Business complaint processing method and device, computer device and storage medium

Abstract

The application relates to the technical field of artificial intelligence, in particular to a business complaint processing method and device, computer equipment and a storage medium. The method comprises the following steps: adopting a trained sentiment classification model to perform text classification on a business complaint text, so as to obtain a first classification result about the sentiment polarity of the business complaint text; the sentiment polarity is divided into positive or non-positive; adopting a trained business classification model to perform text classification on the business complaint text, so as to obtain a second classification result about the business type of the business complaint text; in the case that the first classification result is non-positive, generating a query statement according to the second classification result, performing a query in a business database according to the query statement, and obtaining a query result; the query statement is used for querying a business processing record related to the business complaint text; in the case that the query result does not include the business processing record, the business complaint text is taken as a malicious complaint text. The method can improve the processing efficiency of the business complaint.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and in particular to a business complaint handling method, apparatus, computer equipment, storage medium, and computer program product. Background Technology

[0002] For businesses, the ability to efficiently handle customer complaints and respond to customer requests plays a crucial role in improving service levels, customer satisfaction, and brand image.

[0003] In related technologies, the handling of customer complaints mainly involves manual identification of the complaint content, followed by forwarding the complaint to the appropriate department for analysis. In practice, after multiple layers of processing, it may be discovered that the customer did not perform the relevant actions or that the complaint content does not match the actual situation. This necessitates a follow-up visit to the customer, re-processing and re-investigating the complaint information, resulting in low efficiency in handling business complaints. Summary of the Invention

[0004] Therefore, it is necessary to provide a business complaint handling method, apparatus, computer equipment, computer-readable storage medium, and computer program product that can improve the efficiency of handling business complaints in response to the above-mentioned technical problems.

[0005] Firstly, this application provides a method for handling business complaints. The method includes:

[0006] A trained sentiment classification model is used to classify business complaint texts to obtain a first classification result on the sentiment polarity of the business complaint texts; the sentiment polarity is divided into positive or negative.

[0007] A trained business classification model is used to classify business complaint texts to obtain a second classification result regarding the business type of the business complaint texts.

[0008] If the first classification result is not positive, a query statement is generated based on the second classification result. The query statement is then used to query the business database to obtain the query results. The query statement is used to query business processing records related to business complaint texts.

[0009] If the query results do not include business processing records, the business complaint text will be treated as a malicious complaint text.

[0010] In one embodiment, the trained sentiment classification model is obtained by means of:

[0011] Obtain historical business complaint texts;

[0012] Based on a pre-defined polarity dictionary, obtain the sentiment polarity tags of historical business complaint texts;

[0013] The initial model of the sentiment classification model is trained by inputting historical business complaint texts into the initial model and using the sentiment polarity labels of the historical business complaint texts as the target output.

[0014] In one embodiment, the method for obtaining the trained business classification model includes:

[0015] Obtain historical business complaint texts;

[0016] Based on the historical flow information corresponding to historical business complaint texts, obtain the business type tags of the historical business complaint texts;

[0017] Input historical business complaint texts into the initial model of the business classification model, use the business type labels of the historical business complaint texts as the target output, train the initial model, and obtain the business classification model.

[0018] In one embodiment, generating a query statement based on the second classification result includes:

[0019] The business complaint text is segmented into words to obtain the segmentation results; the segmentation results include multiple words.

[0020] Each term is individually tagged with part-of-speech tags to obtain the part-of-speech tagging results for each term.

[0021] According to preset rules, the words whose part-of-speech tagging results are of the target part of speech are concatenated to obtain the business attributes corresponding to the business complaint text; the business attributes are used to characterize the relevant information of business processing;

[0022] Get the preset query statement template; the preset query statement template includes the query parameter format;

[0023] Based on the query parameter format, the business attributes and the second category results are preprocessed;

[0024] The query statement is generated based on the preset query statement template, as well as the preprocessed business attributes and the second category results.

[0025] In one embodiment, the method further includes:

[0026] If the first classification result is positive, the corresponding transfer object for the business complaint text is determined based on the second classification result and business attributes.

[0027] A business work order is generated based on the business complaint text; the business work order is used to instruct the relevant parties to process the business complaint text.

[0028] In one embodiment, the method further includes:

[0029] If the query results include business processing records, determine the corresponding transfer object of the business complaint text based on the second category results and business attributes;

[0030] A business work order is generated based on the business complaint text; the business work order is used to instruct the relevant parties to process the business complaint text.

[0031] Secondly, this application also provides a business complaint processing device. The device includes:

[0032] The first classification module is used to classify business complaint texts using a trained sentiment classification model to obtain a first classification result on the sentiment polarity of the business complaint texts; the sentiment polarity is divided into positive or negative.

[0033] The second classification module is used to classify business complaint texts using a trained business classification model to obtain a second classification result regarding the business type of the business complaint text.

[0034] The query module is used to generate a query statement based on the second category result when the first category result is not positive. The query statement is then used to query the business database to obtain the query results. The query statement is used to query business processing records related to business complaint texts.

[0035] The tagging module is used to treat business complaint text as malicious complaint text when the query results do not include business processing records.

[0036] Thirdly, this application also provides a computer device. The computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to perform the following steps:

[0037] A trained sentiment classification model is used to classify business complaint texts to obtain a first classification result on the sentiment polarity of the business complaint texts; the sentiment polarity is divided into positive or negative.

[0038] A trained business classification model is used to classify business complaint texts to obtain a second classification result regarding the business type of the business complaint texts.

[0039] If the first classification result is not positive, a query statement is generated based on the second classification result. The query statement is then used to query the business database to obtain the query results. The query statement is used to query business processing records related to business complaint texts.

[0040] If the query results do not include business processing records, the business complaint text will be treated as a malicious complaint text.

[0041] Fourthly, this application also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program thereon, which, when executed by a processor, performs the following steps:

[0042] A trained sentiment classification model is used to classify business complaint texts to obtain a first classification result on the sentiment polarity of the business complaint texts; the sentiment polarity is divided into positive or negative.

[0043] A trained business classification model is used to classify business complaint texts to obtain a second classification result regarding the business type of the business complaint texts.

[0044] If the first classification result is not positive, a query statement is generated based on the second classification result. The query statement is then used to query the business database to obtain the query results. The query statement is used to query business processing records related to business complaint texts.

[0045] If the query results do not include business processing records, the business complaint text will be treated as a malicious complaint text.

[0046] Fifthly, this application also provides a computer program product. The computer program product includes a computer program that, when executed by a processor, performs the following steps:

[0047] A trained sentiment classification model is used to classify business complaint texts to obtain a first classification result on the sentiment polarity of the business complaint texts; the sentiment polarity is divided into positive or negative.

[0048] A trained business classification model is used to classify business complaint texts to obtain a second classification result regarding the business type of the business complaint texts.

[0049] If the first classification result is not positive, a query statement is generated based on the second classification result. The query statement is then used to query the business database to obtain the query results. The query statement is used to query business processing records related to business complaint texts.

[0050] If the query results do not include business processing records, the business complaint text will be treated as a malicious complaint text.

[0051] The aforementioned business complaint handling method, apparatus, computer equipment, storage medium, and computer program product first employ a trained sentiment classification model to classify the business complaint text, obtaining a classification result regarding the sentiment polarity of the text. This allows for a preliminary determination of whether the complaint content is potentially malicious based on the text's sentiment tone. Next, a trained business classification model is used to classify the business complaint text, obtaining a classification result regarding the business type of the complaint text. This identifies the business area targeted by the complaint. Then, if the sentiment polarity is negative, a query statement is generated based on the business type of the complaint content, retrieving relevant business processing records from the business database. This further determines the veracity of the complaint. Finally, if the query results do not include business processing records, the business complaint text is classified as a malicious complaint text. This prevents malicious complaint texts from circulating, thereby improving the efficiency of business complaint processing. Attached Figure Description

[0052] Figure 1 This is an application environment diagram of a business complaint handling method in one embodiment;

[0053] Figure 2 This is a flowchart illustrating a business complaint handling method in one embodiment;

[0054] Figure 3 This is a flowchart illustrating a business complaint handling method in another embodiment;

[0055] Figure 4 This is a structural block diagram of a business complaint processing device in one embodiment;

[0056] Figure 5 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0057] 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.

[0058] The business complaint handling method provided in this application embodiment can be applied to, for example, Figure 1In 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 located on a cloud or other network server. Terminal 102 can upload business complaint texts to server 104. After receiving the business complaint texts uploaded by terminal 102, server 104 can classify and query the business complaint texts, and determine whether the business complaint texts are malicious based on the classification and query results. Terminal 102 can be, but is not limited to, various personal computers, laptops, smartphones, tablets, and portable wearable devices. Portable wearable devices can be smartwatches, smart bracelets, head-mounted devices, etc. Server 104 can be implemented using a standalone server or a server cluster consisting of multiple servers.

[0059] In one embodiment, such as Figure 2 As shown, a business complaint handling method is provided, which is applied to... Figure 1 Taking server 104 as an example, the following steps are included:

[0060] S202: Use a trained sentiment classification model to classify business complaint texts to obtain a first classification result on the sentiment polarity of the business complaint texts; sentiment polarity is divided into positive or negative.

[0061] The trained sentiment classification model can perform sentiment analysis on input text, determining the corresponding sentiment polarity from a pre-defined set of sentiment polarities. This set includes sentiment polarities determined based on different sentiment types and depths; for example, it could include five sentiment polarities: very positive, positive, neutral, negative, and very negative. No specific limitation is made here. The sentiment classification model can employ recurrent neural networks or convolutional neural networks, among others.

[0062] In one optional implementation, the business complaint text is obtained by the server first acquiring a customer complaint file, which may include one or more sub-complaint files. Each sub-complaint file may be, but is not limited to, text, image, and audio files. Then, the server acquires the sub-complaint text corresponding to each sub-complaint file. For example, text-type sub-complaint files are directly used as the corresponding sub-complaint text; image-type sub-complaint files are obtained through text recognition; and audio-type sub-complaint files are obtained through speech recognition. Finally, the server concatenates all the sub-complaint texts to obtain the business complaint text.

[0063] Optionally, after obtaining the business complaint text, the server inputs the business complaint text into a binary sentiment classification model. The binary sentiment classification model will output two sentiment polarities: positive or negative. Then, based on the sentiment polarity corresponding to the business complaint text, it can be determined whether the business complaint text may be a malicious complaint text.

[0064] S204: Use the trained business classification model to classify the business complaint text to obtain a second classification result regarding the business type of the business complaint text.

[0065] Here, "business type" refers to the business domain targeted by the business complaint text. Taking banking as an example, the business type corresponding to the complaint text could be payment, deposit, wealth management, etc., without specific limitations. The trained business classification model can classify the input text into its domain to obtain the corresponding business type. The business classification model can use recurrent neural networks or convolutional neural networks, etc., without specific limitations.

[0066] Optionally, after obtaining the business complaint text, the server inputs the business complaint text into a trained business classification model to obtain the business type corresponding to the business complaint text. Then, the server can transfer the complaint content or query relevant business processing records based on the business type corresponding to the business complaint text.

[0067] S206: If the first classification result is not positive, generate a query statement based on the second classification result, and perform a query in the business database based on the query statement to obtain the query result; the query statement is used to query business processing records related to the business complaint text.

[0068] The query statement refers to an SQL (Structured Query Language) statement. The business processing record refers to records of business operations related to the customer.

[0069] Optionally, a negative first classification result indicates that the business complaint text may be malicious, requiring further verification of its content to determine its purported malicious nature. Therefore, in the case of a negative first classification result, the server first retrieves query parameters from the business complaint text, such as the time, location, and customer name of the complaint event. Then, based on the obtained query parameters and the second classification result, a query statement is generated and performed in the business database storing business processing records.

[0070] S208: If the query results do not include business processing records, treat the business complaint text as a malicious complaint text.

[0071] Optionally, the query results excluding business processing records indicate that no business operation corresponds to the business complaint text, and the content of the business complaint text is untrue. Therefore, the server stores the business complaint text separately as a malicious complaint text and does not circulate it.

[0072] In one alternative implementation, the server generates an audit view based on stored malicious complaint text for secondary manual review.

[0073] In the aforementioned business complaint handling method, a trained sentiment classification model is first used to classify the business complaint text, obtaining a classification result regarding the sentiment polarity of the text. This allows for an initial determination of whether the complaint content may be malicious based on the text's sentiment color. Next, a trained business classification model is used to classify the business complaint text, obtaining a classification result regarding the business type of the complaint text. This identifies the business area targeted by the complaint. Then, if the sentiment polarity is negative, a query is generated based on the business type of the complaint content, and a query is performed in the business database to retrieve relevant business processing records. This further determines whether the complaint content is true. Finally, if the query results do not include business processing records, the business complaint text is classified as a malicious complaint text. This prevents malicious complaint texts from circulating, thereby improving the efficiency of business complaint handling.

[0074] In one embodiment, the method for obtaining the trained sentiment classification model includes: obtaining historical business complaint texts; obtaining sentiment polarity labels of the historical business complaint texts based on a preset polarity dictionary; inputting the historical business complaint texts into the initial model of the sentiment classification model, using the sentiment polarity labels of the historical business complaint texts as the target output, training the initial model, and obtaining the sentiment classification model.

[0075] Among them, historical business complaint texts refer to business complaint texts that have been processed. The preset polarity dictionary includes sentiment words, negation words, degree adverbs, and the sentiment score corresponding to each word.

[0076] Optionally, during the training of the sentiment classification model, the server first acquires historical business complaint texts, segments them into words, and obtains the segmentation results. Then, based on the part-of-speech, position, and polarity dictionary corresponding to each word in the segmentation results, the sentiment score corresponding to the historical business complaint text is determined. Furthermore, based on the sentiment score and the preset sentiment polarity classification interval, the sentiment polarity label of the historical business complaint text is obtained.

[0077] A set of historical business complaint texts and sentiment polarity labels are used as training samples to construct a training set for the sentiment classification model. The training set is then input into a convolutional neural network (CNN) for model training, yielding the predicted sentiment polarity value for each historical business complaint text. Next, based on the predicted sentiment polarity value, sentiment polarity label, and a pre-configured loss function, the model parameters in the CNN are adjusted, and training is repeated until a preset stopping condition is reached. When the preset stopping condition is reached, the sentiment classification model is generated based on the model parameters that minimize the loss value or demonstrate the best robustness during training. The preset stopping condition can be reaching a preset number of iterations or the difference value no longer decreasing.

[0078] In this embodiment, the sentiment polarity labels of historical business complaint texts are obtained based on a preset polarity dictionary. The historical business complaint texts are then input into the initial model of the sentiment classification model. The sentiment polarity labels of the historical business complaint texts are used as the target output to train the initial model and obtain the sentiment classification model, thereby enabling accurate sentiment classification of business complaint texts.

[0079] In one embodiment, the method for obtaining the trained business classification model includes: obtaining historical business complaint texts; obtaining business type labels of historical business complaint texts based on historical flow information corresponding to the historical business complaint texts; inputting the historical business complaint texts into the initial model of the business classification model, using the business type labels of the historical business complaint texts as the target output, training the initial model, and obtaining the business classification model.

[0080] Among them, historical business complaint texts refer to business complaint texts that have been processed, and historical processing information refers to relevant information about historical business complaint texts that have been processed by specific parties.

[0081] Optionally, during the training of the business classification model, the server first obtains historical business complaint texts and their corresponding historical processing information. Then, based on the specific processing object in the historical processing information, it determines the business domain to which the historical business complaint text belongs, thereby obtaining the business type label of the historical business complaint text.

[0082] A set of historical business complaint texts and business type labels are used as training samples to construct a training set for the business classification model. The training set is then input into a convolutional neural network (CNN) for model training, yielding predicted business type values ​​for each historical business complaint text. The model parameters in the CNN are then adjusted based on the predicted business type values, business type labels, and a pre-configured loss function, and training is repeated until a preset stopping condition is reached. When the preset stopping condition is met, the business classification model is generated based on the model parameters that minimized the loss value or demonstrated the best robustness during training. The preset stopping condition can be reaching a preset number of iterations or the difference value no longer decreasing.

[0083] In this embodiment, the business type label of the historical business complaint text is obtained based on the historical flow information corresponding to the historical business complaint text. The historical business complaint text is then input into the initial model of the business classification model. The initial model is trained with the business type label of the historical business complaint text as the target output to obtain the business classification model, thereby enabling accurate business classification of the business complaint text.

[0084] In one embodiment, generating a query statement based on the second classification result includes: performing word segmentation on the business complaint text to obtain word segmentation results; the word segmentation results include multiple terms; performing part-of-speech tagging on each term to obtain the part-of-speech tagging result corresponding to each term; concatenating the terms whose part-of-speech tagging results are of the target part of speech according to preset rules to obtain the business attributes corresponding to the business complaint text; the business attributes are used to characterize relevant information related to business processing; obtaining a preset query statement template; the preset query statement template includes a query parameter format; preprocessing the business attributes and the second classification result according to the query parameter format; and generating a query statement based on the preset query statement template and the preprocessed business attributes and the second classification result.

[0085] The business attributes may include, but are not limited to, one or more of the following: location, time, resource transfer, and personnel. Location refers to the location where the complaint occurred; time refers to the time when the complaint occurred; resource transfer refers to the quantity of resources transferred in the complaint; and personnel refers to the personnel involved in handling the complaint. The target part-of-speech tag is determined based on the business attributes. For example, if the business attributes include location, time, and resource transfer, the target part-of-speech tag would correspondingly include place names, time-related words, and numerals.

[0086] Optionally, during the query generation process, the server first performs word segmentation and part-of-speech tagging on the business complaint text, for example, using a tool like Jieba. Then, according to preset rules, it concatenates entries whose part-of-speech tagging results match the target part of speech, and uses the concatenated result as the business attribute. Taking the time attribute as an example, if the target part of speech is a time word, adjacent time words in the text are concatenated to obtain the time attribute.

[0087] Then, the server obtains a preset query statement template and preprocesses the business attributes and second category results according to the query parameter format. Taking the time attribute as an example, if the time format in the preset query statement template is "YYYY-MM-DD" and the time attribute is "July 24, 2023, 16:00", the time attribute is preprocessed to obtain the processed time attribute "2023-07-24".

[0088] Finally, the server generates a query statement based on a preset query statement template, preprocessed business attributes, and the second category results. In one optional implementation, the query statement is generated using dynamic SQL.

[0089] In this embodiment, by first obtaining business attributes from the business complaint text as query parameters, then obtaining a preset query statement template, preprocessing the business attributes and second classification results according to the query parameter format in the preset query statement template, and automatically generating a query statement based on the preset query statement template and the preprocessed business attributes and second classification results, the efficiency of processing business complaints can be improved.

[0090] In one embodiment, the method further includes: if the first classification result is positive, determining the transfer object corresponding to the business complaint text based on the second classification result and business attributes; generating a business work order based on the business complaint text; the business work order is used to instruct the transfer object to process the business complaint text.

[0091] The "transfer recipient" refers to the specific entity to which the business complaint text will be processed. This recipient can be a department, a team, or a specific personnel handling the complaint; no specific limitations are imposed here. The business work order may include the business complaint text, the complainant, the complainant's contact information, and the processing deadline.

[0092] For example, if the first classification result is positive, the business type of the business complaint text is A, and the location attribute is B, the server generates a business work order, instructing a Class A professional customer service representative in area B to handle the business complaint text.

[0093] In this embodiment, when the first classification result is positive, a business work order is generated to instruct the relevant parties to process the business complaint text based on the business complaint text. In this way, non-malicious complaint texts can be automatically processed, thereby improving the efficiency of business complaint processing.

[0094] In one embodiment, the method further includes: if the query results include business processing records, determining the transfer object corresponding to the business complaint text based on the second classification result and business attributes; generating a business work order based on the business complaint text; the business work order is used to instruct the transfer object to process the business complaint text.

[0095] The "transfer recipient" refers to the specific entity to which the business complaint text will be processed. This recipient can be a department, a team, or a specific personnel handling the complaint; no specific limitations are imposed here. A business work order may include the business complaint text, the complainant, the complainant's contact information, and the processing deadline.

[0096] For example, if the query results include business processing records, and the business type of the business complaint text is A and the location attribute is B, the server generates a business work order, instructing a Class A professional customer service representative in region B to handle the business complaint text.

[0097] In this embodiment, when the query results include business processing records, a business work order is generated based on the business complaint text to instruct the transfer object to process the business complaint text. In this way, non-malicious complaint texts can be automatically transferred, thereby improving the processing efficiency of business complaints.

[0098] In one embodiment, such as Figure 3 As shown, a business complaint handling method is provided, which includes the following steps:

[0099] Obtain the business complaint text;

[0100] The trained sentiment classification model is used to classify business complaint texts, and the first classification result of sentiment polarity of business complaint texts is obtained, which is divided into positive or negative sentiment polarity.

[0101] The trained business classification model is used to classify business complaint texts, resulting in a second classification result regarding the business type of the business complaint texts.

[0102] Determine whether the first classification result is positive; if the first classification result is positive, determine the corresponding transfer object of the business complaint text based on the second classification result and business attributes, and generate a business work order based on the business complaint text; if the first classification result is not positive, continue to execute the subsequent steps.

[0103] The business complaint text is segmented into words to obtain the segmentation results; the segmentation results include multiple words.

[0104] Each term is individually tagged with part-of-speech tags to obtain the part-of-speech tagging results for each term.

[0105] According to preset rules, the words whose part-of-speech tagging results are the target part of speech are concatenated to obtain the business attributes corresponding to the business complaint text;

[0106] Get the preset query statement template, which includes the query parameter format;

[0107] Based on the query parameter format, the business attributes and the second category results are preprocessed;

[0108] Generate a query statement based on the preset query statement template, as well as the preprocessed business attributes and the second category results;

[0109] The query statement is used to query the business database and the query results are obtained.

[0110] The system determines whether the query results include business processing records. If the query results do not include business processing records, the business complaint text is stored as a malicious complaint text. If the query results include business processing records, the system determines the corresponding transfer object of the business complaint text based on the second classification result and business attributes, and generates a business work order based on the business complaint text.

[0111] 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.

[0112] Based on the same inventive concept, this application also provides a business complaint processing apparatus for implementing the business complaint processing method described above. The solution provided by this apparatus is similar to the implementation scheme described in the above method; therefore, the specific limitations in one or more business complaint processing apparatus embodiments provided below can be found in the limitations of the business complaint processing method described above, and will not be repeated here.

[0113] In one embodiment, such as Figure 4 As shown, a business complaint processing device is provided, including: a first classification module 410, a second classification module 420, a query module 430, and a tagging module 440, wherein:

[0114] The first classification module 410 is used to classify business complaint texts using a trained sentiment classification model to obtain a first classification result on the sentiment polarity of the business complaint texts; the sentiment polarity is divided into positive or negative.

[0115] The second classification module 420 is used to classify business complaint texts using a trained business classification model to obtain a second classification result regarding the business type of the business complaint texts.

[0116] The query module 430 is used to generate a query statement based on the second classification result when the first classification result is not positive, and to perform a query in the business database based on the query statement to obtain the query result; the query statement is used to query business processing records related to business complaint texts.

[0117] The tagging module 440 is used to treat business complaint text as malicious complaint text when the query results do not include business processing records.

[0118] In one embodiment, the first classification module 410 is further configured to acquire historical business complaint texts; acquire the sentiment polarity labels of the historical business complaint texts based on a preset polarity dictionary; input the historical business complaint texts into the initial model of the sentiment classification model, use the sentiment polarity labels of the historical business complaint texts as the target output, train the initial model, and obtain the sentiment classification model.

[0119] In one embodiment, the second classification module 420 is further configured to obtain historical business complaint texts; obtain business type labels of historical business complaint texts based on historical flow information corresponding to the historical business complaint texts; input the historical business complaint texts into the initial model of the business classification model, use the business type labels of the historical business complaint texts as the target output, train the initial model, and obtain the business classification model.

[0120] In one embodiment, the query module 430 is further configured to perform word segmentation on the business complaint text to obtain word segmentation results; the word segmentation results include multiple terms; each term is tagged with a part-of-speech tag to obtain the part-of-speech tagging result corresponding to each term; the terms with the part-of-speech tagging result being the target part of speech are concatenated according to preset rules to obtain the business attributes corresponding to the business complaint text; the business attributes are used to characterize the relevant information of business processing; a preset query statement template is obtained; the preset query statement template includes a query parameter format; the business attributes and the second classification results are preprocessed according to the query parameter format; and a query statement is generated according to the preset query statement template and the preprocessed business attributes and the second classification results.

[0121] In one embodiment, the query module 430 is further configured to, if the first classification result is positive, determine the transfer object corresponding to the business complaint text based on the second classification result and business attributes; generate a business work order based on the business complaint text; and use the business work order to instruct the transfer object to process the business complaint text.

[0122] In one embodiment, the tagging module 440 is further configured to, when the query results include business processing records, determine the transfer object corresponding to the business complaint text based on the second classification result and business attributes; generate a business work order based on the business complaint text; and use the business work order to instruct the transfer object to process the business complaint text.

[0123] Each module in the aforementioned business complaint processing 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.

[0124] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 5As 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 business complaint texts, model classification results, and other business data. 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 executed by the processor, the computer program implements a business complaint processing method.

[0125] Those skilled in the art will understand that Figure 5 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.

[0126] In one embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and the processor, when executing the computer program, performs the following steps: classifying a business complaint text using a trained sentiment classification model to obtain a first classification result regarding the sentiment polarity of the business complaint text; the sentiment polarity is divided into positive or negative; classifying the business complaint text using a trained business classification model to obtain a second classification result regarding the business type of the business complaint text; if the first classification result is negative, generating a query statement based on the second classification result, querying a business database based on the query statement, and obtaining query results; the query statement is used to query business processing records related to the business complaint text; if the query results do not include business processing records, the business complaint text is treated as a malicious complaint text.

[0127] In one embodiment, when the processor executes the computer program, it further performs the following steps: acquiring historical business complaint texts; acquiring the sentiment polarity labels of the historical business complaint texts based on a preset polarity dictionary; inputting the historical business complaint texts into the initial model of the sentiment classification model, using the sentiment polarity labels of the historical business complaint texts as the target output, training the initial model, and obtaining the sentiment classification model.

[0128] In one embodiment, when the processor executes the computer program, it further performs the following steps: obtaining historical business complaint texts; obtaining business type labels of historical business complaint texts based on historical flow information corresponding to the historical business complaint texts; inputting the historical business complaint texts into the initial model of the business classification model, using the business type labels of the historical business complaint texts as the target output, training the initial model, and obtaining the business classification model.

[0129] In one embodiment, when the processor executes the computer program, it further performs the following steps: performing word segmentation on the business complaint text to obtain word segmentation results; the word segmentation results include multiple terms; performing part-of-speech tagging on each term to obtain the part-of-speech tagging result corresponding to each term; concatenating the terms whose part-of-speech tagging results are of the target part of speech according to preset rules to obtain the business attributes corresponding to the business complaint text; the business attributes are used to characterize relevant information for business processing; obtaining a preset query statement template; the preset query statement template includes a query parameter format; preprocessing the business attributes and the second classification results according to the query parameter format; and generating a query statement according to the preset query statement template and the preprocessed business attributes and the second classification results.

[0130] In one embodiment, when the processor executes the computer program, it further performs the following steps: if the first classification result is positive, determine the transfer object corresponding to the business complaint text based on the second classification result and the business attribute; generate a business work order based on the business complaint text; the business work order is used to instruct the transfer object to process the business complaint text.

[0131] In one embodiment, when the processor executes the computer program, it further performs the following steps: if the query results include business processing records, determine the transfer object corresponding to the business complaint text based on the second classification result and business attributes; generate a business work order based on the business complaint text; the business work order is used to instruct the transfer object to process the business complaint text.

[0132] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, performs the following steps: classifying business complaint texts using a trained sentiment classification model to obtain a first classification result regarding the sentiment polarity of the business complaint texts; the sentiment polarity is divided into positive or negative; classifying business complaint texts using a trained business classification model to obtain a second classification result regarding the business type of the business complaint texts; if the first classification result is negative, generating a query statement based on the second classification result, querying a business database based on the query statement, and obtaining query results; the query statement is used to query business processing records related to the business complaint texts; if the query results do not include business processing records, classifying the business complaint texts as malicious complaint texts.

[0133] In one embodiment, when the computer program is executed by the processor, it further performs the following steps: acquiring historical business complaint texts; acquiring the sentiment polarity labels of the historical business complaint texts based on a preset polarity dictionary; inputting the historical business complaint texts into the initial model of the sentiment classification model, using the sentiment polarity labels of the historical business complaint texts as the target output, training the initial model, and obtaining the sentiment classification model.

[0134] In one embodiment, when the computer program is executed by the processor, it further performs the following steps: obtaining historical business complaint texts; obtaining business type labels of historical business complaint texts based on historical flow information corresponding to the historical business complaint texts; inputting the historical business complaint texts into the initial model of the business classification model, using the business type labels of the historical business complaint texts as the target output, training the initial model, and obtaining the business classification model.

[0135] In one embodiment, when the computer program is executed by the processor, it further performs the following steps: performing word segmentation on the business complaint text to obtain word segmentation results; the word segmentation results include multiple terms; performing part-of-speech tagging on each term to obtain the part-of-speech tagging result corresponding to each term; concatenating the terms whose part-of-speech tagging results are of the target part of speech according to preset rules to obtain the business attributes corresponding to the business complaint text; the business attributes are used to characterize relevant information for business processing; obtaining a preset query statement template; the preset query statement template includes a query parameter format; preprocessing the business attributes and the second classification results according to the query parameter format; and generating a query statement according to the preset query statement template and the preprocessed business attributes and the second classification results.

[0136] In one embodiment, when the computer program is executed by the processor, it further performs the following steps: if the first classification result is positive, determine the transfer object corresponding to the business complaint text based on the second classification result and the business attribute; generate a business work order based on the business complaint text; the business work order is used to instruct the transfer object to process the business complaint text.

[0137] In one embodiment, when the computer program is executed by the processor, it further performs the following steps: if the query results include business processing records, determine the transfer object corresponding to the business complaint text based on the second classification result and business attributes; generate a business work order based on the business complaint text; the business work order is used to instruct the transfer object to process the business complaint text.

[0138] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, performs the following steps: classifying business complaint text using a trained sentiment classification model to obtain a first classification result regarding the sentiment polarity of the business complaint text; the sentiment polarity is divided into positive or negative; classifying business complaint text using a trained business classification model to obtain a second classification result regarding the business type of the business complaint text; if the first classification result is negative, generating a query statement based on the second classification result, querying a business database based on the query statement, and obtaining query results; the query statement is used to query business processing records related to the business complaint text; if the query results do not include business processing records, classifying the business complaint text as malicious complaint text.

[0139] In one embodiment, when the computer program is executed by the processor, it further performs the following steps: acquiring historical business complaint texts; acquiring the sentiment polarity labels of the historical business complaint texts based on a preset polarity dictionary; inputting the historical business complaint texts into the initial model of the sentiment classification model, using the sentiment polarity labels of the historical business complaint texts as the target output, training the initial model, and obtaining the sentiment classification model.

[0140] In one embodiment, when the computer program is executed by the processor, it further performs the following steps: obtaining historical business complaint texts; obtaining business type labels of historical business complaint texts based on historical flow information corresponding to the historical business complaint texts; inputting the historical business complaint texts into the initial model of the business classification model, using the business type labels of the historical business complaint texts as the target output, training the initial model, and obtaining the business classification model.

[0141] In one embodiment, when the computer program is executed by the processor, it further performs the following steps: performing word segmentation on the business complaint text to obtain word segmentation results; the word segmentation results include multiple terms; performing part-of-speech tagging on each term to obtain the part-of-speech tagging result corresponding to each term; concatenating the terms whose part-of-speech tagging results are of the target part of speech according to preset rules to obtain the business attributes corresponding to the business complaint text; the business attributes are used to characterize relevant information for business processing; obtaining a preset query statement template; the preset query statement template includes a query parameter format; preprocessing the business attributes and the second classification results according to the query parameter format; and generating a query statement according to the preset query statement template and the preprocessed business attributes and the second classification results.

[0142] In one embodiment, when the computer program is executed by the processor, it further performs the following steps: if the first classification result is positive, determine the transfer object corresponding to the business complaint text based on the second classification result and the business attribute; generate a business work order based on the business complaint text; the business work order is used to instruct the transfer object to process the business complaint text.

[0143] In one embodiment, when the computer program is executed by the processor, it further performs the following steps: if the query results include business processing records, determine the transfer object corresponding to the business complaint text based on the second classification result and business attributes; generate a business work order based on the business complaint text; the business work order is used to instruct the transfer object to process the business complaint text.

[0144] 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 shall comply with the relevant laws, regulations and standards of the relevant countries and regions.

[0145] 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 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, etc., and are not limited to these.

[0146] 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 specification.

[0147] 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 handling business complaints, characterized in that, The method includes: A trained sentiment classification model is used to classify business complaint texts to obtain a first classification result of the sentiment polarity of the business complaint texts; the sentiment polarity is divided into positive or negative. The trained business classification model is used to classify the business complaint text to obtain a second classification result regarding the business type of the business complaint text. If the first classification result is not positive, a query statement is generated based on the second classification result, and the query statement is used to query the business database to obtain the query result; the query statement is used to query business processing records related to the business complaint text. The process of generating a query statement based on the second classification result includes: performing word segmentation on the business complaint text to obtain a word segmentation result; the word segmentation result includes multiple terms; performing part-of-speech tagging on each term to obtain a part-of-speech tagging result corresponding to each term; concatenating terms with target part-of-speech tags according to preset rules to obtain the business attributes corresponding to the business complaint text, wherein the business attributes include: location attributes, time attributes, resource transfer attributes, and personnel attributes; the business attributes are used to characterize relevant information related to business processing; obtaining a preset query statement template; the preset query statement template includes a query parameter format; preprocessing the business attributes and the second classification result according to the query parameter format; and generating the query statement based on the preset query statement template, the preprocessed business attributes, and the second classification result. If the query results do not include the business processing record, the business complaint text will be treated as a malicious complaint text. If the first classification result is positive, or if the first classification result is negative and the query result includes the business processing record, the corresponding transfer object for the business complaint text is determined based on the second classification result and the business attribute; a business work order is generated based on the business complaint text; the business work order is used to instruct the transfer object to process the business complaint text.

2. The method according to claim 1, characterized in that, The methods for obtaining the trained sentiment classification model include: Obtain historical business complaint texts; Based on a preset polarity dictionary, obtain the sentiment polarity tags of the historical business complaint texts; The historical business complaint texts are input into the initial model of the sentiment classification model, and the sentiment polarity labels of the historical business complaint texts are used as the target output to train the initial model and obtain the sentiment classification model.

3. The method according to claim 1, characterized in that, The methods for obtaining the trained business classification model include: Obtain historical business complaint texts; Based on the historical flow information corresponding to the historical business complaint text, obtain the business type tag of the historical business complaint text; The historical business complaint texts are input into the initial model of the business classification model, and the business type labels of the historical business complaint texts are used as the target output to train the initial model and obtain the business classification model.

4. A business complaint processing device, characterized in that, The device includes: The first classification module is used to classify business complaint texts using a trained sentiment classification model to obtain a first classification result of the sentiment polarity of the business complaint texts; the sentiment polarity is divided into positive or negative. The second classification module is used to classify the business complaint text using a trained business classification model to obtain a second classification result regarding the business type of the business complaint text. The query module is used to generate a query statement based on the second classification result when the first classification result is not positive, and to perform a query in the business database based on the query statement to obtain the query result; the query statement is used to query business processing records related to the business complaint text. The query module is further configured to: perform word segmentation on the business complaint text to obtain word segmentation results; the word segmentation results include multiple terms; perform part-of-speech tagging on each term to obtain the part-of-speech tagging result corresponding to each term; concatenate the terms whose part-of-speech tagging results are of the target part of speech according to preset rules to obtain the business attributes corresponding to the business complaint text, wherein the business attributes include: location attributes, time attributes, resource transfer attributes, and personnel attributes; the business attributes are used to characterize relevant information related to business processing; obtain a preset query statement template; the preset query statement template includes query parameter format; preprocess the business attributes and the second classification result according to the query parameter format; and generate the query statement according to the preset query statement template, the preprocessed business attributes, and the second classification result. The tagging module is used to identify the business complaint text as malicious complaint text when the query results do not include the business processing record. The device is further configured to: determine the transfer object corresponding to the business complaint text based on the second classification result and the business attribute when the first classification result is positive or when the first classification result is not positive and the query result includes the business processing record; generate a business work order based on the business complaint text; and instruct the transfer object to process the business complaint text.

5. 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 3.

6. 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 3.

7. 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 3.

Images