Methods, devices and electronic equipment for determining broadband network information

Abstract

This application discloses a method, apparatus, and electronic device for determining broadband network information. The method includes: acquiring a first complaint text of the broadband network and a first performance indicator of the broadband network; and determining the broadband network complaint type and broadband network quality score based on the first complaint text and the first performance indicator.

Description

Technical Field

[0001] This application relates to the field of broadband networks, and in particular to a method, apparatus and electronic device for determining broadband network information. Background Technology

[0002] Home broadband networks are an indispensable part of people's daily work and life. In actual use, users may encounter problems such as network latency, network instability, and slowdown, which affect the user experience. Therefore, operators need to quickly and accurately understand and solve network problems in order to improve the user experience.

[0003] In related technologies, traditional topic modeling models (such as LDA (Latent Dirichlet Allocation)) are typically used to identify the types of user complaints about broadband networks. Multiple models are used to independently model and then weighted to evaluate the broadband network quality score. However, because the solutions adopted by these technologies are relatively singular, the accuracy of the broadband network complaint types and broadband network quality scores determined by these technologies is low. Summary of the Invention

[0004] This application discloses a method, apparatus, and electronic device for determining broadband network information, which can improve the accuracy of determining broadband network complaint types and broadband network quality scores.

[0005] To solve the above problems, this application adopts the following technical solution:

[0006] In a first aspect, embodiments of this application disclose a method for determining broadband network information, comprising: obtaining a first complaint text of the broadband network and a first performance indicator of the broadband network; and determining a broadband network complaint type and a broadband network quality score based on the first complaint text and the first performance indicator.

[0007] Secondly, embodiments of this application disclose a device for determining broadband network information, comprising: an acquisition module for acquiring a first complaint text of the broadband network and a first performance indicator of the broadband network; and a determination module for determining a broadband network complaint type and a broadband network quality score based on the first complaint text and the first performance indicator.

[0008] Thirdly, embodiments of this application provide an electronic device including a processor and a memory, wherein the memory stores programs or instructions executable on the processor, and the programs or instructions, when executed by the processor, implement the steps of the method described in the first aspect.

[0009] Fourthly, embodiments of this application provide a readable storage medium on which a program or instructions are stored, which, when executed by a processor, implement the steps of the method described in the first aspect.

[0010] Fifthly, embodiments of this application provide a computer program product, the computer program product including a computer program stored on a non-transitory computer-readable storage medium, the computer program including program instructions, which, when executed by a computer, cause the computer to perform the steps of the method described in the first aspect.

[0011] The technical solution adopted in this application can achieve the following beneficial effects:

[0012] This application provides a method for determining broadband network information. By acquiring a first complaint text and a first performance indicator of the broadband network, and based on these two indicators, the method determines the broadband network complaint type and broadband network quality score. The solution of this application comprehensively considers the correlation between the broadband network complaint text and the broadband network performance indicator, and determines the broadband network complaint type and broadband network quality score based on these indicators, thereby improving the accuracy of the determined broadband network complaint type and broadband network quality score. Attached Figure Description

[0013] Figure 1 This is a flowchart illustrating a method for determining broadband network information disclosed in an embodiment of this application;

[0014] Figure 2 This is a schematic diagram of the structure of a feature fusion module disclosed in an embodiment of this application;

[0015] Figure 3 This is a schematic diagram of the structure of a target model disclosed in an embodiment of this application;

[0016] Figure 4 This is a schematic diagram of the structure of a broadband network information determination device disclosed in an embodiment of this application;

[0017] Figure 5 This is a schematic diagram of the structure of an electronic device disclosed in an embodiment of this application. Detailed Implementation

[0018] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0019] The terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such terms can be used interchangeably where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class and the number of objects is not limited; for example, a first object can be one or more. Furthermore, in the specification and claims, "and / or" indicates at least one of the electrically connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0020] The following description, in conjunction with the accompanying drawings, details the broadband network information determination method, determination device, and electronic device disclosed in this application through specific embodiments and application scenarios.

[0021] This application discloses a method for determining broadband network information. Figure 1 This is a flowchart illustrating a method for determining broadband network information disclosed in an embodiment of this application. Figure 1 As shown, the method includes the following steps:

[0022] S120. Obtain the first complaint text of the broadband network and the first performance index of the broadband network.

[0023] In this application, the complaint text for the broadband network refers to the user's complaint text against the broadband network. Performance metrics for the broadband network may include dial-up success rate, TCP (Transmission Control Protocol) success rate, HTTP response success rate, TCP handshake latency, TCP retransmission rate, number of abnormal disconnections, number of frequent disconnections, and number of service interruptions.

[0024] For example, the user's number can be used as a unique identifier to obtain the broadband network's performance metrics from the operator. Furthermore, there can be a corresponding time relationship between the broadband network complaint text and the broadband network performance metrics obtained in this application. For instance, if the first broadband network complaint text is obtained at 10:00, the corresponding first performance metric for that broadband network between 9:30 and 10:00 can be obtained.

[0025] S140. Based on the first complaint text and the first performance indicator, determine the broadband network complaint type and broadband network quality score.

[0026] Since there is usually a certain correlation between the complaint text of broadband networks and the performance indicators of broadband networks, determining the complaint type and broadband network quality score based on the first complaint text and the first performance indicator of broadband networks can improve the accuracy of the determined complaint type and broadband network quality score.

[0027] For example, broadband network complaints may include, but are not limited to, complaints related to processing, inquiries, malfunctions, and network quality. Network quality complaints may include network-related complaints and service-related complaints. Network-related complaints may include planning and construction complaints, maintenance complaints, and optimization complaints.

[0028] This application provides a method for determining broadband network information. By acquiring a first complaint text and a first performance indicator of the broadband network, and based on these two indicators, the method determines the broadband network complaint type and broadband network quality score. The solution of this application comprehensively considers the correlation between the broadband network complaint text and the broadband network performance indicator, and determines the broadband network complaint type and broadband network quality score based on these indicators, thereby improving the accuracy of the determined broadband network complaint type and broadband network quality score.

[0029] In this application, determining the broadband network complaint type and broadband network quality score based on the first complaint text and the first performance indicator may include: inputting the first complaint text and the first performance indicator into a target model to obtain the broadband network complaint type and broadband network quality score output by the target model, wherein the target model is used to determine the broadband network complaint type and broadband network quality score based on the broadband network complaint text and performance indicator. In other words, this application can improve the efficiency of determining the broadband network complaint type and broadband network quality score by inputting the acquired first broadband network complaint text and first performance indicator into a trained target model and outputting the broadband network complaint type and broadband network quality score based on the target model.

[0030] In one implementation, the step of inputting the first complaint text and the first performance indicator into a target model to obtain the broadband network complaint type and broadband network quality score output by the target model may include: extracting first sparse features and first dense features of the first complaint text by inputting the first complaint text into the sparse feature extraction module and the dense feature extraction module of the target model, respectively; performing feature fusion by inputting the first sparse features, the first dense features, and the first performance indicator into the feature fusion module of the target model to obtain a first fused feature output by the feature fusion module corresponding to the first complaint text and the first performance indicator; and performing mapping by inputting the first fused feature into the first mapping module and the second mapping module of the target model, respectively, to obtain the broadband network complaint type output by the first mapping module and the broadband network quality score output by the second mapping module.

[0031] For example, the sparse feature extraction module in this application can be a skip-gram model. By inputting the first complaint text into the skip-gram model, the first sparse features of the first complaint text are extracted. The dense feature extraction module in this application can be an encoder structure using Transformers (such as a BERT pre-trained model). By inputting the first complaint text into a BERT (Bidirectional Encoder Representations from Transformers) pre-trained model, the BERT pre-trained model represents the first complaint text, learns the complex relationships of words in context, and generates richer and more expressive word vectors, thus obtaining the first dense features. The feature dimensions of the sparse and dense features in this application are generally n*dim_size, where n represents the text length and dim_size represents the token feature vector size (usually 768). The first mapping module in this application can be a Feed-forward network, and the second mapping module can also be a Feed-forward network. By inputting the first fused features into two Feed-forward networks respectively, the broadband network complaint type and broadband network quality score are obtained.

[0032] Adopting the solution of this application, by extracting the first sparse feature and the first dense feature of the first complaint text, fully considering the semantics and context information of the first complaint text, and then inputting the first sparse feature, the first dense feature and the first performance index into the feature fusion module of the target model, the first fusion feature corresponding to the first complaint text and the first performance index output by the feature fusion module is obtained. By respectively inputting the first fusion feature into the first mapping module and the second mapping module of the target model for mapping, the broadband network complaint type output by the first mapping module and the broadband network quality score output by the second mapping module are obtained, which can further improve the accuracy of the determined broadband network complaint type and broadband network quality score.

[0033] In one implementation, before extracting the first sparse feature and the first dense feature of the first complaint text by respectively inputting the first complaint text into the sparse feature extraction module and the dense feature extraction module of the target model, it further includes: preprocessing the first complaint text. Specifically, a word segmentation tool can be used in combination with a custom word segmentation dictionary to cut the first complaint text into individual words, and at the same time remove stop words, which are common words that frequently appear in the text but do not carry much information, such as "of", "is", "and", etc. After completing the preprocessing of the first complaint text by cutting and removing stop words from the first complaint text, the corresponding first sparse feature and first dense feature are extracted by respectively inputting the preprocessed first complaint text into the sparse feature extraction module and the dense feature extraction module of the target model.

[0034] It should be noted that a custom word segmentation dictionary can be constructed by collecting proprietary words related to broadband network complaints, such as broadband, optical modem, WIFI, router, etc.

[0035] In one implementation, the step of inputting the first sparse feature, the first dense feature, and the first performance indicator into the feature fusion module of the target model for feature fusion to obtain a first fused feature output by the feature fusion module corresponding to the first complaint text and the first performance indicator may include: inputting the first sparse feature, the first dense feature, and the first performance indicator into the feature encoding layer of the feature fusion module for unified data dimension processing to obtain a first encoded feature corresponding to the first sparse feature, a second encoded feature corresponding to the first dense feature, and a third encoded feature corresponding to the first performance indicator output by the feature encoding layer; inputting the first encoded feature, the second encoded feature, and the third encoded feature into the feature fusion layer of the feature fusion module for stacking processing to obtain a first stacked feature output by the feature fusion layer; and inputting the first stacked feature into the pooling layer of the feature fusion module for data dimension reduction processing to obtain a first fused feature output by the pooling layer corresponding to the first complaint text and the first performance indicator.

[0036] The structure of the feature fusion module is as follows: Figure 2 As shown, exemplarily, the feature encoding layer of the feature fusion module may include three feed-forward networks. By inputting the first sparse feature f1, the first dense feature f2, and the first performance index f3 into a feed-forward network respectively for unified data dimension calculation, the feature encoding layer outputs a first encoded feature corresponding to the first sparse feature, a second encoded feature corresponding to the first dense feature, and a third encoded feature corresponding to the first performance index. The first, second, and third encoded features can be high-dimensional features containing richer information, and the feature dimension can be 1*d. It should be noted that the feed-forward network also performs a linear transformation operation on the input first sparse feature, first dense feature, or first performance index, and introduces non-linear features through the tanh activation function.

[0037] After obtaining the first encoded feature corresponding to the first sparse feature, the second encoded feature corresponding to the first dense feature, and the third encoded feature corresponding to the first performance index from the feature encoding layer output of the feature fusion module, the first encoded feature, the second encoded feature, and the third encoded feature are input into the feature fusion layer stack of the feature fusion module for stacking processing to obtain the first stacked feature output by the feature fusion layer. The feature dimension of the first stacked feature can be 3*d.

[0038] After obtaining the first stacked features, the data dimensionality is reduced by inputting the first stacked features into the pooling layer of the feature fusion module, resulting in the first fused feature f_merge corresponding to the first complaint text and the first performance indicator. The feature dimension of the first fused feature can be 1*d. It should be noted that the pooling layer can be max pooling, mean pooling, or min pooling.

[0039] In this embodiment of the application, before obtaining the broadband network complaint type and broadband network quality score output by the target model by inputting the first complaint text and the first performance index into the target model, the method may further include: acquiring multiple training samples, wherein the training samples include a second complaint text of the broadband network, a second performance index of the broadband network, a real broadband network complaint type corresponding to the second complaint text and the second performance index, and a real broadband network quality score corresponding to the second complaint text and the second performance index; iteratively training the target model to be trained based on the multiple training samples until the loss function corresponding to the target model converges.

[0040] In one implementation, training the target model to be trained based on the training samples may include: extracting second sparse features and second dense features of the second complaint text by inputting the second complaint text into the sparse feature extraction module and the dense feature extraction module of the target model to be trained, respectively; performing feature fusion by inputting the second sparse features, the second dense features, and the second performance index into the feature fusion module of the target model to be trained, thereby obtaining a second fused feature output by the feature fusion module corresponding to the second complaint text and the second performance index; performing mapping by inputting the second fused feature into the first mapping module and the second mapping module of the target model to be trained, respectively, to obtain the predicted broadband network complaint type output by the first mapping module and the predicted broadband network quality score output by the second mapping module; and training the target model to be trained based on the real broadband network complaint type, the real broadband network quality score, the predicted broadband network complaint type, and the predicted broadband network quality score.

[0041] The step of inputting the second sparse feature, the second dense feature, and the second performance index into the feature fusion module of the target model to be trained for feature fusion to obtain the second fused feature output by the feature fusion module corresponding to the second complaint text and the second performance index may include: inputting the second sparse feature, the second dense feature, and the second performance index into the feature encoding layer of the feature fusion module for unified data dimension processing to obtain a fourth encoded feature corresponding to the second sparse feature, a fifth encoded feature corresponding to the second dense feature, and a sixth encoded feature corresponding to the second performance index output by the feature encoding layer; inputting the fourth encoded feature, the fifth encoded feature, and the sixth encoded feature into the feature fusion layer of the feature fusion module for stacking processing to obtain a second stacked feature output by the feature fusion layer; and inputting the second stacked feature into the output layer of the feature fusion module for data dimension reduction processing to obtain the second fused feature output by the output layer corresponding to the second complaint text and the second performance index.

[0042] In this application, the structure of the target model is as follows: Figure 3 As shown, the loss function corresponding to the target model Where i is a sample, c is the complaint type, N is the sample size, and M is the number of categories. If the actual complaint type of sample i is equal to c, then y ic Select 1, otherwise y ic Take 0, p ic Let y' represent the predicted probability that sample i belongs to complaint type c. i The predicted broadband network quality score output by the second mapping module, y i This corresponds to a real broadband network score.

[0043] It should be noted that L1 can be calculated through the first mapping module and L2 can be calculated through the second mapping module.

[0044] The proposed solution inputs complaint text and performance metrics into a target model, yielding the broadband network complaint type and broadband network quality score. This allows for the modeling of both complaint content classification and network quality scoring within a single end-to-end model, reducing interference from other factors and fully utilizing statistical information and patterns in the data to provide more accurate and stable predictions. This helps operators resolve network quality issues and improve user experience. Furthermore, this solution determines the broadband network complaint type and broadband network quality score through a single model, eliminating the need to separately calculate weights for different network problem quality scores. This avoids interference from other factors and the accumulation of losses, improving overall generalization and accuracy. Moreover, this solution utilizes a BERT pre-trained model to represent the complaint text and combines it with broadband network performance metrics for joint modeling. This allows for the learning of richer and more continuous text representations, capturing relationships between words and contextual information, providing better text representation capabilities, and thus addressing the problem that traditional topic models cannot handle complex Chinese semantics.

[0045] It should be noted that the dense features in this application are also called dense features.

[0046] The broadband network information determination method provided in this application can be executed by a broadband network information determination device. This application uses the broadband network information determination device executing the broadband network information determination method as an example to illustrate the device for determining broadband network information provided in this application.

[0047] Figure 4 This is a schematic diagram of the structure of a broadband network information determination device disclosed in an embodiment of this application. Figure 4 As shown, the broadband network information determination device 400 includes: an acquisition module 410 and a determination module 420.

[0048] In this application, the acquisition module 410 is used to acquire a first complaint text of the broadband network and a first performance indicator of the broadband network; the determination module 420 is used to determine the broadband network complaint type and the broadband network quality score based on the first complaint text and the first performance indicator.

[0049] In one implementation, the determining module 420 determines the broadband network complaint type and broadband network quality score based on the first complaint text and the first performance indicator, including: inputting the first complaint text and the first performance indicator into a target model to obtain the broadband network complaint type and broadband network quality score output by the target model, wherein the target model is used to determine the broadband network complaint type and broadband network quality score according to the broadband network complaint text and performance indicator.

[0050] In one implementation, the determining module 420 obtains the broadband network complaint type and broadband network quality score output by the target model by inputting the first complaint text and the first performance index into the target model, including: extracting first sparse features and first dense features of the first complaint text by inputting the first complaint text into the sparse feature extraction module and the dense feature extraction module of the target model respectively; performing feature fusion by inputting the first sparse features, the first dense features and the first performance index into the feature fusion module of the target model to obtain the first fused feature output by the feature fusion module corresponding to the first complaint text and the first performance index; and performing mapping by inputting the first fused feature into the first mapping module and the second mapping module of the target model respectively to obtain the broadband network complaint type output by the first mapping module and the broadband network quality score output by the second mapping module.

[0051] In one implementation, the determining module 420 performs feature fusion by inputting the first sparse feature, the first dense feature, and the first performance indicator into the feature fusion module of the target model to obtain a first fused feature output by the feature fusion module corresponding to the first complaint text and the first performance indicator. This includes: inputting the first sparse feature, the first dense feature, and the first performance indicator into the feature encoding layer of the feature fusion module for unified data dimension processing to obtain a first encoded feature corresponding to the first sparse feature, a second encoded feature corresponding to the first dense feature, and a third encoded feature corresponding to the first performance indicator output by the feature encoding layer; inputting the first encoded feature, the second encoded feature, and the third encoded feature into the feature fusion layer of the feature fusion module for stacking processing to obtain a first stacked feature output by the feature fusion layer; and inputting the first stacked feature into the pooling layer of the feature fusion module for data dimension reduction processing to obtain a first fused feature output by the pooling layer corresponding to the first complaint text and the first performance indicator.

[0052] In one implementation, the above apparatus further includes: the acquisition module 410, which is further configured to acquire multiple training samples before obtaining the broadband network complaint type and broadband network quality score output by the target model by inputting the first complaint text and the first performance index into the target model, wherein the training samples include a second complaint text of the broadband network, a second performance index of the broadband network, a real broadband network complaint type corresponding to the second complaint text and the second performance index, and a real broadband network quality score corresponding to the second complaint text and the second performance index; and a training module, which is configured to iteratively train the target model to be trained based on the multiple training samples until the loss function corresponding to the target model converges.

[0053] In one implementation, the training module trains the target model to be trained based on the training samples, including: extracting second sparse features and second dense features of the second complaint text by inputting the second complaint text into the sparse feature extraction module and the dense feature extraction module of the target model to be trained, respectively; performing feature fusion by inputting the second sparse features, the second dense features, and the second performance index into the feature fusion module of the target model to be trained, to obtain a second fused feature output by the feature fusion module corresponding to the second complaint text and the second performance index; performing mapping by inputting the second fused feature into the first mapping module and the second mapping module of the target model to be trained, respectively, to obtain the predicted broadband network complaint type output by the first mapping module and the predicted broadband network quality score output by the second mapping module; and training the target model to be trained based on the real broadband network complaint type, the real broadband network quality score, the predicted broadband network complaint type, and the predicted broadband network quality score.

[0054] The broadband network information determination device provided in this application embodiment can realize the various processes implemented in the broadband network information determination method embodiment, and will not be described again here to avoid repetition.

[0055] Optionally, such as Figure 5 As shown, this application embodiment also provides an electronic device 500, including a processor 501 and a memory 502. The memory 502 stores a program or instructions that can run on the processor 501. When the program or instructions are executed by the processor 501, they implement the various steps of the above-described method embodiment for determining broadband network information and can achieve the same technical effect. To avoid repetition, they will not be described again here.

[0056] It should be noted that the electronic devices in the embodiments of this application include mobile electronic devices and non-mobile electronic devices.

[0057] This application also provides a readable storage medium storing a program or instructions. When the program or instructions are executed by a processor, they implement the various processes of the above-described method for determining broadband network information and achieve the same technical effect. To avoid repetition, they will not be described again here.

[0058] The processor is the processor in the electronic device described in the above embodiments. The readable storage medium includes computer-readable storage media, such as computer read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk.

[0059] This application also provides a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium. The computer program includes program instructions that, when executed by a computer, cause the computer to perform the steps of the broadband network information determination method described above.

[0060] The above embodiments of this application focus on describing the differences between the various embodiments. As long as the different optimization features between the various embodiments are not contradictory, they can be combined to form a better embodiment. For the sake of brevity, they will not be described in detail here.

[0061] The above description is merely an embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of this application should be included within the scope of the claims of this application.

Claims

1. A method for determining broadband network information, characterized in that, include: Obtain the first complaint text of the broadband network and the first performance index of the broadband network; Based on the first complaint text and the first performance indicator, determine the broadband network complaint type and broadband network quality score. The process of determining the broadband network complaint type and broadband network quality score based on the first complaint text and the first performance indicator includes: By inputting the first complaint text and the first performance indicator into the target model, the broadband network complaint type and broadband network quality score output by the target model are obtained, wherein the target model is used to determine the broadband network complaint type and broadband network quality score based on the broadband network complaint text and performance indicator; The process of inputting the first complaint text and the first performance indicator into the target model to obtain the broadband network complaint type and broadband network quality score output by the target model includes: By inputting the first complaint text into the sparse feature extraction module and the dense feature extraction module of the target model respectively, the first sparse feature and the first dense feature of the first complaint text are extracted. By inputting the first sparse feature, the first dense feature and the first performance index into the feature fusion module of the target model for feature fusion, the first fused feature output by the feature fusion module corresponding to the first complaint text and the first performance index is obtained. By inputting the first fusion feature into the first mapping module and the second mapping module of the target model respectively for mapping, the broadband network complaint type output by the first mapping module and the broadband network quality score output by the second mapping module are obtained.

2. The determination method according to claim 1, characterized in that, The step involves inputting the first sparse feature, the first dense feature, and the first performance index into the feature fusion module of the target model for feature fusion to obtain the first fused feature output by the feature fusion module corresponding to the first complaint text and the first performance index, including: By inputting the first sparse feature, the first dense feature, and the first performance index into the feature encoding layer of the feature fusion module for unified data dimension processing, the first encoded feature corresponding to the first sparse feature, the second encoded feature corresponding to the first dense feature, and the third encoded feature corresponding to the first performance index output by the feature encoding layer are obtained. By inputting the first encoded feature, the second encoded feature, and the third encoded feature into the feature fusion layer of the feature fusion module and stacking them, a first stacked feature output by the feature fusion layer is obtained. By inputting the first stacked feature into the pooling layer of the feature fusion module to reduce the data dimensionality, the first fused feature output by the pooling layer corresponding to the first complaint text and the first performance indicator is obtained.

3. The determination method according to claim 1, characterized in that, Before the step of inputting the first complaint text and the first performance indicator into the target model to obtain the broadband network complaint type and broadband network quality score output by the target model, the method further includes: Multiple training samples are obtained, wherein the training samples include a second complaint text of the broadband network, a second performance index of the broadband network, a real broadband network complaint type corresponding to the second complaint text and the second performance index, and a real broadband network quality score corresponding to the second complaint text and the second performance index. The target model to be trained is iteratively trained based on multiple training samples until the loss function corresponding to the target model converges.

4. The determination method according to claim 3, characterized in that, Training the target model to be trained based on the training samples includes: By inputting the second complaint text into the sparse feature extraction module and the dense feature extraction module of the target model to be trained, the second sparse feature and the second dense feature of the second complaint text are extracted. By inputting the second sparse feature, the second dense feature and the second performance index into the feature fusion module of the target model to be trained, feature fusion is performed to obtain the second fused feature output by the feature fusion module, which corresponds to the second complaint text and the second performance index. By mapping the second fused features into the first mapping module and the second mapping module of the target model to be trained, the predicted broadband network complaint type output by the first mapping module and the predicted broadband network quality score output by the second mapping module are obtained. The target model to be trained is trained based on the actual broadband network complaint types, the actual broadband network quality scores, the predicted broadband network complaint types, and the predicted broadband network quality scores.

5. A device for determining broadband network information, characterized in that, include: The acquisition module is used to acquire the first complaint text of the broadband network and the first performance index of the broadband network; The determination module is used to determine the broadband network complaint type and broadband network quality score based on the first complaint text and the first performance indicator; The determining module determines the broadband network complaint type and broadband network quality score based on the first complaint text and the first performance indicator, including: inputting the first complaint text and the first performance indicator into the target model to obtain the broadband network complaint type and broadband network quality score output by the target model, wherein the target model is used to determine the broadband network complaint type and broadband network quality score according to the broadband network complaint text and performance indicator; The determining module obtains the broadband network complaint type and broadband network quality score output by the target model by inputting the first complaint text and the first performance index into the target model, including: extracting the first sparse feature and the first dense feature of the first complaint text by inputting the first complaint text into the sparse feature extraction module and the dense feature extraction module of the target model respectively; performing feature fusion by inputting the first sparse feature, the first dense feature and the first performance index into the feature fusion module of the target model to obtain the first fused feature output by the feature fusion module corresponding to the first complaint text and the first performance index; and performing mapping by inputting the first fused feature into the first mapping module and the second mapping module of the target model respectively to obtain the broadband network complaint type output by the first mapping module and the broadband network quality score output by the second mapping module.

6. An electronic device, characterized in that, It includes a processor and a memory, the memory storing a program or instructions that can run on the processor, the program or instructions being executed by the processor to implement the steps of the method for determining broadband network information as described in any one of claims 1-4.

7. A readable storage medium, characterized in that, The readable storage medium stores a program or instructions that, when executed by a processor, implement the steps of the method for determining broadband network information as described in any one of claims 1-4.

8. A computer program product, characterized in that, The computer program product includes a computer program stored on a non-transitory computer-readable storage medium, the computer program including program instructions that, when executed by a computer, cause the computer to perform the steps of the method for determining broadband network information as described in any one of claims 1-4.

Images