit is Time output scale Contribution of represents the learnable parameter matrix of the scale selection unit; represents the bias vector of the scale selection unit; The final output features of the forward and reverse long short-term memory network layers at different scales are calculated based on the contribution of different scales, as shown in the following formula: ； The multi-scale feature fusion unit is used to calculate the final output features of the forward and reverse long short-term memory network layers at different scales by the contribution of different scales, as described in the following formula: ； ； in: Representation scale exist The position adaptive weight at the moment; represents the attention query vector; a learnable parameter matrix representing the position adaptation weights; Bias vector representing the position adaptation weights.

6. A method for predicting the timeliness demand for handling complaint events according to claim 5, characterized in that: The processing time requirement of the current complaint event is calculated based on the narrative tension of the complaint event and the future trend of the user's emotions, as shown in the following formula: ； in: Indicates the time limit required for handling the current complaint event; It represents the timeliness demand sensitivity adjustment coefficient; Represents the future prediction of the current complaint event by the user sentiment prediction model The future trend of the user's emotions at the moment; Indicates the critical threshold of situational characteristics.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the method according to any one of claims 1 to 6 is implemented.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images