Intelligent business analysis method, system and electronic device based on big data

Abstract

The present application relates to the field of intelligent analysis of business, and specifically discloses a method, system and electronic device for intelligent business analysis based on big data, which first obtains the text data information and voice emotion data information of the customer to be analyzed from the customer service call voice data, and then uses the customer text information feature extraction module based on the context encoder to extract and analyze the text data information of the customer to be analyzed, and at the same time, extracts and analyzes the voice emotion data information of the customer to be analyzed through the voice noise reduction module based on the automatic encoder and the voice emotion feature extractor based on the convolutional neural network model, and then obtains the classification result for indicating whether the customer to be analyzed is satisfied with the service based on the feature information of the two. In this way, the enterprise can timely adjust the service mode and attitude of the customer service center to improve customer satisfaction, thereby improving the efficiency of business marketing.

Description

Technical Field

[0001] The present application relates to the field of intelligent business analysis, and more specifically, to an intelligent business analysis method, system and electronic device based on big data. Background Art

[0002] In modern enterprises, customer service centers bear the important responsibility of maintaining corporate customer relationships and business marketing. It is of great significance to monitor the quality of customer service center calls. Emotional analysis of call speech can identify the emotional state of customer service and customers during customer service calls, thereby effectively tracking and monitoring the quality of service. Existing customer sentiment analysis methods mostly convert speech into text content through speech recognition technology, and then perform sentiment analysis based on the text. In this way, the intensity, intonation, speed and other characteristics of the customer's voice that effectively reflect their emotions are often ignored, thus affecting the accuracy of speech emotion recognition.

[0003] Therefore, an intelligent business analysis method, system and electronic device based on big data are provided to analyze the customer's emotions based on the text data information and voice emotion data information in the voice call between the customer service and the customer, so as to judge the customer's satisfaction with the service. Summary of the invention

[0004] In order to solve the above technical problems, the present application is proposed. The embodiment of the present application provides an intelligent business analysis method, system and electronic device based on big data, which first obtains the text data information of the customer to be analyzed and the voice emotion data information of the customer to be analyzed from the customer service call voice data, and then uses the customer text information feature extraction module based on the context encoder to extract and analyze the text data information of the customer to be analyzed, and at the same time, the voice emotion data information of the customer to be analyzed is extracted and analyzed by the voice noise reduction module based on the automatic encoder and the voice emotion feature extractor based on the convolutional neural network model, and then based on the feature information of the two, a classification result is obtained to indicate whether the customer to be analyzed is satisfied with the service. In this way, the enterprise can adjust the service mode and attitude of the customer service center in a timely manner to improve customer satisfaction, thereby improving the efficiency of business marketing.

[0005] According to a first aspect of the present application, there is provided a method for intelligent business analysis based on big data, which includes:

[0006] Obtaining text data information of the customer to be analyzed and voice emotion data information of the customer to be analyzed from the customer service call voice data;

[0007] Obtaining a customer satisfaction analysis feature vector based on the text data information of the customer to be analyzed and the voice emotion data information of the customer to be analyzed;

[0008] The customer satisfaction analysis feature vector is analyzed to obtain a classification result.

[0009] According to a second aspect of the present application, there is provided an intelligent business analysis system based on big data, comprising:

[0010] The voice information collection module is used to obtain the text data information of the customer to be analyzed and the voice emotion data information of the customer to be analyzed from the customer service call voice data;

[0011] A voice information feature analysis module, used to obtain a customer satisfaction analysis feature vector based on the text data information of the customer to be analyzed and the voice emotion data information of the customer to be analyzed;

[0012] The user satisfaction feature classification module is used to analyze the customer satisfaction analysis feature vector to obtain a classification result.

[0013] According to a third aspect of the present application, an electronic device is provided, comprising: a processor; and a memory, in which computer program instructions are stored, and when the computer program instructions are executed by the processor, the processor executes the intelligent business analysis method based on big data as described above.

[0014] According to a fourth aspect of the present application, a computer-readable medium is provided, on which computer program instructions are stored. When the computer program instructions are executed by a processor, the processor executes the intelligent business analysis method based on big data as described above.

[0015] In combination with the first aspect of the present application, in an intelligent business analysis method based on big data of the first aspect of the present application, a customer satisfaction analysis feature vector is obtained based on the text data information of the customer to be analyzed and the voice emotion data information of the customer to be analyzed, including: performing semantic understanding analysis on the text data information of the customer to be analyzed through a customer text information feature extraction module to obtain a customer text semantic understanding feature vector; performing preprocessing operations on the voice emotion data information of the customer to be analyzed to obtain a denoised voice emotion waveform; passing the denoised voice emotion waveform through a voice emotion feature extractor to obtain a customer voice emotion feature vector; and fusing the customer text semantic understanding feature vector and the customer voice emotion feature vector based on the smoothed posterior expression of the output relative objective function to obtain a customer satisfaction analysis feature vector.

[0016] In combination with the second aspect of the present application, in a big data-based intelligent business analysis system of the second aspect of the present application, the user satisfaction feature classification module is used to: perform feature classification on the customer satisfaction analysis feature vector through a classifier to obtain a classification result indicating whether the customer to be analyzed is satisfied with the service.

[0017] Compared with the prior art, the present application provides a method, system and electronic device for intelligent business analysis based on big data, which first obtains the text data information and voice emotion data information of the customer to be analyzed from the customer service call voice data, and then uses the customer text information feature extraction module based on the context encoder to extract and analyze the text data information of the customer to be analyzed, and at the same time, the voice emotion data information of the customer to be analyzed is extracted and analyzed by the voice noise reduction module based on the automatic encoder and the voice emotion feature extractor based on the convolutional neural network model, and then the classification result indicating whether the customer to be analyzed is satisfied with the service is obtained based on the feature information of the two. In this way, the enterprise can adjust the service mode and attitude of the customer service center in a timely manner to improve customer satisfaction, thereby improving the efficiency of business marketing. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0019] Figure 1 The figure illustrates a flow chart of an intelligent business analysis method based on big data according to an embodiment of the present application.

[0020] Figure 2 The figure illustrates a schematic block diagram of an intelligent business analysis system based on big data according to an embodiment of the present application.

[0021] Figure 3 The diagram illustrates an architecture diagram of an intelligent business analysis system based on big data according to an embodiment of the present application.

[0022] Figure 4 A block diagram of an electronic device according to an embodiment of the present application is illustrated.

[0023] Figure 5 A block diagram of a non-transitory computer-readable storage medium according to an embodiment of the present application is illustrated. DETAILED DESCRIPTION

[0024] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0025] Exemplary Methods

[0026] Figure 1 The figure shows a flow chart of an intelligent business analysis method based on big data according to an embodiment of the present application. Figure 1 As shown, according to the big data-based intelligent business analysis method of the embodiment of the present application, the method includes: S110, obtaining text data information of the customer to be analyzed and voice emotion data information of the customer to be analyzed from the customer service call voice data; S120, obtaining a customer satisfaction analysis feature vector based on the text data information of the customer to be analyzed and the voice emotion data information of the customer to be analyzed; S130, analyzing the customer satisfaction analysis feature vector to obtain a classification result.

[0027] With the development of social economy and the improvement of living standards, people know how to enjoy life better, which makes people's requirements for service comfort and other aspects of the service industry have increased. For the service industry, customer satisfaction is one of the important factors to promote the development of enterprises. It should be understood that understanding customer satisfaction with customer service can help enterprises understand customer needs and expectations, thereby improving customer experience. By collecting and analyzing customer feedback, enterprises can find problems and take corresponding measures to improve customer satisfaction. In addition, satisfied customers are more likely to become loyal customers. If customers are satisfied with the service of customer service, they are more likely to choose to continue to buy and use the company's products or services, and may recommend them to others. In other words, improving customer satisfaction can increase customer loyalty and promote business growth. Therefore, more and more companies use various customer sentiment analysis methods to understand customer satisfaction with services.

[0028] However, as described in the above background technology, most existing customer sentiment analysis methods convert speech into text content through speech recognition technology, and then perform sentiment analysis based on the text. In this way, the intensity, intonation, speed and other changing characteristics of the customer's speech that effectively reflect their emotions are often ignored, thereby affecting the accuracy of speech emotion recognition. Therefore, an intelligent business analysis method, system and electronic device based on big data are provided to analyze the customer's emotions based on the text data information and speech emotion data information in the voice of the customer service and the customer, so as to judge the customer's satisfaction with the service.

[0029] Those skilled in the art should understand that natural language processing (NLP) is a field that studies the interaction between human language and computers, aiming to enable computers to understand, process and generate natural language text. NLP combines the theories and methods of multiple disciplines such as computer science, artificial intelligence and linguistics, and involves tasks such as text analysis, semantic understanding, language generation, and machine translation. In the task of semantic understanding, NLP technology can help computers understand the meaning and structure of natural language text. This includes lexical analysis (segmentation, part-of-speech tagging), syntactic analysis (sentence structure analysis), semantic analysis (semantic role tagging, named entity recognition), semantic role tagging, etc. Therefore, in the embodiment of the present application, the text data information in the voice of customer service and customer calls is understood and analyzed by natural language processing technology. In addition, considering that deep learning is widely used in speech recognition, it has become the mainstream method in the field of speech recognition. Deep learning can learn higher-level speech feature representations from raw audio data by constructing a convolutional neural network model, thereby achieving more accurate and robust speech recognition. Based on this, the convolutional neural network model in deep learning is used to identify and analyze the voice emotion data information in the voice calls between customer service and customers.

[0030] In step S110, the text data information of the customer to be analyzed and the voice emotion data information of the customer to be analyzed are obtained from the customer service call voice data. It should be understood that many companies have their own customer service centers, and customers can understand and communicate about the business by talking to the company's customer service. That is, the customer service center bears the important responsibility of maintaining the company's customer relationship and business marketing. Therefore, it is of great significance to monitor the service quality of the customer service center call. Further, considering that the text can be semantically understood and content analyzed by transcribing the customer service call voice data into text, the text data information in the customer service call voice data can be collected based on this. In this way, the customer's problems, needs and feedback can be better understood, so as to provide more accurate and personalized responses and solutions. In addition, by obtaining the customer's voice emotion data, the customer service center can provide personalized services and emotion-driven marketing according to the customer's emotional state. For example, for customers with negative emotions, a more gentle and soothing tone can be adopted to respond to improve customer satisfaction and emotional connection. Therefore, the voice emotion data information in the customer service call voice data can also be collected.

[0031] Furthermore, when obtaining the text data information of the customer to be analyzed, the customer service call voice data can be transcribed into text data with the help of speech-to-text technology, for example, using automatic speech recognition (ASR) technology. When collecting the voice emotion data information of the customer to be analyzed, the acoustic feature extraction technology is used to extract emotion-related features from the voice signal of the customer to be analyzed. These features may include pitch, speech speed, audio energy, sound intensity, etc. In this way, the customer's emotions and attitudes can be more comprehensively identified, thereby improving the accuracy of customer satisfaction judgment.

[0032] In step S120, a customer satisfaction analysis feature vector is obtained based on the text data information of the customer to be analyzed and the voice emotion data information of the customer to be analyzed. It should be understood that the text data information can provide the customer's specific problems, feedback and needs, while the voice emotion data information can reveal the customer's emotional state during the call. Combining these two types of information can provide a more comprehensive understanding of the customer's needs and emotional reactions, thereby better analyzing customer satisfaction. Further, considering that in the embodiment of the present application, the text data information of the customer to be analyzed is semantically understood by natural language processing technology, and the voice emotion data information of the customer to be analyzed is feature recognized by speech recognition technology in deep learning. Therefore, it is necessary to convert the text data information of the customer to be analyzed and the voice emotion data information of the customer to be analyzed into a form such as a vector or matrix that can be recognized by the machine.

[0033] Specifically, step S120, based on the text data information of the customer to be analyzed and the voice emotion data information of the customer to be analyzed to obtain the customer satisfaction analysis feature vector, includes: S121, performing semantic understanding analysis on the text data information of the customer to be analyzed through a customer text information feature extraction module to obtain a customer text semantic understanding feature vector; S122, performing preprocessing operations on the voice emotion data information of the customer to be analyzed to obtain a denoised voice emotion waveform; S123, passing the denoised voice emotion waveform through a voice emotion feature extractor to obtain a customer voice emotion feature vector; and S124, fusing the customer text semantic understanding feature vector and the customer voice emotion feature vector based on the smoothed posterior expression of the output relative objective function to obtain a customer satisfaction analysis feature vector.

[0034] In step S121, the text data information of the customer to be analyzed is subjected to semantic understanding analysis by the customer text information feature extraction module to obtain a customer text semantic understanding feature vector. Considering that the customer text information is text data, there is contextual information in these data, and there may be some abstract expressions in these text information. Considering that the convolutional neural network model mainly extracts local features through convolution layers and pooling layers, the modeling ability of the overall context is relatively weak, and it is difficult to capture these abstract information and contextual information. Based on this, the contextual information and abstract information in the text data information of the customer to be analyzed are feature extracted and analyzed by the customer text information feature extraction module based on the context encoder. Among them, the context encoder can perform context modeling on the customer text information, that is, consider the before and after information in the text. This helps to understand the semantics and context in the text and extract more accurate feature representations. In addition, the model based on the context encoder can learn richer and more abstract semantic expressions. Through the multi-layer neural network structure, the model can gradually extract higher-level semantic features, so as to better understand the customer text information. In this way, it can help understand the customer's needs, emotions, intentions and other information, so as to better perform business analysis and decision support.

[0035] Specifically, step S121, performs semantic understanding analysis on the text data information of the customer to be analyzed through the customer text information feature extraction module to obtain a customer text semantic understanding feature vector, which is used to: perform word segmentation processing on the text data information of the customer to be analyzed to obtain a word sequence; use the embedding layer of the sequence encoder of the Clip model to map each word in the word sequence to a word embedding vector to obtain a sequence of word embedding vectors; use the converter-based Bert model of the sequence encoder of the Clip model to perform global context semantic encoding on the sequence of embedding vectors to obtain multiple feature vectors; and cascade the multiple feature vectors to obtain the customer text semantic understanding feature vector.

[0036] In step S122, the voice emotion data information of the customer to be analyzed is preprocessed to obtain a voice emotion waveform after noise reduction. It should be understood that obtaining the voice signal in the voice emotion data information of the customer to be analyzed to generate the customer voice emotion waveform is an important step in the intelligent business analysis method based on big data. In this way, emotion analysis can be performed based on voice signals, non-invasive data collection can be achieved, a large-scale data set can be constructed, and the needs of real-time emotion analysis can be met, thereby providing more comprehensive and accurate emotion information for intelligent business analysis. Therefore, before feature extraction is performed on the voice emotion data information of the customer to be analyzed, the voice signal must first be obtained. In addition, considering that the emotion information in the voice emotion data information of the customer to be analyzed is usually reflected in the tone, volume, spectrum, etc. of the sound. However, the voice emotion data information of the customer to be analyzed is usually interfered by environmental noise, such as background noise, microphone noise, etc. The presence of noise can cover up or change these emotion-related features, making emotion recognition difficult. Therefore, it is necessary to perform noise reduction processing on it. In this way, the influence of noise can be reduced, making the voice emotion features more prominent and recognizable, which helps to more accurately analyze the emotional state of the customer.

[0037] Specifically, step S122, preprocessing the voice emotion data information of the customer to be analyzed to obtain a voice emotion waveform after noise reduction, includes: S122-1, obtaining the voice signal of the voice emotion data information of the customer to be analyzed to obtain the customer voice emotion waveform; and S122-2, performing noise reduction processing on the customer voice emotion waveform through a voice noise reduction module to obtain the voice emotion waveform after noise reduction. Wherein, the voice noise reduction module is a voice noise reduction module based on an automatic encoder.

[0038] Among them, step S122-2, performs noise reduction processing on the customer's voice emotion waveform through a voice noise reduction module to obtain the voice emotion waveform after noise reduction, which is used to: use the feature encoder of the voice noise reduction module to extract the customer's voice emotion features from the customer's voice emotion waveform; and use the feature decoder of the voice noise reduction module to decode the customer's voice emotion features respectively to obtain the voice emotion waveform after noise reduction.

[0039] In step S123, the denoised speech emotion waveform is passed through a speech emotion feature extractor to obtain a customer speech emotion feature vector. It should be understood that the denoised speech emotion waveform contains rich emotional information, such as tone, speaking speed, volume, etc. However, directly using the original waveform for analysis and modeling may lead to problems of high dimensionality and high redundancy. In addition, in a big data scenario, the speech emotion data may be very large, and feature vectors with too high dimensions will increase the cost of calculation and storage. Therefore, a speech emotion feature extractor is further used to extract speech emotion features and convert high-dimensional waveform data into more compact and more informative low-dimensional feature vectors. Among them, the speech emotion feature extractor is a speech emotion feature extractor based on a convolutional neural network model. In this way, the customer's voice emotions can be better represented, while reducing data redundancy and improving computing efficiency.

[0040] Specifically, step S123, passes the denoised speech emotion waveform through a speech emotion feature extractor to obtain a customer speech emotion feature vector, including: S123-1, extracting features from the denoised speech emotion waveform through a speech emotion feature extractor based on a convolutional neural network model to obtain a customer speech emotion feature map; and S123-2, performing global mean pooling processing on each feature matrix of the customer speech emotion feature map to obtain the customer speech emotion feature vector.

[0041] Among them, step S123-1, through the speech emotion feature extractor based on the convolutional neural network model, the feature extraction of the speech emotion waveform after noise reduction is performed to obtain the customer speech emotion feature map, which is used to: use the layers of the convolutional neural network model to perform the following on the input data in the forward transfer of the layer: convolution processing on the input data to obtain a convolution feature map; pooling processing on the convolution feature map based on the local feature matrix to obtain a pooled feature map; nonlinear activation on the pooled feature map to obtain an activation feature map; wherein the input of the first layer of the convolutional neural network model is the speech emotion waveform after noise reduction, and the output of the last layer of the convolutional neural network model is the customer speech emotion feature map.

[0042] In step S124, the customer text semantic understanding feature vector and the customer voice emotion feature vector are fused based on the smoothed posterior expression of the output relative objective function to obtain a customer satisfaction analysis feature vector. It should be understood that the customer's text information and voice emotion information are complementary to a certain extent. The customer's text information can provide a clearer and more specific expression, while the voice emotion information can provide a more intuitive and real emotional experience. By fusing these two types of information, their complementarity can be fully utilized to improve the accuracy and robustness of satisfaction analysis. Therefore, in order to more accurately obtain the customer's emotional information characteristics, the customer's text semantic understanding features and the customer's voice emotion features are subjected to feature fusion operations. In this way, by fusing the customer's text semantic understanding feature vector and the customer's voice emotion feature vector, the customer's expression in different modes can be comprehensively considered to provide a more comprehensive and accurate satisfaction analysis.

[0043] In particular, in the technical solution of the present application, considering that the customer text semantic understanding feature vector and the customer voice emotion feature vector come from different data sources and representation methods, and have different characteristics and expressions, there is a large difference between the high-dimensional feature distribution of the customer text semantic understanding feature vector and the customer voice emotion feature vector, which is not suitable for direct fusion. First, the customer text semantic understanding feature vector is extracted from the customer's text data information, usually based on natural language processing technology, such as word embedding, sentence embedding, etc. Text data is represented in the form of discrete symbols, each symbol represents a word or phrase, and the feature vector is encoded based on these symbols. Due to the characteristics of text data, the dimension of the text feature vector is usually high, and there may be large differences between different texts. For example, different word combinations, sentence structures, grammatical rules and other factors will affect the text feature vector, resulting in differences in feature distribution between different texts. On the other hand, the customer voice emotion feature vector is extracted from the customer's voice emotion data information, usually based on speech signal processing and pattern recognition technology. Voice emotion data is represented in the form of a continuous waveform, which contains the customer's voice and voice features. The extraction of speech emotion feature vectors usually includes spectral analysis of sound, measurement of sound intensity, extraction of fundamental frequency period, etc. Due to the continuity and temporal nature of speech data, the dimension of speech feature vectors is usually low, and there may be small differences between different speech emotions. Therefore, due to the large differences in data source, representation method and feature distribution between the customer text semantic understanding feature vector and the customer voice emotion feature vector, directly fusing these two feature vectors may cause confusion and distortion of information. Directly fusing high-dimensional text feature vectors with low-dimensional speech feature vectors may cause imbalance in feature space and information loss, affecting the final customer satisfaction analysis results. Therefore, the customer text semantic understanding feature vector and the customer voice emotion feature vector are fused based on the smoothed posterior expression of the output relative objective function.

[0044] Specifically, step S124, the customer text semantic understanding feature vector and the customer voice emotion feature vector are fused based on the smoothed posterior expression of the output relative objective function to obtain a customer satisfaction analysis feature vector, including: calculating the smoothed parameterized expression of the customer text semantic understanding feature vector relative to the target classification distribution to obtain an optimized customer text semantic understanding feature vector; calculating the smoothed parameterized expression of the customer voice emotion feature vector relative to the target classification distribution to obtain an optimized customer voice emotion feature vector; fusing the optimized customer text semantic understanding feature vector and the optimized customer voice emotion feature vector to obtain a customer satisfaction analysis feature vector.

[0045] The method comprises: calculating a smooth parameterized expression of the customer text semantic understanding feature vector relative to the target classification distribution to obtain an optimized customer text semantic understanding feature vector; calculating a smooth parameterized expression of the customer voice emotion feature vector relative to the target classification distribution to obtain an optimized customer voice emotion feature vector, including:

[0046]

[0047]

[0048] Wherein, V1 represents the customer text semantic understanding feature vector, V2 represents the customer voice emotion feature vector, represents the mean vector of the customer text semantic understanding feature vector and the customer voice emotion feature vector, ⊕ represents positional subtraction and addition of vectors, λ represents a predetermined hyperparameter, V′1 represents the optimized customer text semantic understanding feature vector, and V′2 represents the optimized customer voice emotion feature vector.

[0049] In view of the above technical problems, the present application proposes a feature vector fusion method based on smooth parameterized expression, which can use the prior and posterior distribution of the target classification distribution to probabilistically map and align the feature values ​​of the customer text semantic understanding feature vector and the customer voice emotion feature vector, thereby providing a fusion effect between the customer text semantic understanding feature vector and the customer voice emotion feature vector. Smooth parameterized expression is a feature transformation method based on parameterized surfaces, which can maintain the original information and semantics of the feature and avoid the information loss and confusion of the feature. Specifically, the method uses the mean vector of the customer text semantic understanding feature vector and the customer voice emotion feature vector as a pseudo target distribution to measure the information correlation between the two, thereby determining the approximate method of the point correlation between the two, which is used to distinguish and classify the information between the two, so as to determine the perceptual discrimination of the position feature matching of the two by measuring the information consistency between the feature values ​​of the two and the target classification distribution, which is used to enhance and supplement the information of the two, and then use the perceptual discrimination of the position feature matching to measure the information correlation between the feature values ​​of the two and the target classification distribution, thereby determining the fusion effect of the two, which is used to optimize and integrate the information of the two.

[0050] In step S130, the customer satisfaction analysis feature vector is analyzed to obtain a classification result. It should be understood that by analyzing the customer satisfaction analysis feature vector, the customer's satisfaction with the service or product can be understood. This is very important for enterprises, and enterprises can make corresponding decisions based on the analysis results of customer satisfaction. For example, if it is found that a certain type of customer is dissatisfied with a certain service, the enterprise can take measures to improve the service to improve customer satisfaction and increase customer loyalty.

[0051] Further, considering that a classifier is a machine learning model or algorithm for classifying input data into different categories or labels. In feature classification, the role of the classifier is to classify samples according to the input feature vector. Feature classification is a supervised learning task in which each sample has a known category label. The classifier constructs a model or function by learning the relationship between features and categories in the training data, and can classify new unknown samples. Therefore, in an embodiment of the present application, the customer satisfaction analysis feature vector can be feature classified by a classifier to obtain a classification result.

[0052] Specifically, step S130, analyzing the customer satisfaction analysis feature vector to obtain a classification result, is used to: perform feature classification on the customer satisfaction analysis feature vector through a classifier to obtain a classification result for indicating whether the customer to be analyzed is satisfied with the service.

[0053] It is worth mentioning that in addition to using classifiers to identify and classify customer satisfaction information features, there is another possible implementation method to judge and identify customer satisfaction. This implementation method is to use clustering algorithms for customer satisfaction analysis. Clustering algorithms can cluster similar samples together to form different clusters or clusters without knowing the category labels in advance. This method can help discover potential patterns and cluster structures in the data, thereby analyzing customer satisfaction. The following is a possible implementation method for customer satisfaction analysis using clustering algorithms: 1. Data preparation: Collect and prepare feature vector data sets for customer satisfaction analysis. 2. Feature preprocessing: Preprocess feature vectors, such as standardization, normalization, or processing missing values. 3. Clustering algorithm selection: Select a clustering algorithm suitable for customer satisfaction analysis, such as K-means clustering, hierarchical clustering, DBSCAN, etc. 4. Clustering model training: Use the selected clustering algorithm to train the clustering model on the feature vector data. 5. Clustering result analysis: Analyze the clustering results to understand the customer satisfaction patterns and characteristics in different clusters or clusters. 6. Result interpretation and decision support: Based on the clustering results, explain the customer satisfaction characteristics in different clusters or clusters, and make corresponding decisions and improvement measures as needed.

[0054] The advantage of the clustering algorithm is that it can discover potential patterns and cluster structures in the data without pre-defining category labels. This can help companies understand the characteristics of customer satisfaction in different clusters and develop personalized strategies and improvement plans for different clusters. However, compared with classifiers, clustering algorithms do not provide clear category labels, but rather provide cluster or cluster division results that require further interpretation and analysis.

[0055] In summary, the intelligent business analysis method based on big data according to the embodiment of the present application is explained, which first obtains the text data information of the customer to be analyzed and the voice emotion data information of the customer to be analyzed from the customer service call voice data, and then uses the customer text information feature extraction module based on the context encoder to extract and analyze the text data information of the customer to be analyzed, and at the same time, the voice emotion data information of the customer to be analyzed is extracted and analyzed by the voice noise reduction module based on the automatic encoder and the voice emotion feature extractor based on the convolutional neural network model, and then based on the feature information of the two, a classification result is obtained to indicate whether the customer to be analyzed is satisfied with the service. In this way, the enterprise can adjust the service mode and attitude of the customer service center in a timely manner to improve customer satisfaction, thereby improving the efficiency of business marketing.

[0056] Exemplary Systems

[0057] Figure 2 The figure shows a schematic block diagram of an intelligent business analysis system based on big data according to an embodiment of the present application. Figure 2 As shown, the big data-based intelligent business analysis system 100 according to the embodiment of the present application includes: a voice information collection module 110, which is used to collect customer service call voice data and obtain text data information of the customer to be analyzed and voice emotion data information of the customer to be analyzed; a voice information feature analysis module 120, which is used to obtain a customer satisfaction analysis feature vector based on the text data information of the customer to be analyzed and the voice emotion data information of the customer to be analyzed; and a user satisfaction feature classification module 130, which is used to analyze the customer satisfaction analysis feature vector to obtain a classification result.

[0058] Further, Figure 3 The diagram shows an architecture diagram of an intelligent business analysis system based on big data according to an embodiment of the present application. Figure 3As shown, firstly, the text data information of the customer to be analyzed and the voice emotion data information of the customer to be analyzed are obtained from the customer service call voice data; then, the text data information of the customer to be analyzed is passed through the customer text information feature extraction module based on the context encoder to obtain the customer text semantic understanding feature vector; in addition, the customer voice emotion waveform is obtained from the voice emotion data information of the customer to be analyzed; then, the customer voice emotion waveform is passed through the voice denoising module based on the autoencoder to obtain the denoised voice emotion waveform; further, the denoised voice emotion waveform is passed through the voice emotion feature extractor based on the convolutional neural network model to obtain the customer voice emotion feature vector; then, the smooth parameterized expression of the customer text semantic understanding feature vector and the customer voice emotion feature vector relative to the target classification distribution is calculated respectively to obtain the optimized customer text semantic understanding feature vector and the optimized customer voice emotion feature vector; then, the optimized customer text semantic understanding feature vector and the optimized customer voice emotion feature vector are feature fused to obtain the customer satisfaction analysis feature vector; finally, the customer satisfaction analysis feature vector is passed through the classifier to obtain the classification result, and the classification result is used to indicate whether the customer to be analyzed is satisfied with the service.

[0059] As described above, the intelligent business analysis system 100 based on big data according to the embodiment of the present application can be implemented in various wireless terminals, such as a server for intelligent business analysis based on big data. In one example, the intelligent business analysis system 100 based on big data according to the embodiment of the present application can be integrated into the wireless terminal as a software module and / or a hardware module. For example, the intelligent business analysis system 100 based on big data can be a software module in the operating system of the wireless terminal, or can be an application developed for the wireless terminal; of course, the intelligent business analysis system 100 based on big data can also be one of the many hardware modules of the wireless terminal.

[0060] Alternatively, in another example, the big data-based intelligent business analysis system 100 and the wireless terminal may also be separate devices, and the big data-based intelligent business analysis system 100 may be connected to the wireless terminal via a wired and / or wireless network and transmit interactive information in accordance with an agreed data format.

[0061] Here, those skilled in the art can understand that the specific functions and operations of each step in the above-mentioned intelligent business analysis system based on big data have been referred to above. Figure 1 The invention has been introduced in detail in the description of the intelligent business analysis method based on big data, and therefore, its repeated description will be omitted.

[0062] Exemplary Electronic Devices

[0063] Below, reference Figure 4 To describe an electronic device according to an embodiment of the present application.

[0064] Figure 4 A block diagram of an electronic device according to an embodiment of the present application is illustrated.

[0065] like Figure 4 As shown, the electronic device includes a processor 601, a communication interface 602, a memory 603 and a communication bus 604. The processor 601, the communication interface 602, and the memory 603 communicate with each other through the communication bus 604. The processor 601, the communication interface 602, the memory 603 and other components can also communicate through a network connection. The present disclosure does not limit the type and function of the network. It should be noted that Figure 4 The components of the electronic device shown are merely exemplary and non-limiting. The electronic device may also have other components according to actual application requirements.

[0066] For example, the memory 603 is used to store computer-readable instructions non-transiently. When the processor 601 is used to execute the computer-readable instructions, the edge detection method described in any of the above embodiments is implemented. The specific implementation of each step of the edge detection method and the related explanation content can be referred to the embodiment of the edge detection method above, which will not be repeated here.

[0067] For example, other implementations of the edge detection method implemented by the processor 601 executing computer-readable instructions stored in the memory 603 are the same as the implementations mentioned in the aforementioned method embodiment part, and will not be repeated here.

[0068] For example, the communication bus 604 may be a Peripheral Component Interconnect (PCI) bus or an Extended Industrial Standard Architecture (EISA) bus, etc. The communication bus may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, only one thick line is used in the figure, but it does not mean that there is only one bus or one type of bus.

[0069] For example, the communication interface 602 is used to implement communication between the electronic device and other devices.

[0070] For example, the processor 601 and the memory 603 may be arranged on a server side (or a cloud side).

[0071] For example, the processor 601 can control other components in the electronic device to perform the desired functions. The processor 601 can be a device with data processing capabilities and / or program execution capabilities, such as a central processing unit (CPU), a network processor (NP), a tensor processing unit (TPU), or a graphics processing unit (GPU); it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The central processing unit (CPU) can be an X86 or ARM architecture, etc.

[0072] For example, the memory 603 may include any combination of one or more computer program products, and the computer program product may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and / or cache memory (cache), etc. Non-volatile memory may include, for example, read-only memory (ROM), hard disk, erasable programmable read-only memory (EPROM), portable compact disk read-only memory (CD-ROM), USB memory, flash memory, etc. One or more computer-readable instructions may be stored on the computer-readable storage medium, and the processor 601 may execute the computer-readable instructions to implement various functions of the electronic device. Various applications and various data may also be stored in the storage medium.

[0073] For example, in some embodiments, the electronic device may further include an image acquisition component. The image acquisition component is used to obtain an input image of the object. The memory 603 is also used to store the input image.

[0074] For example, the image acquisition component may be a camera of a smartphone, a camera of a tablet computer, a camera of a personal computer, a lens of a digital camera, or even a webcam.

[0075] For example, the input image may be an original image directly acquired by the image acquisition component, or may be an image obtained after preprocessing the original image. Preprocessing may eliminate irrelevant information or noise information in the original image so as to better process the input image. Preprocessing may include, for example, image augmentation (DataAugment), image scaling, gamma correction, image enhancement, or noise reduction filtering on the original image.

[0076] For example, for a detailed description of the process of edge detection performed by the electronic device, reference may be made to the relevant description in the embodiment of the edge detection method, and repeated parts will not be repeated.

[0077] Exemplary non-transitory computer-readable storage medium

[0078] Below, reference Figure 5 To describe the non-transitory computer-readable storage medium according to an embodiment of the present application.

[0079] Figure 5 A block diagram of a non-transitory computer-readable storage medium according to an embodiment of the present application is illustrated.

[0080] like Figure 5 As shown, one or more computer readable instructions 701 may be non-transitorily stored on a storage medium 700 .

[0081] For example, when the computer readable instructions 701 are executed by a processor, one or more steps in the edge detection method described above may be performed.

[0082] For example, the storage medium 700 may be applied to the above-mentioned electronic device. For example, the storage medium 700 may include the memory 603 in the electronic device.

[0083] For example, the description of the storage medium 700 may refer to the description of the memory in the embodiment of the electronic device, and the repeated parts will not be repeated.

Claims

1. An intelligent business analysis method based on big data, characterized in that: include: Obtaining text data information of the customer to be analyzed and voice emotion data information of the customer to be analyzed from the customer service call voice data; Obtaining a customer satisfaction analysis feature vector based on the text data information of the customer to be analyzed and the voice emotion data information of the customer to be analyzed; Analyzing the customer satisfaction analysis feature vector to obtain a classification result; Wherein, obtaining a customer satisfaction analysis feature vector based on the text data information of the customer to be analyzed and the voice emotion data information of the customer to be analyzed includes: Performing semantic understanding analysis on the text data information of the customer to be analyzed by a customer text information feature extraction module to obtain a customer text semantic understanding feature vector; Preprocessing the voice emotion data information of the customer to be analyzed to obtain a noise-reduced voice emotion waveform; Passing the noise-reduced speech emotion waveform through a speech emotion feature extractor to obtain a customer speech emotion feature vector; and The customer text semantic understanding feature vector and the customer voice emotion feature vector are fused based on a smoothed posterior expression of an output relative objective function to obtain a customer satisfaction analysis feature vector; The customer text semantic understanding feature vector and the customer voice emotion feature vector are fused based on the smoothed posterior expression of the output relative objective function to obtain the customer satisfaction analysis feature vector, including: Calculating a smooth parameterized expression of the customer text semantic understanding feature vector relative to a target classification distribution to obtain an optimized customer text semantic understanding feature vector; Calculating a smoothed parameterized expression of the customer voice emotion feature vector relative to a target classification distribution to obtain an optimized customer voice emotion feature vector; Fusion of the optimized customer text semantic understanding feature vector and the optimized customer voice emotion feature vector to obtain a customer satisfaction analysis feature vector; The method comprises: calculating a smooth parameterized expression of the customer text semantic understanding feature vector relative to the target classification distribution to obtain an optimized customer text semantic understanding feature vector; calculating a smooth parameterized expression of the customer voice emotion feature vector relative to the target classification distribution to obtain an optimized customer voice emotion feature vector, including: Wherein, V1 represents the customer text semantic understanding feature vector, V2 represents the customer voice emotion feature vector, represents the mean vector of the customer text semantic understanding feature vector and the customer voice emotion feature vector, They respectively represent positional subtraction and addition of vectors, λ represents a predetermined hyperparameter, V'1 represents the optimized customer text semantic understanding feature vector, and V'2 represents the optimized customer voice emotion feature vector.

2. The intelligent business analysis method based on big data according to claim 1 is characterized in that: The customer text information feature extraction module is a customer text information feature extraction module based on a context encoder.

3. The intelligent business analysis method based on big data according to claim 2 is characterized in that: The voice emotion data information of the customer to be analyzed is preprocessed to obtain a noise-reduced voice emotion waveform diagram, including: Acquire the voice signal of the voice emotion data information of the customer to be analyzed to obtain a customer voice emotion waveform diagram; and The customer's voice emotion waveform is subjected to noise reduction processing by a voice noise reduction module to obtain the noise-reduced voice emotion waveform.

4. The intelligent business analysis method based on big data according to claim 3 is characterized in that: The speech denoising module is a speech denoising module based on an automatic encoder.

5. The intelligent business analysis method based on big data according to claim 4 is characterized in that: Analyze the customer satisfaction analysis feature vector to obtain a classification result, which is used to: perform feature classification on the customer satisfaction analysis feature vector through a classifier to obtain a classification result indicating whether the customer to be analyzed is satisfied with the service.

6. An intelligent business analysis system based on big data, characterized in that: include: The voice information collection module is used to obtain the text data information of the customer to be analyzed and the voice emotion data information of the customer to be analyzed from the customer service call voice data; A voice information feature analysis module, used to obtain a customer satisfaction analysis feature vector based on the text data information of the customer to be analyzed and the voice emotion data information of the customer to be analyzed; A user satisfaction feature classification module, used to analyze the customer satisfaction analysis feature vector to obtain a classification result; Wherein, obtaining a customer satisfaction analysis feature vector based on the text data information of the customer to be analyzed and the voice emotion data information of the customer to be analyzed includes: Performing semantic understanding analysis on the text data information of the customer to be analyzed by a customer text information feature extraction module to obtain a customer text semantic understanding feature vector; Preprocessing the voice emotion data information of the customer to be analyzed to obtain a noise-reduced voice emotion waveform; Passing the noise-reduced speech emotion waveform through a speech emotion feature extractor to obtain a customer speech emotion feature vector; and The customer text semantic understanding feature vector and the customer voice emotion feature vector are fused based on a smoothed posterior expression of an output relative objective function to obtain a customer satisfaction analysis feature vector; The customer text semantic understanding feature vector and the customer voice emotion feature vector are fused based on the smoothed posterior expression of the output relative objective function to obtain the customer satisfaction analysis feature vector, including: Calculating a smooth parameterized expression of the customer text semantic understanding feature vector relative to a target classification distribution to obtain an optimized customer text semantic understanding feature vector; Calculating a smoothed parameterized expression of the customer voice emotion feature vector relative to a target classification distribution to obtain an optimized customer voice emotion feature vector; Fusion of the optimized customer text semantic understanding feature vector and the optimized customer voice emotion feature vector to obtain a customer satisfaction analysis feature vector; The method comprises: calculating a smooth parameterized expression of the customer text semantic understanding feature vector relative to the target classification distribution to obtain an optimized customer text semantic understanding feature vector; calculating a smooth parameterized expression of the customer voice emotion feature vector relative to the target classification distribution to obtain an optimized customer voice emotion feature vector, including: Wherein, V1 represents the customer text semantic understanding feature vector, V2 represents the customer voice emotion feature vector, represents the mean vector of the customer text semantic understanding feature vector and the customer voice emotion feature vector, They respectively represent positional subtraction and addition of vectors, λ represents a predetermined hyperparameter, V'1 represents the optimized customer text semantic understanding feature vector, and V'2 represents the optimized customer voice emotion feature vector.

7. An electronic device comprising: processor; as well as A memory, in which computer program instructions are stored, and when the computer program instructions are executed by the processor, the processor executes the big data-based intelligent business analysis method as described in any one of claims 1 to 5.

Images