Method, device, equipment, storage medium and program product for predicting collection style information

Abstract

This application relates to a method, apparatus, computer device, computer-readable storage medium, and computer program product for predicting collection style information. The method includes: acquiring first basic feature data of the user to be predicted; inputting the first basic feature data into a prediction model to obtain the prediction result output by the prediction model, wherein the prediction result includes target collection style information corresponding to the user; the target collection style information is determined based on target dialogue data; the prediction model stores a mapping relationship between target dialogue data and customer group types; the mapping relationship is determined based on second basic feature data of sample users and initial dialogue data in each scenario; the initial dialogue data in each scenario is based on the output of a scenario classification model, which is trained based on the second basic feature data. This method improves the flexibility of the collection robot while also increasing its adaptability to customers and the resource utilization rate of the intelligent robot.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and in particular to a method, apparatus, computer device, computer-readable storage medium, and computer program product for predicting collection style information. Background Technology

[0002] With the development of artificial intelligence technology, intelligent debt collection robots have emerged. These robots can automate the debt collection process, including automatic outbound calls, automatic SMS messages, and automatic emails. Automation significantly reduces manual operations and improves debt collection efficiency.

[0003] In traditional technologies, the collection style, process, and scripts of intelligent collection robots are rigid and monotonous, operating through fixed collection procedures and scripts.

[0004] However, current collection methods or traditional approaches suffer from poor flexibility in collection style and low adaptability to users, resulting in low resource utilization of intelligent robots. Summary of the Invention

[0005] Therefore, it is necessary to provide a method, apparatus, computer equipment, computer-readable storage medium, and computer program product for predicting collection style information that can improve the flexibility and resource utilization of intelligent robots, in order to address the above-mentioned technical problems.

[0006] Firstly, this application provides a method for predicting collection style information, the method comprising:

[0007] Obtain the first basic feature data of the predicted user;

[0008] The first basic feature data is input into the prediction model to obtain the prediction result output by the prediction model. The prediction result includes the target collection style information corresponding to the predicted user. The target collection style information is determined based on the target dialogue data. The prediction model stores the mapping relationship between the target dialogue data and the customer group type. The mapping relationship is determined based on the second basic feature data of the sample user and the initial dialogue data in each scenario. The initial dialogue data in each scenario is output based on the scenario classification model, which is used to classify the collection agent's interactive text.

[0009] In one embodiment, the mapping relationship is generated in the following ways:

[0010] For each scenario, the initial dialogue data is used to determine the second basic feature data associated with the scenario; wherein, the initial dialogue data includes the initial dialogue content and the initial dialogue order corresponding to the initial dialogue content;

[0011] The customer group type is determined based on the second basic feature data;

[0012] The initial dialogue content and the initial dialogue order in each scenario are sorted and superimposed to determine each target dialogue content and the target dialogue order corresponding to the target dialogue content;

[0013] The target speech content, the target speech order, and the customer group type are associated to determine the mapping relationship between the target speech data and the customer group type. The target speech data includes the target speech content and the target speech order.

[0014] In one embodiment, the method for determining the initial speech data includes:

[0015] Obtain the target collection agent interaction text for the target time period; wherein, the target collection agent interaction text indicates text that there is user repayment data within a preset time period after the collection agent interaction;

[0016] Semantic understanding is performed on the target collection human interaction text to obtain the semantics corresponding to each text in the target collection human interaction text, and the number of the semantics is counted;

[0017] Based on the number of semantics, word embedding encoding is performed on the semantics corresponding to each text to determine the vector representation of the target collection human interaction text;

[0018] The vector representation is labeled with intent based on the sample annotations;

[0019] The vector representations after intent annotation are classified to obtain initial speech data for each scenario.

[0020] In one embodiment, obtaining the target collection interaction text for the target time period includes:

[0021] Acquire historical collection agent interaction texts and historical user repayment data for the target time period;

[0022] For each target user corresponding to the historical user repayment data, determine the historical collection agent interaction text corresponding to the target user;

[0023] The historical collection agent interaction text corresponding to the target user is used as the target collection agent interaction text for the target time period.

[0024] In one embodiment, the method further includes:

[0025] Based on the first basic characteristic data of the predicted users, the customer group type is determined;

[0026] Based on the customer group type, determine the feature vector;

[0027] By processing the feature vector through the mapping relationship between the target speech data and the customer group type, the target speech data corresponding to the predicted user can be obtained;

[0028] The target collection style information is determined based on the target dialogue data.

[0029] In one embodiment, determining the feature vector based on the customer group type includes:

[0030] Based on the customer group type, determine the feature fields;

[0031] Obtain the first feature field that is empty from the feature fields, and predict the first feature field to obtain the fitted value of the first feature field;

[0032] The fitted value of the first feature field and the actual value of the second feature field are encoded, normalized, and processed into a vector to obtain the feature vector; wherein the second feature field indicates the feature field that is not null.

[0033] Secondly, this application provides a device for predicting collection style information, the device comprising:

[0034] The acquisition module is used to acquire the first basic feature data of the predicted user;

[0035] The input / output module is used to input the first basic feature data into the prediction model and obtain the prediction result output by the prediction model. The prediction result includes the target collection style information corresponding to the predicted user. The target collection style information is determined based on the target dialogue data. The prediction model stores the mapping relationship between the target dialogue data and the customer group type. The mapping relationship is determined based on the second basic feature data of the sample user and the initial dialogue data in each scenario. The initial dialogue data in each scenario is output by the scenario classification model, which is used to classify the collection human interaction text.

[0036] Thirdly, this application provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the above-described method.

[0037] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described method.

[0038] Fifthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the above-described method.

[0039] The aforementioned method, apparatus, computer equipment, computer-readable storage medium, and computer program product for predicting collection style information firstly utilize initial dialogue data for each scenario output by a scenario classification model. Secondly, based on the initial dialogue data for each scenario and the second basic feature data of sample users, a mapping relationship between target dialogue data and customer group types is determined. Different target dialogue data is matched for different customer group types, and the mapping relationship is stored in the prediction model. Finally, the first basic feature data of the predicted user is input into the prediction model to obtain the target collection style information corresponding to the predicted user output by the prediction model. The prediction model processes the first basic feature data of the predicted user and outputs the collection style information for the customer group type corresponding to the predicted user. By outputting different collection style information for different predicted users, the problem of single collection style information in traditional technologies is avoided, improving the flexibility of the collection robot, enhancing its adaptability to customers, and increasing the resource utilization rate of the intelligent robot. Attached Figure Description

[0040] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0041] Figure 1 This is an application environment diagram of a method for predicting collection style information in one embodiment.

[0042] Figure 2 This is a flowchart illustrating a method for predicting collection style information in one embodiment;

[0043] Figure 3 This is a flowchart illustrating how the mapping relationship is generated in one embodiment;

[0044] Figure 4 This is a flowchart illustrating how initial dialogue data is determined in one embodiment;

[0045] Figure 5 This is a flowchart illustrating the process of obtaining target collection interaction text for a target time period in one embodiment;

[0046] Figure 6 This is a flowchart illustrating the processing method of the prediction model in one embodiment;

[0047] Figure 7 This is a flowchart illustrating the process of determining feature vectors based on customer group type in one embodiment.

[0048] Figure 8 This is a structural block diagram of a device for predicting collection style information in one embodiment;

[0049] Figure 9 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0050] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0051] The method for predicting collection style information provided in this application can be applied to, for example... Figure 1 In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system can store the data that server 104 needs to process. The data storage system can be integrated onto server 104 or placed on a cloud or other network server. Server 104 acquires the first basic feature data of the predicted user; inputs the first basic feature data into the prediction model, and obtains the prediction result output by the prediction model. The prediction result includes the target collection style information corresponding to the predicted user; the target collection style information is determined based on target dialogue data; the prediction model stores the mapping relationship between target dialogue data and customer group type; the mapping relationship is determined based on the second basic feature data of the sample user and the initial dialogue data in each scenario; the initial dialogue data in each scenario is based on the output of a scenario classification model, which is trained based on the second basic feature data. Terminal 102 can be, but is not limited to, various personal computers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. IoT devices can be smart speakers, smart TVs, smart air conditioners, smart in-vehicle devices, projection devices, etc. Portable wearable devices can be smartwatches, smart bracelets, head-mounted devices, etc. Headset devices can be virtual reality (VR) devices, augmented reality (AR) devices, smart glasses, etc. Server 104 can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services.

[0052] In one exemplary embodiment, such as Figure 2 As shown, a method for predicting collection style information is provided, which can be applied to... Figure 1Taking the server in the example, the explanation includes the following steps S202 to S204. Wherein:

[0053] Step S202: Obtain the first basic feature data of the predicted user.

[0054] The first basic feature data is the basic feature data for predicting users, such as name, gender, card number, address, etc.

[0055] Optionally, the server obtains the predicted user and retrieves the corresponding first basic feature data based on the predicted user, such as name, gender, card number, address, etc.

[0056] Step S204: Input the first basic feature data into the prediction model and obtain the prediction result output by the prediction model.

[0057] The prediction results include the target collection style information corresponding to the predicted user; the target collection style information is determined based on the target dialogue data; the prediction model stores the mapping relationship between the target dialogue data and the customer group type; the mapping relationship is determined based on the second basic feature data of the sample users and the initial dialogue data in each scenario; the initial dialogue data in each scenario is based on the output of the scenario classification model, which is used to classify the collection manual interaction text.

[0058] Optionally, before inputting the first basic feature data into the prediction model, the server inputs the second basic feature data, user repayment data, and collection agent interaction text into the already trained classification model, and obtains the initial dialogue data for each scenario output by the classification model.

[0059] Optionally, before inputting the first basic feature data into the prediction model, the server trains the initial prediction model using the second basic feature data of the sample users and the initial dialogue data of each scenario output by the scenario classification model. By using the second basic feature data of the sample users and the initial dialogue data of each scenario, the server determines the mapping relationship between the target dialogue data and the customer group type, and stores the mapping relationship in the prediction model to obtain the trained prediction model.

[0060] Optionally, the server inputs the first basic feature data into the trained prediction model and obtains the prediction results output by the prediction model, which include the target collection style information corresponding to the predicted user.

[0061] In the aforementioned method for predicting collection style information, firstly, initial dialogue data for each scenario is output by a scenario classification model. Secondly, based on the initial dialogue data for each scenario and the second basic feature data of sample users, the mapping relationship between target dialogue data and customer group type is determined. Different target dialogue data is matched for different customer group types, and the mapping relationship is stored in the prediction model. Finally, the first basic feature data of the predicted user is input into the prediction model to obtain the target collection style information corresponding to the predicted user output by the prediction model. The prediction model processes the first basic feature data of the predicted user and outputs the collection style information for the customer group type corresponding to the predicted user. By outputting different collection style information for different predicted users, the problem of single collection style information in traditional technologies is avoided. This improves the flexibility of the collection robot, enhances its adaptability to customers, and increases the resource utilization rate of the intelligent robot.

[0062] Following the above embodiments, as Figure 3 As shown, the generation method of the mapping relationship includes steps S302 to S308. Wherein:

[0063] Step S302: For the initial dialogue data in each scenario, determine the second basic feature data associated with the scenario.

[0064] The initial script data includes the initial script content and the initial script order corresponding to the initial script content.

[0065] Optionally, for the initial dialogue data in each scenario, the server determines the sample users associated with each scenario, and determines the corresponding second basic feature data based on the sample users. For example, the second basic feature data of sample user A associated with scenario 1; the second basic feature data of sample user B associated with scenario 2, and so on. For example, there are 100 associated sample users in 100 scenarios.

[0066] Step S304: Determine the customer group type based on the second basic feature data.

[0067] Optionally, the server filters out common feature data based on the second basic feature data of each sample user, such as the 100 sample users being between 20 and 30 years old and having an income between xx and xx, and determines the customer group type of the 100 sample users as a customer group type.

[0068] Step S306: Sort and superimpose the initial dialogue content and initial dialogue order in each scenario to determine the target dialogue content and the target dialogue order corresponding to the target dialogue content.

[0069] Optionally, the server sorts and superimposes the initial dialogue content and initial dialogue order for each scenario. For example, the initial dialogue content and initial dialogue order for scenario 1 are abcde; the initial dialogue content and initial dialogue order for scenario 2 are abch; the initial dialogue content and initial dialogue order for scenario 3 are abegi... The initial dialogue content and initial dialogue order for 100 scenarios are sorted and superimposed, and the sorted and superimposed initial dialogue data is used as the target dialogue content and target dialogue order.

[0070] It should be noted that there are no restrictions on the initial dialogue content or the order of the initial dialogue in each scenario.

[0071] Step S308: Associate the target speech content, target speech order, and customer group type to determine the mapping relationship between the target speech data and the customer group type. The target speech data includes the target speech content and the target speech order.

[0072] Optionally, the server processes the initial dialogue content and initial dialogue sequence data for several scenarios to obtain the target dialogue content and target dialogue sequence; the server processes the second basic feature data of several sample users to obtain the customer group type. The server then associates the target dialogue content, target dialogue sequence, and customer group type to determine the mapping relationship between the target dialogue data and the customer group type.

[0073] In this embodiment, by associating and matching user feature data with the initial dialogue content and initial dialogue order of each scenario output by the corresponding classification model, target dialogue content and target dialogue order of the corresponding customer group type are formed and stored in the prediction model as the basis for the subsequent prediction model to process the first basic feature data of the predicted user.

[0074] Following the previous embodiment, as Figure 4 As shown, the method for determining the initial dialogue data includes steps S402 to S410. Wherein:

[0075] Step S402: Obtain the target collection interaction text for the target time period.

[0076] The target collection agent's interactive text indicates text containing user repayment data within a preset timeframe following the collection agent's interaction. For example, if new user repayment data is added within two hours of the collection agent's interaction, the target collection agent's interactive text will be used as the target text. This target collection agent's interactive text is considered a high-performing interactive text, and the collection agent's style information will be extracted from the target collection agent's interactive text to increase the human-like nature of the collection robot.

[0077] Optionally, the server obtains target collection human interaction texts for the target time period through a classification model, such as several target collection human interaction texts in December 2024.

[0078] Step S404: Perform semantic understanding on the target collection human interaction text to obtain the semantics corresponding to each text in the target collection human interaction text, and count the number of semantics.

[0079] Optionally, the server performs semantic understanding on each target collection human interaction text using a classification model to obtain the semantics corresponding to each text in the target collection human interaction text, and counts the number of semantics. For example, if there are 200 sets of dialogue text in one target collection human interaction text, the server performs semantic understanding on each text to obtain the corresponding semantics. Suppose that there are 3 sets of dialogue text with the same semantics and 10 sets of dialogue text with the same semantics. The number of semantics corresponding to the target collection human interaction text may be any number less than the number of texts.

[0080] Furthermore, the server performs semantic understanding on each text. For the same text WB, it outputs three semantics. For example, the probability of text WB corresponding to semantic 1 is 80%, the probability of corresponding to semantic 2 is 70%, and the probability of corresponding to semantic 3 is 60%. The semantic with the highest probability is selected as the semantic of the text.

[0081] Step S406: Based on the number of semantics, perform word embedding encoding on the semantics corresponding to each text to determine the vector representation of the target collection manual interaction text.

[0082] Optionally, the server encodes each semantic element using a classification model to obtain a fixed-dimensional vector, which is used to represent that semantic element. The number of rows corresponds to the number of words in the input text, i.e., the number of semantic elements, and the number of columns is the dimension of the encoding. The server performs word embedding encoding on the semantic element corresponding to each text to determine the vector representation of the target collection human interaction text.

[0083] Step S408: Perform intent annotation on the vector representation based on the sample annotations.

[0084] Among them, the sample annotations are reference annotations and can be manually completed annotations.

[0085] Optionally, the server uses a classification model to perform intent labeling on a large number of vector representations based on the sample labels. The intent label for vector representation 1 is A; the intent label for vector representation 2 is A; the intent label for vector representation 3 is B, and so on.

[0086] Step S410: Classify the vector representations after intent annotation to obtain initial dialogue data for each scenario.

[0087] Optionally, the server classifies the vector representations of intent annotations according to scenarios using a classification model to obtain initial dialogue data for each scenario.

[0088] In this embodiment, the high-performing human-interaction text of debt collectors is processed by a classification model and classified according to the scenario. This allows the initial dialogue data for each scenario to be obtained. The initial dialogue data for each scenario is used as training data to train the prediction model. When using the debt collection robot for debt collection, the style of human debt collectors is absorbed to increase the human-likeness of the debt collection robot.

[0089] Continuing with the above embodiments, as follows: Figure 5 As shown, obtaining the target collection interaction text for the target time period includes steps S502 to S506. Wherein:

[0090] Step S502: Obtain historical collection interaction texts and historical user repayment data for the target time period.

[0091] Optionally, the server retrieves historical collection agent interaction texts and historical user repayment data for a target time period. For example, it retrieves historical collection agent interaction texts and historical user repayment data for December 2024. In this case, the historical collection agent interaction texts and historical user repayment data are two independent sets of data.

[0092] Step S504: For each target user corresponding to the historical user repayment data, determine the historical collection manual interaction text corresponding to the target user.

[0093] Optionally, for each target user corresponding to historical user repayment data, the historical collection interaction text corresponding to the target user can be matched from the historical collection interaction text.

[0094] Step S506: Use the historical collection interaction text corresponding to the target user as the target collection interaction text for the target time period.

[0095] Optionally, the server may use the historical collection interaction text corresponding to the target user as the target collection interaction text for the target time period.

[0096] In this embodiment, by filtering historical user repayment data to obtain corresponding historical collection human interaction texts, and using them as target collection human interaction texts for the target time period, it is possible to filter out collection human interaction texts with collection effectiveness, and use them as training data for classification and prediction models, so as to enrich the customer segmentation rules of the intelligent collection robot and enhance the adaptability of the collection robot to changes in the market collection environment.

[0097] In one exemplary embodiment, such as Figure 6As shown, the processing procedure of the prediction model includes steps S602 to S608. Wherein:

[0098] Step S602: Determine the customer group type based on the first basic characteristic data of the predicted users.

[0099] Optionally, the server analyzes the first basic characteristic data of the predicted users through a predictive model to determine the customer group type, such as the characteristic fields included in the customer group type.

[0100] Step S604: Determine the feature vector based on the customer group type.

[0101] Optionally, the server determines different feature fields based on customer group type using a prediction model. For example, customer group type 1 corresponds to feature field 1, feature field 2, and feature field 3; customer group type 2 corresponds to feature field 1, feature field 2, and feature field 4, and so on. The server encodes, normalizes, and vectorizes each feature field to obtain a feature vector.

[0102] Step S606: By processing the feature vector through the mapping relationship between the target speech data and the customer group type, the target speech data corresponding to the predicted user is obtained.

[0103] Optionally, the prediction model includes a hidden layer and an output layer. The server calculates the feature vector based on the mapping relationship between the target speech data and the customer group type through the hidden layer of the prediction model, and passes the calculated feature vector to the output layer. The Softmax activation function in the output layer transforms the output value into a probability. For example, the probability of the output target speech data 1 is 75%, the probability of target speech data 2 is 95%, and the probability of target speech data 3 is 60%. The target speech data with the highest probability value is obtained as the target speech data corresponding to the predicted user.

[0104] Step S608: Determine the target collection style information based on the target script data.

[0105] Optionally, the server determines target collection style information based on target script data using a predictive model. For example, the server determines target collection style information that will produce good collection results for the predicted users based on the target script content and order in the target script data. For instance, for a certain customer group type, a strong and decisive target collection style will have a good collection effect; the server then uses the target script content and order corresponding to the strong and decisive target collection style to conduct collection operations on the users.

[0106] In this embodiment, a predictive model is used to determine the target collection style information and target communication script data for the predicted user. Compared with collection methods with fixed scenarios and fixed communication scripts, this approach is more targeted, flexible, and better suited to the collection user.

[0107] In one exemplary embodiment, such as Figure 7 As shown, the feature vector is determined based on the customer group type, including steps S702 to S706. Wherein:

[0108] Step S702: Determine the feature fields based on the customer group type.

[0109] Optionally, the server determines different feature fields based on the customer group type. For example, customer group type 1 corresponds to feature field 1, feature field 2, and feature field 3; customer group type 2 corresponds to feature field 1, feature field 2, and feature field 4, and so on.

[0110] Step S704: Obtain the first feature field that is empty from the feature fields, and predict the first feature field to obtain the fitted value of the first feature field.

[0111] Among them, feature fields with empty values ​​are used as the first feature field, and feature fields with non-empty values ​​are used as the second feature field.

[0112] Optionally, the server obtains the first feature field that is empty from the feature fields, and performs fitting and completion using different calculation methods (statistics, algorithm prediction, etc.) to determine the fitted value of the first feature field.

[0113] Step S706: The fitted value of the first feature field and the actual value of the second feature field are encoded, normalized, and processed into a vector to obtain a feature vector.

[0114] The second feature field indicates the feature field that is not null.

[0115] Optionally, the server encodes, normalizes, and vectorizes the fitted values ​​of the first feature field and the actual values ​​of the second feature field to obtain a feature vector.

[0116] Optionally, the fitted values ​​of the first feature field and the actual values ​​of the second feature field are encoded, normalized, and processed into vectors to obtain feature vectors. This includes: extracting the first and second feature fields from the first basic feature data; encoding the first feature field, its fitted value, and the actual values ​​of the second and second feature fields; converting the encoded fitted values ​​of the first feature field and the encoded actual values ​​of the second feature field into dimensionless data, and normalizing the dimensionless data; and determining the feature vector corresponding to the first basic feature data based on the normalized dimensionless data. The server will mathematically encode different feature fields of the existing data, including the first and second feature fields, such as occupation, risk level, and whether it is overdue, etc., according to their values. Through data standardization, the original data is converted into dimensionless data, so that the value range of each indicator is at the same order of magnitude, eliminating the differences (dimensions) between features. All the dimensionally normalized feature data are taken as multi-dimensional (n-dimensional) vectors and combined into an m*n matrix according to a fixed number of rows (m rows).

[0117] In this embodiment, by encoding, normalizing, and vectorizing the feature fields, feature vectors or matrices are obtained, which can improve the computational efficiency of the prediction model and reduce the training time of the prediction model.

[0118] In an exemplary embodiment, in the first stage: the server inputs the second basic feature data, user repayment data, and collection agent interaction text into the already trained classification model, and obtains the initial dialogue data for each scenario output by the classification model. The server trains the initial prediction model using the second basic feature data of the sample users and the initial dialogue data for each scenario output by the scenario classification model. Using the second basic feature data of the sample users and the initial dialogue data for each scenario, the server determines the mapping relationship between the target dialogue data and the customer group type, and stores the mapping relationship in the prediction model, thus obtaining the trained prediction model.

[0119] The second stage involves the server acquiring predicted users and obtaining their corresponding primary basic feature data, such as name, gender, card number, and address. The server inputs this primary basic feature data into the trained prediction model and obtains the model's output prediction results. These results include the target collection style information corresponding to the predicted user. The target collection style information is determined based on target dialogue data. The prediction model stores a mapping relationship between target dialogue data and customer group types. This mapping relationship is determined based on the sample users' secondary basic feature data and the initial dialogue data for each scenario. The initial dialogue data for each scenario is based on the output of a scenario classification model, which is used to classify the collection agent's human interaction text.

[0120] The mapping relationship is generated as follows: For the initial dialogue data in each scenario, the server determines the sample users associated with each scenario and determines the corresponding second basic feature data based on the sample users. For example, the second basic feature data of sample user A associated with scenario 1; the second basic feature data of sample user B associated with scenario 2, and so on. For example, there are 100 sample users associated with 100 scenarios. Based on the second basic feature data of each sample user, the server filters out common feature data, such as the 100 sample users being between 20 and 30 years old and having an income between xx and xx, thus determining the customer group type of the 100 sample users as one customer group type. The server sorts and superimposes the initial dialogue content and order for each scenario. For example, the initial dialogue content and order for scenario 1 are abcde; for scenario 2, they are abch; for scenario 3, they are abegi... The server sorts and superimposes the initial dialogue content and order for 100 scenarios, using the resulting sorted and superimposed initial dialogue data as the target dialogue content and order. The server processes the initial dialogue content and order data from several scenarios to obtain the target dialogue content and order. The server also processes the second basic feature data of several sample users to obtain the customer group type. The server then associates the target dialogue content, target dialogue order, and customer group type to determine the mapping relationship between the target dialogue data and the customer group type.

[0121] How to determine the initial script data:

[0122] The server retrieves historical collection agent interaction texts and historical user repayment data for a target time period. For example, it retrieves historical collection agent interaction texts and historical user repayment data for December 2024. In this case, the historical collection agent interaction texts and historical user repayment data are two independent sets of data. For each target user corresponding to the historical user repayment data, the server matches the corresponding historical collection agent interaction text from the historical collection agent interaction texts. The server uses the historical collection agent interaction texts corresponding to the target user as the target collection agent interaction texts for the target time period. The server performs semantic understanding on each target collection agent interaction text using a classification model to obtain the semantics corresponding to each text in the target collection agent interaction text and counts the number of semantics. For example, if there are 200 sets of dialogue texts in one target collection agent interaction text, the server performs semantic understanding on each text to obtain the corresponding semantics. Assuming that one type of semantics appears in 3 sets of dialogue texts and another type of semantics appears in 10 sets of dialogue texts, the number of semantics corresponding to the target collection agent interaction text may be any number less than the number of texts.

[0123] Furthermore, the server performs semantic understanding on each text. For the same text WB, it outputs three semantics. For example, text WB has an 80% probability of corresponding to semantic 1, a 70% probability of corresponding to semantic 2, and a 60% probability of corresponding to semantic 3. The semantic with the highest probability is selected as the semantic of the text. The server encodes each semantic into a fixed-dimensional vector using a classification model to represent that semantic. The number of rows corresponds to the number of words in the input text, i.e., the number of semantics, and the number of columns is the dimension of the encoding. The server performs word embedding encoding on the semantics corresponding to each text to determine the vector representation of the target collection human interaction text. The server uses a classification model to perform intent annotation on a large number of vector representations based on sample annotations. The intent annotation for vector representation 1 is A; the intent annotation for vector representation 2 is A; the intent annotation for vector representation 3 is B, and so on. The server uses a classification model to classify the vector representations with intent annotation according to scenarios to obtain the initial dialogue data for each scenario.

[0124] The processing steps of the prediction model include: The server analyzes the first basic characteristic data of the predicted users using the prediction model to determine the customer group type, such as the feature fields included in the customer group type. Based on the customer group type, the server determines the corresponding different feature fields. For example, customer group type 1 corresponds to feature field 1, feature field 2, and feature field 3; customer group type 2 corresponds to feature field 1, feature field 2, and feature field 4, etc. The server retrieves the first feature field that is empty from the feature fields and performs fitting and completion using different calculation methods (statistical, algorithmic prediction, etc.) to determine the fitted value of the first feature field. The server extracts the first feature field and the second feature field from the first basic characteristic data, and encodes the first feature field, the fitted value of the first feature field, and the actual values ​​of the second and third feature fields. The encoded fitted values ​​of the first feature field and the encoded actual values ​​of the second feature field are converted into dimensionless data, and the dimensionless data is normalized. Based on the normalized dimensionless data, the feature vector corresponding to the first basic characteristic data is determined. The prediction model consists of a hidden layer and an output layer. The server calculates feature vectors based on the mapping relationship between target dialogue data and customer group types through the hidden layer of the prediction model. These calculated feature vectors are then passed to the output layer. The softmax activation function in the output layer transforms the output values ​​into probabilities. For example, the probability of target dialogue data 1 is 75%, target dialogue data 2 is 95%, and target dialogue data 3 is 60%. The target dialogue data with the highest probability value is used as the target dialogue data for the predicted user. Based on the target dialogue data, the server determines target collection style information. For instance, based on the target dialogue content and order in the target dialogue data, the server determines the target collection style information that will have a good collection effect on the predicted user. For example, for a certain customer group type, a strong and decisive target collection style will have a good collection effect; the server uses the target dialogue content and order corresponding to the strong and decisive target collection style information to carry out collection operations on the user.

[0125] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0126] Based on the same inventive concept, this application also provides a collection style information prediction device for implementing the collection style information prediction method described above. The solution provided by this device is similar to the implementation described in the above method; therefore, the specific limitations in one or more embodiments of the collection style information prediction device provided below can be found in the limitations of the collection style information prediction method described above, and will not be repeated here.

[0127] In one exemplary embodiment, such as Figure 8 As shown, a device for predicting collection style information is provided, comprising: an acquisition module 801 and an input / output module 802, wherein:

[0128] The acquisition module 801 is used to acquire the first basic feature data of the predicted user;

[0129] The input / output module 802 is used to input the first basic feature data into the prediction model and obtain the prediction results output by the prediction model. The prediction results include the target collection style information corresponding to the predicted user. The target collection style information is determined based on the target dialogue data. The prediction model stores the mapping relationship between the target dialogue data and the customer group type. The mapping relationship is determined based on the second basic feature data of the sample user and the initial dialogue data in each scenario. The initial dialogue data in each scenario is based on the output of the scenario classification model, which is used to classify the collection human interaction text.

[0130] In one exemplary embodiment, a collection style information prediction device further includes:

[0131] The mapping relationship generation module is used to determine the second basic feature data associated with the initial dialogue data in each scenario. The initial dialogue data includes the initial dialogue content and the corresponding initial dialogue order. Based on the second basic feature data, the customer group type is determined. The initial dialogue content and initial dialogue order in each scenario are sorted and superimposed to determine each target dialogue content and the corresponding target dialogue order. The target dialogue content, target dialogue order, and customer group type are associated to determine the mapping relationship between the target dialogue data and the customer group type. The target dialogue data includes the target dialogue content and target dialogue order.

[0132] In an exemplary embodiment, the mapping relationship generation module includes an initial dialogue data determination unit, used to acquire target collection human interaction text for a target time period; wherein, the target collection human interaction text indicates text indicating that there is user repayment data within a preset time after the collection human interaction; semantic understanding is performed on the target collection human interaction text to obtain the semantics corresponding to each text in the target collection human interaction text, and the number of semantics is counted; based on the number of semantics, word embedding encoding is performed on the semantics corresponding to each text to determine the vector representation of the target collection human interaction text; intent annotation is performed on the vector representation based on sample annotation; the vector representation with completed intent annotation is classified to obtain initial dialogue data for each scenario.

[0133] In an exemplary embodiment, the initial script data determination unit includes a target collection text acquisition subunit, which is used to acquire historical collection human interaction text and historical user repayment data for a target time period; for each target user corresponding to the historical user repayment data, determine the historical collection human interaction text corresponding to the target user; and use the historical collection human interaction text corresponding to the target user as the target collection human interaction text for the target time period.

[0134] In one exemplary embodiment, a collection style information prediction device further includes:

[0135] The prediction model processing module is used to determine the customer group type based on the first basic feature data of the predicted user; determine the feature vector based on the customer group type; process the feature vector through the mapping relationship between the target dialogue data and the customer group type to obtain the target dialogue data corresponding to the predicted user; and determine the target collection style information based on the target dialogue data.

[0136] In an exemplary embodiment, the prediction model processing module includes a feature vector determination unit, used to determine feature fields according to customer group type; obtain a first feature field that is null from the feature fields, and predict the first feature field to obtain a fitted value of the first feature field; encode, normalize and vectorize the fitted value of the first feature field and the actual value of the second feature field to obtain a feature vector; wherein the second feature field indicates a feature field that is not null.

[0137] Each module in the aforementioned collection style information prediction device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0138] In one exemplary embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 9 As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operating system and computer programs in the non-volatile storage media to run. The database stores basic characteristic data and collection style information. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network connection. When the computer program is executed by the processor, it implements a method for predicting collection style information.

[0139] Those skilled in the art will understand that Figure 9 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0140] In one embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above method embodiments.

[0141] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the steps in the above method embodiments.

[0142] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0143] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0144] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0145] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0146] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for predicting collection style information, characterized in that, The method includes: Obtain the first basic feature data of the predicted user; The first basic feature data is input into the prediction model to obtain the prediction result output by the prediction model. The prediction result includes the target collection style information corresponding to the predicted user. The target collection style information is determined based on the target dialogue data. The prediction model stores the mapping relationship between the target dialogue data and the customer group type. The mapping relationship is determined based on the second basic feature data of the sample user and the initial dialogue data in each scenario. The initial dialogue data in each scenario is output based on the scenario classification model, which is used to classify the collection agent's interactive text.

2. The method according to claim 1, characterized in that, The methods for generating the mapping relationship include: For each scenario, the initial dialogue data is used to determine the second basic feature data associated with the scenario; wherein, the initial dialogue data includes the initial dialogue content and the initial dialogue order corresponding to the initial dialogue content; The customer group type is determined based on the second basic feature data; The initial dialogue content and the initial dialogue order in each scenario are sorted and superimposed to determine the target dialogue content and the target dialogue order corresponding to the target dialogue content; The target speech content, the target speech order, and the customer group type are associated to determine the mapping relationship between the target speech data and the customer group type. The target speech data includes the target speech content and the target speech order.

3. The method according to claim 2, characterized in that, The method for determining the initial dialogue data includes: Obtain the target collection agent interaction text for the target time period; wherein, the target collection agent interaction text indicates text that there is user repayment data within a preset time period after the collection agent interaction; Semantic understanding is performed on the target collection human interaction text to obtain the semantics corresponding to each text in the target collection human interaction text, and the number of the semantics is counted; Based on the number of semantics, word embedding encoding is performed on the semantics corresponding to each text to determine the vector representation of the target collection human interaction text; The vector representation is labeled with intent based on the sample annotations; The vector representations after intent annotation are classified to obtain initial speech data for each scenario.

4. The method according to claim 3, characterized in that, The acquisition of the target collection interaction text for the target time period includes: Acquire historical collection agent interaction texts and historical user repayment data for the target time period; For each target user corresponding to the historical user repayment data, determine the historical collection agent interaction text corresponding to the target user; The historical collection agent interaction text corresponding to the target user is used as the target collection agent interaction text for the target time period.

5. The method according to claim 1, characterized in that, The method further includes: Based on the first basic characteristic data of the predicted users, the customer group type is determined; Based on the customer group type, determine the feature vector; By processing the feature vector through the mapping relationship between the target speech data and the customer group type, the target speech data corresponding to the predicted user can be obtained; The target collection style information is determined based on the target dialogue data.

6. The method according to claim 5, characterized in that, The step of determining the feature vector based on the customer group type includes: Based on the customer group type, determine the feature fields; Obtain the first feature field that is empty from the feature fields, and predict the first feature field to obtain the fitted value of the first feature field; The fitted value of the first feature field and the actual value of the second feature field are encoded, normalized, and processed into a vector to obtain the feature vector; wherein the second feature field indicates the feature field that is not null.

7. A device for predicting collection style information, characterized in that, The device includes: The acquisition module is used to acquire the first basic feature data of the predicted user; The input / output module is used to input the first basic feature data into the prediction model and obtain the prediction result output by the prediction model. The prediction result includes the target collection style information corresponding to the predicted user. The target collection style information is determined based on the target dialogue data. The prediction model stores the mapping relationship between the target dialogue data and the customer group type. The mapping relationship is determined based on the second basic feature data of the sample user and the initial dialogue data in each scenario. The initial dialogue data in each scenario is output by the scenario classification model, which is used to classify the collection human interaction text.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

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

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images