Artificial intelligence-based data analysis method, device, equipment and storage medium

Abstract

The embodiment of the application belongs to the field of artificial intelligence, and relates to a data analysis method based on artificial intelligence, comprising the following steps: acquiring voice dialogue data of a user and an agent in a business conversation process; converting the voice dialogue data into text data, and obtaining dialogue text by converting the text data; performing classification processing on the dialogue text according to a preset rule to obtain a theme label corresponding to the dialogue text; constructing a corresponding directed graph based on the dialogue text and the theme label; and performing similarity analysis processing on the directed graph based on a preset similarity analysis model to generate a similarity result between each directed graph. The application also provides a data analysis device based on artificial intelligence, a computer device and a storage medium. In addition, the application also relates to blockchain technology, and the similarity result can be stored in the blockchain. The application analyzes the similarity relationship of the theme path of the dialogue technique based on the use of the similarity analysis model to quickly and accurately obtain the corresponding similarity result.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of artificial intelligence development, and in particular relates to a data analysis method and device based on artificial intelligence, a computer device and a storage medium. BACKGROUND

[0002] In the field of financial technology, it is usually necessary for a seat to communicate with a customer for business recommendation. At present, the dialogue technique only stays in the topic classification of dialogue data, and cannot provide useful information for the marketing of the seat according to the topic classification, so as to fail to provide effective help for the work of the seat. However, different dialogues often have similar logic, and how to discover the similar logic between different dialogues and reasonably utilize it to create more conditions for the marketing of the seat is a technical problem that needs to be solved at present. SUMMARY

[0003] The purpose of the embodiments of the present application is to provide a data analysis method and device based on artificial intelligence, a computer device and a storage medium, so as to solve the technical problem that the current dialogue technique only stays in the topic classification of dialogue data, and cannot provide useful information for the marketing of the seat according to the topic classification, and cannot discover the similar logic between different dialogues and reasonably utilize it to create more conditions for the marketing of the seat.

[0004] In order to solve the above technical problem, the embodiments of the present application provide a data analysis method based on artificial intelligence, which adopts the following technical solution:

[0005] Obtain voice dialogue data of a user and a seat in a business communication process;

[0006] Convert the voice dialogue data into text data, and convert the text data according to a preset information type to obtain corresponding dialogue text;

[0007] Classify the dialogue text according to a preset rule to obtain a topic label corresponding to the dialogue text;

[0008] Construct a corresponding directed graph based on the dialogue text and the topic label; wherein the number of the directed graphs includes a plurality;

[0009] Perform similarity analysis processing on the directed graphs based on a preset similarity analysis model to generate a similarity result between each of the directed graphs; wherein the similarity analysis model is composed of a graph convolution network, a tensor neural network and a full connection layer.

[0010] Further, the step of performing similarity analysis processing on the directed graphs based on the preset similarity analysis model to generate the similarity result between each of the directed graphs specifically includes:

[0011] reconstructing features of each node in each of the directed graphs through a graph convolution network in the similarity analysis model to obtain a corresponding node representation vector;

[0012] encoding each of the directed graphs based on a preset attention mechanism to obtain a graph feature vector of each of the directed graphs;

[0013] processing the graph feature vectors of each of the directed graphs based on a tensor neural network in the similarity analysis model to obtain a similarity vector between target directed graphs; wherein the target directed graphs are any two of all the directed graphs;

[0014] generating a histogram feature of the directed graph based on the node representation vector;

[0015] processing the similarity vector between the target directed graphs and a target histogram feature between the target directed graphs based on a fully connected layer in the similarity analysis model to obtain a target similarity result between the target directed graphs.

[0016] Further, the step of generating the histogram feature of the directed graph based on the node representation vector specifically comprises:

[0017] performing inner product calculation on the node representation vector of the directed graph to obtain a corresponding correlation matrix;

[0018] performing conversion processing on the correlation matrix to obtain a histogram feature corresponding to the directed graph.

[0019] Further, the step of processing the similarity vector between the target directed graphs and the target histogram feature between the target directed graphs based on the fully connected layer in the similarity analysis model to obtain the target similarity result between the target directed graphs specifically comprises:

[0020] performing splicing processing on the similarity vector between the target directed graphs and the target histogram feature between the target directed graphs to obtain a corresponding splicing feature;

[0021] inputting the splicing feature into the fully connected layer in the similarity analysis model to obtain an output result corresponding to the target directed graphs output by the fully connected layer;

[0022] taking the output result as the target similarity result.

[0023] Further, the step of classifying the dialogue text according to a preset rule to obtain a topic label corresponding to the dialogue text specifically comprises:

[0024] processing the dialogue text based on a preset text representation model to generate text representation data corresponding to the dialogue text;

[0025] performing feature extraction on the text representation data based on a preset feature extraction model to obtain corresponding feature data;

[0026] using the feature data as the topic label.

[0027] Further, the step of constructing a corresponding directed graph based on the dialogue text and the topic label specifically includes:

[0028] obtaining a preset target field;

[0029] obtaining target data corresponding to the target field from the dialogue text based on the target field;

[0030] constructing the directed graph based on the target data and the topic label.

[0031] Further, after the step of performing similarity analysis on the directed graph based on a preset similarity analysis model to generate a similarity result between each of the directed graphs, the method further includes:

[0032] obtaining the similarity result;

[0033] generating a corresponding dialogue analysis report based on the similarity result;

[0034] storing the dialogue analysis report.

[0035] To solve the above technical problems, the embodiment of the present application also provides a data analysis device based on artificial intelligence, which adopts the following technical solution:

[0036] a first obtaining module configured to obtain voice dialogue data of a user and an agent in a business process;

[0037] a conversion module configured to convert the voice dialogue data into text data and convert the text data according to a preset information type to obtain corresponding dialogue text;

[0038] a classification module configured to classify the dialogue text according to a preset rule to obtain a topic label corresponding to the dialogue text;

[0039] a construction module configured to construct a corresponding directed graph based on the dialogue text and the topic label; wherein the number of the directed graphs includes a plurality;

[0040] The processing module is configured to perform similarity analysis on the directed graphs based on a preset similarity analysis model to generate similarity results between the directed graphs.

[0041] To solve the above technical problems, the embodiment of the application further provides a computer device, which adopts the technical scheme as follows:

[0042] Obtain voice conversation data of a user and an agent in a business process.

[0043] Convert the voice conversation data into text data, and convert the text data according to a preset information type to obtain corresponding conversation text.

[0044] Classify the conversation text according to a preset rule to obtain a topic label corresponding to the conversation text.

[0045] Construct a corresponding directed graph based on the conversation text and the topic label, wherein the number of the directed graphs includes multiple.

[0046] Perform similarity analysis on the directed graphs based on a preset similarity analysis model to generate similarity results between the directed graphs, wherein the similarity analysis model is composed of a graph convolution network, a tensor neural network and a fully connected layer.

[0047] To solve the above technical problems, the embodiment of the application further provides a computer readable storage medium, which adopts the technical scheme as follows:

[0048] Obtain voice conversation data of a user and an agent in a business process.

[0049] Convert the voice conversation data into text data, and convert the text data according to a preset information type to obtain corresponding conversation text.

[0050] Classify the conversation text according to a preset rule to obtain a topic label corresponding to the conversation text.

[0051] Construct a corresponding directed graph based on the conversation text and the topic label, wherein the number of the directed graphs includes multiple.

[0052] Perform similarity analysis on the directed graphs based on a preset similarity analysis model to generate similarity results between the directed graphs, wherein the similarity analysis model is composed of a graph convolution network, a tensor neural network and a fully connected layer.

[0053] Compared with the prior art, the embodiment of the application has the following beneficial effects:

[0054] The embodiment of the application first loads the voice dialogue data of the user and the agent in the business process; then converts the voice dialogue data into text data, and converts the text data according to a preset information type to obtain corresponding dialogue text; then classifies and processes the dialogue text according to a preset rule to obtain a topic label corresponding to the dialogue text; subsequently, a corresponding directed graph is constructed based on the dialogue text and the topic label; finally, a similarity analysis model is used to perform similarity analysis processing on the directed graph, and a similarity result between each directed graph is generated. The embodiment of the application completes the construction of the directed graph based on the dialogue text and the topic label of the dialogue text of the user and the agent in the business process, and then uses a similarity analysis model to perform similarity analysis processing on the constructed directed graph, so that the similarity result between the directed graphs can be quickly and accurately generated, the generation efficiency of the similarity result is improved, and the data accuracy of the similarity result is ensured. The similarity relationship of the topic path of the dialogue skill is analyzed based on the similarity analysis model to obtain the corresponding similarity result, which is beneficial to helping the agent end to discover the common dialogue path from the voice dialogue data, so that the similarity logic of the common dialogue path can be used to timely and reasonably adjust the dialogue skill to guide the user to form an optimal purchase path, and the work experience of the agent is improved. BRIEF DESCRIPTION OF DRAWINGS

[0055] In order to more clearly illustrate the scheme in the application, the drawings needed in the description of the embodiments of the application will be briefly introduced. Obviously, the drawings in the following description are some embodiments of the application, and other drawings can be obtained by those skilled in the art without creative labor.

[0056] Figure 1 is an exemplary system architecture diagram to which the application can be applied;

[0057] Figure 2 Flowchart of one embodiment of the data analysis method based on artificial intelligence according to the application;

[0058] Figure 3 is a structural schematic diagram of one embodiment of the data analysis device based on artificial intelligence according to the application;

[0059] Figure 4 is a structural schematic diagram of one embodiment of the computer device according to the application. DETAILED DESCRIPTION

[0060] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs; the terminology used in the description herein is for describing particular embodiments only and is not intended to be limiting upon the application; the terms "comprising," "including," and "having," and variations thereof, as used in enrolling and claims herein, are intended to be open-ended and to mean including, but not limited to; the terms "first," "second," and the like, as used in the description herein, are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. Unless otherwise indicated herein, the materials described herein can be used in a variety of applications.

[0061] Reference herein to "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment can be included in at least one embodiment of the application. The appearances of the phrase that in various places in the specification are not necessarily all referring to the same embodiment, nor are they necessarily mutually exclusive of one another. As will be apparent to those of ordinary skill in the art, embodiments described herein can be combined with other embodiments.

[0062] In order to make the technical personnel in the art better understand the scheme of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings.

[0063] As shown in Figure 1 The system architecture 100 can include terminal devices 101, 102, 103, a network 104, and a server 105. The network 104 is used to provide a communication link medium between the terminal devices 101, 102, 103 and the server 105. The network 104 can include various connection types, such as wired, wireless communication links, or optical fiber cables, etc.

[0064] A user can use the terminal devices 101, 102, 103 to interact with the server 105 through the network 104 to receive or send messages, etc. Various communication client applications can be installed on the terminal devices 101, 102, 103, such as web browser applications, shopping applications, search applications, instant messaging tools, email clients, social platform software, etc.

[0065] The terminal devices 101, 102, 103 can be various electronic devices with display screens and supporting web browsing, including but not limited to smart phones, tablet computers, e-book readers, MP3 (Moving Picture Experts Group Audio Layer III) players, MP4 (Moving Picture Experts Group Audio Layer IV) players, laptop portable computers, desktop computers, and the like.

[0066] The server 105 can be a server providing various services, for example, a background server providing support for a page displayed on the terminal devices 101, 102, 103.

[0067] It should be noted that the artificial intelligence-based data analysis method provided in the embodiments of the present application is generally executed by a server / terminal device, and accordingly, the artificial intelligence-based data analysis apparatus is generally arranged in a server / terminal device.

[0068] It should be understood that Figure 1 The number of terminal devices, networks, and servers in

[0069] With reference to Figure 2 , a flowchart of one embodiment of the artificial intelligence-based data analysis method according to the present application is shown. The artificial intelligence-based data analysis method includes the following steps:

[0070] In step S201, voice conversation data of a user and an agent in a business process is acquired.

[0071] In the present embodiment, the electronic device (for example, the terminal device 101, 102, 103 or the server 105) on which the artificial intelligence-based data analysis method runs can be a terminal device, a network, or a server. Figure 1The server / terminal device shown) can obtain voice dialogue data through wired connection or wireless connection. It should be noted that the wireless connection can include but is not limited to 3G / 4G / 5G connection, WiFi connection, Bluetooth connection, WiMAX connection, Zigbee connection, UWB (ultra wideband) connection, and other now known or future developed wireless connection. The dialogue between the customer and the agent involves the jump of different dialogue topics, which can be represented in the form of a graph through the construction of the topic path. There are often similar logics between these graphs, and it is essential to discover these logics and use them reasonably to create more conditions for agent marketing. The previous dialogue technique only stays in the topic classification of dialogue sentences, and has not involved the similarity research of dialogue technique topic path. Therefore, the application proposes a similarity analysis scheme of agent topic path based on a similarity analysis model.

[0072] In step S202, the voice dialogue data is converted into text data, and the text data is converted according to a preset information type to obtain corresponding dialogue text.

[0073] In this embodiment, the voice can be converted into text data in real time by using existing ASR technology (such as the ASR technology of Patsnap). The information types include session ID, agent ID, timestamp, dialogue content, etc. The text data can be converted based on the information type to form formatted data including session ID, agent ID, timestamp, and dialogue content.

[0074] In step S203, the dialogue text is classified according to a preset rule to obtain a topic label corresponding to the dialogue text.

[0075] In this embodiment, the topic label can specifically include confirmation of offer willingness, confirmation of offer scheme, broadcast of offer and cashback, broadcast of value-added service, confirmation of insurance information, introduction of payment process, and recommendation of credit card. A dialogue may involve the reciprocal jump of different topics, and the topic path formed thereby may be inconsistent. Different dialogue paths corresponding to different call IDs often have certain differences and connections. However, the similarity between them has not been mined in previous research, missing the exploration and discussion of common problems, and affecting the adjustment of the navigation scene. In addition, the specific implementation process of classifying the dialogue text according to the preset rule to obtain the topic label corresponding to the dialogue text will be further described in detail in the subsequent specific embodiments, and will not be described here.

[0076] In step S204, a corresponding directed graph is constructed based on the dialogue text and the topic label; wherein the number of the directed graph includes a plurality.

[0077] In the embodiment, the specific implementation process of constructing the corresponding directed graph based on the dialogue text and the topic label will be further described in detail in subsequent embodiments, and will not be described in detail here.

[0078] In step S205, the similarity analysis model is used to analyze the similarity of the directed graphs, and a similarity result between the directed graphs is generated.

[0079] In the embodiment, the specific implementation process of analyzing the similarity of the directed graphs based on the similarity analysis model will be further described in detail in subsequent embodiments, and will not be described in detail here. The training and generation process of the similarity analysis model can include: after the training data set is constructed, the training data set is normalized, the initial parameters of the directed graph deep neural network are set, and then the directed graph deep neural network is trained on the training data set by using the Adam optimization algorithm, and the model parameters are iteratively optimized to generate the final similarity analysis model.

[0080] The voice dialogue data of the user and the agent in the business process is first loaded, and then the voice dialogue data is converted into text data, and the text data is converted into corresponding dialogue text according to a preset information type. The dialogue text is classified according to a preset rule to obtain a topic label corresponding to the dialogue text. The dialogue text and the topic label are used to construct a corresponding directed graph. Finally, the similarity of the directed graph is analyzed based on a preset similarity analysis model, and a similarity result between the directed graphs is generated. The dialogue text and the topic label of the user and the agent in the business process are used to construct a directed graph, and then the similarity analysis model is used to analyze the similarity of the constructed directed graph, which can quickly and accurately generate a similarity result between the directed graphs, improve the generation efficiency of the similarity result, and ensure the data accuracy of the similarity result. The similarity relationship of the topic path of the dialogue skill based on the similarity analysis model is analyzed to obtain a corresponding similarity result, which can help the agent to find common dialogue paths from the voice dialogue data, and thus the similarity logic of the common dialogue path can be used to adjust the dialogue skill in a timely manner to guide the user to form an optimal purchase path, and the work experience of the agent is improved.

[0081] In some optional implementations, step S205 includes the following steps:

[0082] The feature of each node in each directed graph is reconstructed by a graph convolution network in the similarity analysis model to obtain a corresponding node representation vector.

[0083] In this embodiment, the feature of each node in each directed graph can be reconstructed by weighting using an aggregation function, and the specific process can be seen from the following formula: wherein, is the n node encoding, is a weight matrix, is a bias, and N(n) is an adjacency matrix, (·) is a ReLU activation function.

[0084] Each of the directed graphs is encoded based on a preset attention mechanism to obtain a graph feature vector of each of the directed graphs.

[0085] In this embodiment, the above attention mechanism can specifically adopt an Att attention mechanism. In order to obtain global features, the Att attention mechanism can be used to obtain the importance of each point in the global in the graph-level encoding link, and the specific process can be seen from the following formula: wherein, h is a graph embedding, which is a weighted sum of point-level encoding, is an m node encoding, wherein is a learnable weight matrix, (·) is a sigmoid activation function.

[0086] The graph feature vectors of each of the directed graphs are processed based on a tensor neural network in the similarity analysis model to obtain a similarity vector between target directed graphs; wherein the target directed graphs are any two of all the directed graphs.

[0087] In this embodiment, the graph feature vector is obtained after graph-level encoding. In order to calculate the similarity between the two vectors, a tensor neural network (Neural Tensor Networks, NTN) can be used to calculate the mutual relationship between each point in the two directed graphs, and the relationship features between the graph and the graph can be calculated according to the following formula: wherein, is a weight tensor, [] is a connection operation, (·) is an activation function, and K is a hyperparameter.

[0088] The histogram feature of the directed graph is generated based on the node representation vector.

[0089] In the embodiment, the specific implementation process of generating the histogram feature of the directed graph based on the node representation vector is further described in detail in subsequent specific embodiments, and is not described in detail here.

[0090] The target similarity result between the target directed graphs is obtained by processing the similarity vector between the target directed graphs and the target histogram feature between the target directed graphs based on the full connection layer in the similarity analysis model.

[0091] In the embodiment, the specific implementation process of generating the histogram feature of the directed graph based on the node representation vector is further described in detail in subsequent specific embodiments, and is not described in detail here.

[0092] The node representation vector corresponding to each node in each directed graph is obtained by reconstructing and processing the feature of each node in each directed graph through the graph convolution network in the similarity analysis model. Then, the graph feature vector of each directed graph is obtained by encoding each directed graph based on the preset attention mechanism. Then, the similarity vector between the target directed graphs is obtained by processing the graph feature vector of each directed graph based on the tensor neural network in the similarity analysis model. Subsequently, the histogram feature of the directed graph is generated based on the node representation vector. Finally, the target similarity result between the target directed graphs is obtained by processing the similarity vector between the target directed graphs and the target histogram feature between the target directed graphs based on the full connection layer in the similarity analysis model.

[0093] In some optional implementations of the embodiment, the histogram feature of the directed graph is generated based on the node representation vector, including the following steps:

[0094] The node representation vector of the directed graph is calculated by inner product to obtain a corresponding correlation matrix.

[0095] In the embodiment, in order to consider the information of the local node in the directed graph, the inner product of the node representation vector is calculated to obtain a correlation matrix.

[0096] The correlation matrix is converted to obtain a histogram feature corresponding to the directed graph.

[0097] The application obtains a corresponding correlation matrix h by performing inner product calculation on the node representation vectors of the directed graph. The correlation matrix is subsequently converted to obtain a histogram feature corresponding to the directed graph, so as to realize fast acquisition of the histogram feature of the directed graph. The histogram feature of the directed graph can be processed based on the full connection layer in the similarity analysis model, so as to realize fast and accurate generation of the similarity result between directed graphs. The application can realize fast and accurate generation of the similarity result between directed graphs based on the use of the similarity analysis model, improve the generation efficiency of the similarity result, and ensure the data accuracy of the similarity result.

[0098] In some optional implementations, the processing of the similarity vector between the target directed graphs and the target histogram feature between the target directed graphs based on the full connection layer in the similarity analysis model to obtain the target similarity result between the target directed graphs includes the following steps:

[0099] The similarity vector between the target directed graphs and the target histogram feature between the target directed graphs are spliced to obtain a corresponding spliced feature.

[0100] The spliced feature is input into the full connection layer in the similarity analysis model to obtain an output result corresponding to the target directed graph output by the full connection.

[0101] The output result is taken as the target similarity result.

[0102] The application splices the similarity vector between the target directed graphs and the target histogram feature between the target directed graphs to obtain a corresponding spliced feature. Then, the spliced feature is input into the full connection layer in the similarity analysis model to obtain an output result corresponding to the target directed graph output by the full connection. Subsequently, the output result is taken as the target similarity result. The application can realize fast and accurate generation of the similarity result between directed graphs by processing the spliced feature generated by splicing the similarity vector and the histogram feature of the directed graph based on the use of the similarity analysis model, improve the generation efficiency of the similarity result, and ensure the data accuracy of the similarity result.

[0103] In some optional implementations, the step S203 includes the following steps:

[0104] The dialogue text is processed based on a preset text representation model to generate text representation data of the dialogue text.

[0105] In the embodiment, the text representation model can be specifically a RoBERTa text representation model. After inputting the dialogue text into the RoBERTa text representation model, the semantic knowledge in the dialogue text is fully mined by the RoBERTa text representation model, and corresponding is generated.

[0106] The text representation data is subjected to feature extraction based on a preset feature extraction model, to obtain corresponding feature data.

[0107] In the embodiment, the feature extraction model can be specifically an RCNN feature extraction model. The RCNN feature extraction model is a bidirectional recurrent structure RNN network, based on which the context syntax and semantic information of the input text can be obtained, and then the most important features are automatically screened out by cooperating with a max-pooling CNN network, and finally the theme label corresponding to the input text is output.

[0108] The feature data is taken as the theme label.

[0109] The dialogue text is processed based on the preset text representation model to generate text representation data corresponding to the dialogue text, and then the text representation data is subjected to feature extraction based on a preset feature extraction model to obtain corresponding feature data, and subsequently the feature data is taken as the theme label. By using the text representation model and the feature extraction model, the theme label corresponding to the dialogue text can be quickly and accurately generated, the generation efficiency of the theme label of the dialogue text is effectively improved, and the data accuracy of the theme label is ensured.

[0110] In some optional implementation manners of the embodiment, step S203 comprises the following steps:

[0111] A preset target field is obtained.

[0112] In the embodiment, the target field specifically comprises a session ID, an agent ID, and a timestamp.

[0113] Based on the target field, target data corresponding to the target field is obtained from the dialogue text.

[0114] In the embodiment, according to the target field, the field data matched with the target field in the dialogue text, i.e., the session ID, the agent ID, and the timestamp in the dialogue text, can be obtained.

[0115] The target data and the theme label are used to construct the directed graph.

[0116] In the embodiment, the directed graph construction process of the dialogue text topic can include: first traversing the conversation ID of each group of conversations in the dialogue text, and making it into a format that can be used by torch_geometric; then, the topic label records of each agent are sorted according to the timestamps to obtain a dialogue topic sequence file of each agent; then, the dialogue topic sequence file is encoded by Labelencoder to make the edge index numbered from 0, and the feature representation of the corresponding point is the encoding of the source node and the next node generated for each point; subsequently, the dialogue feedback topics of the agents are taken as the research object, and each dialogue feedback topic is taken as a point in the directed graph; the relationship between each dialogue feedback topic and other dialogue feedback topics is determined according to the upstream and downstream relationship of various dialogue feedback topics, and the points with causal relationship are connected by vectors; and all points and vectors constitute a directed graph.

[0117] The application obtains a preset target field, then obtains target data corresponding to the target field from the dialogue text based on the target field, and subsequently constructs the directed graph based on the target data and the topic label. The use of the target field can quickly and accurately obtain matching target data from the dialogue text, and then the subsequent accurate construction of the directed graph can be completed based on the obtained target data and the topic label, which is beneficial to the subsequent processing of the constructed directed graph by using a similarity analysis model to automatically and intelligently generate a similarity result between each directed graph.

[0118] In some optional implementation manners of the embodiment, after step S205, the electronic device can further perform the following steps:

[0119] Obtaining the similarity result.

[0120] Generating a corresponding dialogue analysis report based on the similarity result.

[0121] In the embodiment, the dialogue analysis report can be generated by obtaining a preset analysis report template, and filling the similarity result into the analysis report template based on the data filling specification in the report template. The analysis report template is generated according to the actual business requirements.

[0122] Storing the dialogue analysis report.

[0123] In the embodiment, the storage manner of the dialogue analysis report is not specifically limited, for example, a database storage, cloud storage, blockchain storage, etc.

[0124] The application obtains the similar result, generates a corresponding dialogue analysis report based on the similar result, and stores the dialogue analysis report. By generating the dialogue analysis report by using the similar result and storing the dialogue analysis report, the subsequent agent can analyze the dialogue analysis report to help the agent end discover common dialogue paths from the voice dialogue data and adjust the reasonable speech to guide the user to form an optimal purchase path, thereby improving the work experience of the agent.

[0125] It should be emphasized that, in order to further ensure the privacy and security of the similar result, the similar result can also be stored in a node of a block chain.

[0126] The block chain referred to in the application is a new application mode of distributed data storage, peer-to-peer transmission, consensus mechanism, encryption algorithm and other computer technologies. The block chain is essentially a decentralized database, which is a series of data blocks associated using cryptographic methods, each data block containing information of a batch of network transactions, for verifying the validity (anti-fake) of the information and generating the next block. The block chain can include a block chain underlying platform, a platform product service layer, and an application service layer.

[0127] The embodiments of the application can acquire and process related data based on artificial intelligence technology. Artificial intelligence (AI) is the use of digital computers or computer-controlled machines to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0128] The basic technology of artificial intelligence generally includes technologies such as sensors, special artificial intelligence chips, cloud computing, distributed storage, big data processing technology, operation / interaction system, mechatronics, etc. The software technology of artificial intelligence mainly includes computer vision technology, robot technology, biometric technology, speech processing technology, natural language processing technology, and machine learning / deep learning, etc.

[0129] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be completed by computer readable instructions instructing related hardware, and the computer readable instructions can be stored in a computer readable storage medium. The program can include the processes of the above-mentioned embodiments when executed, wherein the storage medium can be a non-volatile storage medium such as a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

[0130] It should be understood that although each step in the flowchart of the accompanying drawings is shown in sequence according to the direction of the arrow, these steps are not necessarily executed in sequence according to the direction of the arrow. Unless otherwise explicitly stated herein, the execution of these steps is not strictly limited in sequence, and they can be executed in other sequences. Moreover, at least some of the steps in the flowchart of the accompanying drawings can include multiple sub-steps or multiple stages, which are not necessarily executed at the same time, but can be executed at different times, and the execution sequence is not necessarily sequential, but can be alternately or alternately executed with at least part of other steps or sub-steps or stages of other steps.

[0131] Further referring to Figure 3 , as an implementation of the method shown in the above Figure 2 , the present application provides an embodiment of an artificial intelligence-based data analysis device, which corresponds to the method embodiment shown in Figure 2 , and the device can be specifically applied to various electronic devices.

[0132] As shown in Figure 3 , the artificial intelligence-based data analysis device 300 described in the embodiment includes a first acquisition module 301, a conversion module 302, a classification module 303, a construction module 304, and a processing module 305. Among them:

[0133] The first acquisition module 301 is configured to acquire voice conversation data of a user and an agent in a business process.

[0134] The conversion module 302 is configured to convert the voice conversation data into text data, and convert the text data according to a preset information type to obtain corresponding conversation text.

[0135] The classification module 303 is configured to classify and process the conversation text according to a preset rule to obtain a topic label corresponding to the conversation text.

[0136] The construction module 304 is configured to construct a corresponding directed graph based on the conversation text and the topic label; wherein the number of the directed graphs includes a plurality.

[0137] The processing module 305 is configured to perform similarity analysis processing on the directed graphs based on a preset similarity analysis model to generate a similarity result between each of the directed graphs; wherein the similarity analysis model is composed of a graph convolution network, a tensor neural network, and a fully connected layer.

[0138] In the embodiment, the operations performed by the above-mentioned modules or units correspond one-to-one to the steps of the artificial intelligence-based data analysis method of the foregoing embodiments, and will not be described here.

[0139] In some optional implementations of the embodiment, the processing module 305 includes:

[0140] a first processing submodule, configured to perform reconstruction processing on the features of each node in each directed graph by using a graph convolution network in the similarity analysis model, to obtain a corresponding node representation vector;

[0141] a submodule, configured to perform encoding processing on each directed graph based on a preset attention mechanism, to obtain a graph feature vector of each directed graph;

[0142] a second processing submodule, configured to perform processing on the graph feature vector of each directed graph based on a tensor neural network in the similarity analysis model, to obtain a similarity vector between target directed graphs; the target directed graphs are any two of all the directed graphs;

[0143] a first generation submodule, configured to generate histogram features of the directed graphs based on the node representation vectors;

[0144] a third processing submodule, configured to perform processing on the similarity vector between the target directed graphs and target histogram features between the target directed graphs based on a full connection layer in the similarity analysis model, to obtain a target similarity result between the target directed graphs.

[0145] In the embodiment, the above modules or units are respectively used for operations corresponding to the steps of the artificial intelligence-based data analysis method of the foregoing embodiments, and thus will not be described here again.

[0146] In some optional implementations of the embodiment, the first generation submodule includes:

[0147] a calculation unit, configured to perform inner product calculation on the node representation vectors of the directed graphs, to obtain a corresponding correlation matrix;

[0148] a conversion unit, configured to perform conversion processing on the correlation matrix, to obtain histogram features corresponding to the directed graphs.

[0149] In the embodiment, the above modules or units are respectively used for operations corresponding to the steps of the artificial intelligence-based data analysis method of the foregoing embodiments, and thus will not be described here again.

[0150] In some optional implementations of the embodiment, the third processing submodule includes:

[0151] a processing unit, configured to perform splicing processing on the similarity vector between the target directed graphs and the target histogram features between the target directed graphs, to obtain a corresponding splicing feature;

[0152] The acquisition unit is configured to input the splicing feature into a full connection layer in the similarity analysis model, and acquire an output result corresponding to the target directed graph and output by the full connection.

[0153] The determination unit is configured to determine the output result as the target similarity result.

[0154] In the embodiment, the above modules or units are respectively used for performing operations corresponding to steps of the artificial intelligence-based data analysis method of the foregoing embodiments, and details are not described herein again.

[0155] In some optional implementations of the embodiment, the classification module 303 includes:

[0156] The second generation sub-module is configured to process the dialogue text based on a preset text representation model, and generate text representation data corresponding to the dialogue text.

[0157] The extraction sub-module is configured to perform feature extraction on the text representation data based on a preset feature extraction model, and obtain corresponding feature data.

[0158] The determination sub-module is configured to determine the feature data as the topic label.

[0159] In the embodiment, the above modules or units are respectively used for performing operations corresponding to steps of the artificial intelligence-based data analysis method of the foregoing embodiments, and details are not described herein again.

[0160] In some optional implementations of the embodiment, the construction module 304 includes:

[0161] The first acquisition sub-module is configured to acquire a preset target field.

[0162] The second acquisition sub-module is configured to acquire target data corresponding to the target field from the dialogue text based on the target field.

[0163] The construction sub-module is configured to construct the directed graph based on the target data and the topic label.

[0164] In the embodiment, the above modules or units are respectively used for performing operations corresponding to steps of the artificial intelligence-based data analysis method of the foregoing embodiments, and details are not described herein again.

[0165] In some optional implementations of the embodiment, the artificial intelligence-based data analysis apparatus further includes:

[0166] The second acquisition module is configured to acquire the similarity result.

[0167] The generation module is configured to generate a corresponding dialogue analysis report based on the similarity result.

[0168] a storage module configured to store the dialogue analysis report.

[0169] In the embodiments, the modules or units described above are respectively configured to perform the operations corresponding to the steps of the artificial intelligence-based data analysis method of the foregoing embodiments, and thus no further description is provided here.

[0170] To solve the above technical problems, the embodiments of the present application further provide a computer device. For details, please refer to Figure 4 , Figure 4 The basic structure block diagram of the computer device of the present embodiment is shown in FIG. 4.

[0171] The computer device 4 includes a memory 41, a processor 42, and a network interface 43, which are connected to each other through a system bus. It should be noted that only the computer device 4 with components 41-43 is shown in the figure, but it should be understood that it is not required to implement all the shown components, and more or fewer components can be alternatively implemented. Among them, those skilled in the art can understand that the computer device herein is a device capable of automatically performing numerical calculation and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to microprocessors, application specific integrated circuits (ASICs), field programmable gate arrays (FPGAs), digital signal processors (DSPs), embedded devices, etc.

[0172] The computer device can be a desktop computer, a notebook computer, a palm computer, a cloud server, and the like. The computer device can interact with the user through a keyboard, a mouse, a remote controller, a touchpad, a voice control device, and the like.

[0173] The memory 41 includes at least one type of readable storage medium, such as a flash memory, a hard disk, a multimedia card, a card-type memory (e.g., an SD or DX memory, etc.), a random access memory (RAM), a static random access memory (SRAM), a read-only memory (ROM), an electrically erasable programmable read-only memory (EEPROM), a programmable read-only memory (PROM), a magnetic memory, a magnetic disk, an optical disk, etc. In some embodiments, the memory 41 can be an internal storage unit of the computer device 4, such as a hard disk or a memory of the computer device 4. In other embodiments, the memory 41 can also be an external storage device of the computer device 4, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the computer device 4. Of course, the memory 41 can also include both the internal storage unit and the external storage device of the computer device 4. In this embodiment, the memory 41 is generally used to store an operating system and various application software installed on the computer device 4, such as computer readable instructions of the artificial intelligence-based data analysis method, etc. In addition, the memory 41 can also be used to temporarily store various data that have been output or will be output.

[0174] The processor 42 can be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chip in some embodiments. The processor 42 is generally used to control the overall operation of the computer device 4. In this embodiment, the processor 42 is used to run computer readable instructions or process data stored in the memory 41, such as computer readable instructions of the artificial intelligence-based data analysis method.

[0175] The network interface 43 can include a wireless network interface or a wired network interface, and is generally used to establish a communication connection between the computer device 4 and other electronic devices.

[0176] Compared with the prior art, the embodiments of the present application have the following beneficial effects:

[0177] In the embodiments of the present application, first, voice dialogue data of a user and an agent in a business process is acquired; then the voice dialogue data is converted into text data, and the text data is converted according to a preset information type to obtain corresponding dialogue text; then the dialogue text is classified according to a preset rule to obtain a theme label corresponding to the dialogue text; subsequently, a corresponding directed graph is constructed based on the dialogue text and the theme label; finally, a similarity analysis model is used to perform similarity analysis on the directed graph to generate a similarity result between each directed graph. The embodiments of the present application complete the construction of a directed graph based on dialogue text and theme labels of the dialogue text of a user and an agent in a business process, and then use a similarity analysis model to perform similarity analysis on the constructed directed graph, which can quickly and accurately generate a similarity result between directed graphs, improve the generation efficiency of the similarity result, and ensure the data accuracy of the similarity result. The present application analyzes the similarity relationship of the theme path of the dialogue technique based on the similarity analysis model to obtain the corresponding similarity result, which is beneficial to helping the agent side discover common dialogue paths from voice dialogue data in the subsequent process, so that reasonable dialogue techniques can be adjusted in time based on the similarity logic of the common dialogue path to guide the user to form an optimal purchase path, which is beneficial to improving the work experience of the agent.

[0178] The present application also provides another embodiment, that is, a computer readable storage medium storing computer readable instructions, which can be executed by at least one processor to make the at least one processor execute the steps of the artificial intelligence-based data analysis method as described above.

[0179] Compared with the prior art, the embodiments of the present application have the following beneficial effects:

[0180] In the embodiments of the present application, firstly, voice dialogue data of a user and an agent in a business process is acquired; then the voice dialogue data is converted into text data, and the text data is converted according to a preset information type to obtain corresponding dialogue text; then the dialogue text is classified and processed according to a preset rule to obtain a theme label corresponding to the dialogue text; subsequently, a corresponding directed graph is constructed based on the dialogue text and the theme label; finally, a similarity analysis model is used to perform similarity analysis processing on the directed graph to generate a similarity result between the directed graphs. In the embodiments of the present application, the dialogue text of the user and the agent in the business process and the theme label of the dialogue text are used to complete the construction of the directed graph, and then the similarity analysis model is used to perform similarity analysis processing on the constructed directed graph, so that the similarity result between the directed graphs can be quickly and accurately generated, the generation efficiency of the similarity result is improved, and the data accuracy of the similarity result is ensured. The similarity analysis model is used to analyze and research the similarity relationship of the theme path of the dialogue skill to obtain the corresponding similarity result, which is beneficial to helping the agent end to discover a common dialogue path from the voice dialogue data, so that reasonable skill adjustment can be made in time based on the similarity logic of the common dialogue path to guide the user to form an optimal purchase path, and the work experience of the agent is improved.

[0181] Through the above description of the embodiments, those skilled in the art can clearly understand that the above-mentioned embodiment method can be realized by means of software and a general hardware platform, of course, it can also be realized by hardware, but in many cases, the former is a better embodiment. Based on such understanding, the technical solutions of the present application can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a plurality of instructions for making a terminal device (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) execute the methods described in the embodiments of the present application.

[0182] Obviously, the above-described embodiments are only some of the embodiments of the present application, not all the embodiments, and the preferred embodiments of the present application are given in the drawings, but do not limit the patent scope of the present application. The present application can be realized in many different forms, and conversely, the purpose of providing these embodiments is to make the disclosure of the present application more thorough and comprehensive. Although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art can modify the technical solutions recorded in the foregoing specific embodiments, or make equivalent replacements to some technical features. Any equivalent structure made by using the contents of the specification and drawings, directly or indirectly applied to other related technical fields, is also within the scope of the patent protection of the present application.

Claims

1. An artificial intelligence-based data analysis method, characterized by, The method comprises the following steps: obtaining voice conversation data of a user and an agent in a business conversation process; converting the voice conversation data into text data, and converting the text data according to a preset information type to obtain corresponding conversation text; classifying the conversation text according to a preset rule to obtain a topic label corresponding to the conversation text; constructing a corresponding directed graph based on the conversation text and the topic label; wherein the number of the directed graphs comprises a plurality of directed graphs; performing similarity analysis on the directed graphs based on a preset similarity analysis model to generate a similarity result between each of the directed graphs; wherein the similarity analysis model comprises a graph convolution network, a tensor neural network and a fully connected layer; wherein the step of performing similarity analysis on the directed graphs based on the preset similarity analysis model to generate a similarity result between each of the directed graphs comprises: reconstructing features of each node in each of the directed graphs through the graph convolution network in the similarity analysis model to obtain a corresponding node representation vector; encoding each of the directed graphs based on a preset attention mechanism to obtain a graph feature vector of each of the directed graphs; processing the graph feature vectors of each of the directed graphs based on the tensor neural network in the similarity analysis model to obtain a similarity vector between target directed graphs; wherein the target directed graphs are any two of all the directed graphs; generating a histogram feature of the directed graph based on the node representation vector; processing the similarity vector between the target directed graphs and a target histogram feature between the target directed graphs based on the fully connected layer in the similarity analysis model to obtain a target similarity result between the target directed graphs. 2.The artificial intelligence-based data analysis method of claim 1, wherein, The step of generating a histogram feature of the directed graph based on the node representation vector comprises: performing inner product calculation on the node representation vector of the directed graph to obtain a corresponding correlation matrix; performing conversion processing on the correlation matrix to obtain a histogram feature corresponding to the directed graph. 3.The artificial intelligence-based data analysis method of claim 1, wherein, The step of processing the similarity vector between the target directed graphs and the target histogram feature between the target directed graphs based on the fully connected layer in the similarity analysis model to obtain a target similarity result between the target directed graphs comprises: performing splicing processing on the similarity vector between the target directed graphs and the target histogram feature between the target directed graphs to obtain a corresponding splicing feature; inputting the splicing feature into the fully connected layer in the similarity analysis model to obtain an output result corresponding to the target directed graphs output by the fully connected layer; taking the output result as the target similarity result. 4.The artificial intelligence-based data analysis method of claim 1, wherein, The step of classifying the conversation text according to a preset rule to obtain a topic label corresponding to the conversation text comprises: processing the conversation text based on a preset text representation model to generate text representation data of the conversation text; extracting features of the text representation data based on a preset feature extraction model to obtain corresponding feature data; taking the feature data as the topic label. 5.The artificial intelligence-based data analysis method of claim 1, wherein, The step of constructing the corresponding directed graph based on the dialogue text and the topic label specifically comprises: acquiring a preset target field; acquiring target data corresponding to the target field from the dialogue text based on the target field; constructing the directed graph based on the target data and the topic label. 6.The artificial intelligence-based data analysis method of claim 1, wherein, After the step of performing similarity analysis on the directed graph based on the preset similarity analysis model to generate a similarity result between each directed graph, the method further comprises: acquiring the similarity result; generating a corresponding dialogue analysis report based on the similarity result; storing the dialogue analysis report. 7.A data analysis apparatus based on artificial intelligence, which executes the data analysis method based on artificial intelligence according to any one of claims 1 to 6 at the time of operation, characterized by The artificial intelligence-based data analysis device comprises: a first acquisition module configured to acquire voice dialogue data of a user and an agent in a business process; a conversion module configured to convert the voice dialogue data into text data and convert the text data according to a preset information type to obtain corresponding dialogue text; a classification module configured to classify the dialogue text according to a preset rule to obtain a topic label corresponding to the dialogue text; a construction module configured to construct a corresponding directed graph based on the dialogue text and the topic label; wherein the number of the directed graphs comprises a plurality of directed graphs; a processing module configured to perform similarity analysis on the directed graph based on a preset similarity analysis model to generate a similarity result between each directed graph; wherein the similarity analysis model is composed of a graph convolution network, a tensor neural network, and a full connection layer.

8. A computer device comprising a memory and a processor, wherein the memory stores computer readable instructions, and the processor executes the computer readable instructions to implement the steps of the artificial intelligence-based data analysis method according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer readable storage medium stores computer readable instructions, and the computer readable instructions are executed by the processor to implement the steps of the artificial intelligence-based data analysis method according to any one of claims 1 to 6.

Images