Incoming call customer intention analysis system and method

Abstract

According to the incoming call customer intention analysis system and method, voice signals of incoming call customers are collected, the intentions of the incoming call customers are analyzed and judged through the voice recognition technology, automatic recognition and analysis of the intentions of the incoming call customers are achieved, the incoming call reception efficiency and the service quality are improved, and the user experience is improved. And meanwhile, better service experience is provided for customers.

Description

Technical Field

[0001] The present invention relates to the technical field of caller intention analysis, and in particular to a caller intention analysis system and method. Background Art

[0002] The extensive development in the past is no longer sustainable. Traditional manufacturing retail is facing problems such as low traffic, low stickiness, difficulty in retention, high costs, and a poor transaction rate. The existing business development and follow-up process is that the salesperson visits the customer on site or communicates with the customer online, and then waits for the interested customer to contact the salesperson for consultation or further negotiation. In the actual follow-up process, it is found that it is easy for the salesperson to lose contact with the customer after the inquiry communication with the customer, and it is impossible to judge the customer's intention for the business. They can only wait for the customer to contact them actively, and even lose the interested customer; and when the salesperson takes the initiative to contact the customer, it is difficult to choose the right time and grasp the customer's interest and intention. Summary of the invention

[0003] The purpose of the present invention is to provide a system and method based on caller intention analysis to solve the problems raised in the background technology and facilitate promotion.

[0004] In order to achieve the above-mentioned object of the invention, the technical solution adopted by the present invention is:

[0005] A caller intention analysis system, comprising a caller voice recording module, a voice recognition module, an intention analysis module and a command execution module, wherein each module interacts with each other based on digital signals;

[0006] The caller voice recording module is responsible for recording and storing the caller voice signal;

[0007] The speech recognition module is responsible for performing speech recognition on the voice signal of the incoming caller and converting it into text data;

[0008] The intention analysis module analyzes and judges the text data converted from speech through natural language processing technology and machine learning algorithms to judge the specific intentions and needs of the calling customer, and the machine learning algorithms are based on convolutional neural networks or recursive neural networks;

[0009] The command execution module performs different operations according to the needs and intentions of the calling customer, including connecting to a designated customer service staff and playing a pre-recorded voice answer.

[0010] Furthermore, the caller voice recording module includes a recording device module, a recording control software, a recording storage system, a recording quality detection and preprocessing module, and a recording index and classification management module. The recording device module is used to collect the voice signal of the caller. The recording control software is used to control the recording device module, including starting and stopping recording, adjusting the recording volume, and setting the recording format and parameters. The recording storage system includes a local memory and a cloud storage device for storing recorded voice signals. The recording quality detection and preprocessing module is used to detect the quality of the incoming call voice and use denoising, noise reduction, signal amplification, and voice compensation technology to preprocess the recording data. The recording index and classification management module includes an index and classification management system.

[0011] Furthermore, the speech recognition module includes a speech signal input module, a front-end processing module, a feature extraction module, a signal model module, a language model module, a vocabulary and speech category and a back-end processing module. The speech signal input module inputs the recorded speech signal into the speech recognition system. The front-end processing module pre-processes the input speech signal to improve the accuracy and robustness of subsequent speech recognition. The front-end processing includes signal segmentation, noise suppression, speech enhancement, speech detection, and speech endpoint detection. The feature extraction module includes feature extraction methods such as MFCC, PLP, and LPC. The signal model module establishes an acoustic model in the speech recognition system to describe the characteristics of the speech signal. The language model module processes text data and gives the probability of a word appearing in a certain context to help the recognition system perform speech analysis and command recognition. The vocabulary and speech category module establish a vocabulary corresponding to the speech signal to achieve mapping from the phoneme sequence predicted and output by the acoustic model to the corresponding word sequence, thereby improving the recognition accuracy and efficiency of the system. After recognizing the speech command or text in text form, the back-end processing module needs to perform subsequent processing on it, such as semantic understanding, command analysis, and execution operation.

[0012] Furthermore, the intention analysis module includes a classification algorithm module, a feature extraction module, a semantic understanding module, a knowledge graph module, a context analysis module and a dialogue strategy module. The classification algorithm module classifies the text input by the user to determine the user's intention, including rules, decision trees, Bayesian classifiers, support vector machines, and deep learning networks. In the intention analysis module, the feature extraction module needs to extract features from the text input by the user to help the classifier make correct decisions. Common features include text keywords, phrases, intonation, and emotions. The semantic understanding module performs in-depth semantic analysis on the input text, including natural language processing, word vector representation, and topic model. The knowledge graph module helps the machine better understand the user's intention. The context analysis module helps the machine better understand the user's intention and generate more accurate answers based on the context, including dialogue state tracking and dialogue history management. The dialogue strategy module includes dialogue generation, planning, and updating.

[0013] A method for analyzing the intention of incoming call customers comprises the following steps:

[0014] a. Record and store the voice signals of incoming customers through the recording module;

[0015] b. Convert the voice signal of the calling customer into text data through the voice recognition module;

[0016] c. The intention analysis module uses natural language processing technology and machine learning algorithms to analyze and judge the text data converted from voice to determine the specific intentions and needs of the calling customer;

[0017] d. Perform corresponding operations based on the needs and intentions of the calling customer.

[0018] As an improvement, the beneficial effects of the present invention are:

[0019] The present invention provides a system and method based on caller intention analysis, which collects the voice signals of callers and uses voice recognition technology to analyze and judge their intentions, thereby realizing automatic identification and analysis of caller intentions, improving call reception efficiency and service quality, and providing customers with a better service experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 A flow chart of a system and method for analyzing calling customer intentions according to the present invention; DETAILED DESCRIPTION

[0021] In order to make the contents of the present invention more clearly understood, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0022] A caller intention analysis system, comprising a caller voice recording module, a voice recognition module, an intention analysis module and a command execution module, wherein each module interacts with each other based on digital signals;

[0023] The caller voice recording module is responsible for recording and storing the caller voice signal;

[0024] The speech recognition module is responsible for voice recognition of incoming customer voice signals and converting them into text data;

[0025] The intention analysis module uses natural language processing technology and machine learning algorithms to analyze and judge the text data converted from speech to determine the specific intentions and needs of the calling customer. The machine learning algorithm is based on convolutional neural networks or recursive neural networks.

[0026] The command execution module performs different operations according to the needs and intentions of the calling customer, including connecting to the designated customer service staff and playing pre-recorded voice answers.

[0027] The caller voice recording module includes a recording device module, a recording control software, a recording storage system, a recording quality detection and preprocessing module, and a recording index and classification management module. The recording device module is used to collect the voice signal of the caller. The recording control software is used to control the recording device module, including starting and stopping recording, adjusting the recording volume, and setting the recording format and parameters. The recording storage system includes a local storage device and a cloud storage device for storing recorded voice signals. The recording quality detection and preprocessing module is used to detect the quality of the caller voice and use denoising, noise reduction, signal amplification, and voice compensation technology to preprocess the recording data. The recording index and classification management module includes an index and classification management system.

[0028] The speech recognition module includes a speech signal input module, a front-end processing module, a feature extraction module, a signal model module, a language model module, a vocabulary and speech category and a back-end processing module. The speech signal input module inputs the recorded speech signal into the speech recognition system. The front-end processing module pre-processes the input speech signal to improve the accuracy and robustness of subsequent speech recognition. The front-end processing includes signal segmentation, noise suppression, speech enhancement, speech detection, and speech endpoint detection. The feature extraction module includes MFCC, PLP, LPC and other feature extraction methods. In the speech recognition system, the signal model module establishes an acoustic model to describe the characteristics of the speech signal. The language model module processes text data and gives the probability of a word appearing in a certain context to help the recognition system perform speech analysis and command recognition. The vocabulary and speech category module establish a vocabulary corresponding to the speech signal to achieve the mapping from the phoneme sequence predicted and output by the acoustic model to the corresponding word sequence, thereby improving the recognition accuracy and efficiency of the system. After recognizing the speech command or text in text form, the back-end processing module needs to perform subsequent processing on it, such as semantic understanding, command analysis, and execution operations.

[0029] The intention analysis module includes a classification algorithm module, a feature extraction module, a semantic understanding module, a knowledge graph module, a context analysis module and a dialogue strategy module. The classification algorithm module classifies the text input by the user to determine the user's intention, including rules, decision trees, Bayesian classifiers, support vector machines, and deep learning networks. In the intention analysis module, the feature extraction module needs to extract features from the text input by the user to help the classifier make correct decisions. Common features include text keywords, phrases, intonation, and emotions. The semantic understanding module performs in-depth semantic analysis of the input text, including natural language processing, word vector representation, and topic model. The knowledge graph module helps the machine better understand the user's intention. The context analysis module helps the machine better understand the user's intention and generate more accurate answers based on the context, including dialogue state tracking and dialogue history management. The dialogue strategy module includes dialogue generation, planning, and updating.

[0030] A method for analyzing the intention of incoming call customers comprises the following steps:

[0031] a. Record and store the voice signals of incoming customers through the recording module;

[0032] b. Convert the voice signal of the calling customer into text data through the voice recognition module;

[0033] c. The intention analysis module uses natural language processing technology and machine learning algorithms to analyze and judge the text data converted from voice to determine the specific intentions and needs of the calling customer;

[0034] d. Perform corresponding operations based on the needs and intentions of the calling customer.

[0035] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention shall be included in the protection scope of the present invention.

Claims

1. A caller intention analysis system, including a caller voice recording module, a voice recognition module, an intention analysis module and a command execution module, wherein each module interacts with each other based on digital signals; The caller voice recording module is responsible for recording and storing the caller voice signal; The speech recognition module is responsible for performing speech recognition on the voice signal of the incoming caller and converting it into text data; The intention analysis module analyzes and judges the text data converted from speech through natural language processing technology and machine learning algorithms to judge the specific intentions and needs of the calling customer, and the machine learning algorithms are based on convolutional neural networks or recursive neural networks; The command execution module performs different operations according to the needs and intentions of the calling customer, including connecting to a designated customer service staff and playing a pre-recorded voice answer.

2. A system based on calling customer intention analysis according to claim 1, characterized in that: The incoming call voice recording module includes a recording device module, a recording control software, a recording storage system, a recording quality detection and preprocessing module, and a recording index and classification management module. The recording device module is used to collect the voice signal of the incoming caller. The recording control software is used to control the recording device module, including starting and stopping recording, adjusting the recording volume, and setting the recording format and parameters. The recording storage system includes a local memory and a cloud memory for storing recorded voice signals. The recording quality detection and preprocessing module is used to detect the quality of the incoming call voice and use denoising, noise reduction, signal amplification, and voice compensation technology to preprocess the recording data. The recording index and classification management module includes an index and classification management system.

3. The system based on calling customer intention analysis according to claim 1, characterized in that: The speech recognition module includes a speech signal input module, a front-end processing module, a feature extraction module, a signal model module, a language model module, a vocabulary, a speech category and a back-end processing module. The speech signal input module inputs the recorded speech signal into the speech recognition system. The front-end processing module pre-processes the input speech signal to improve the accuracy and robustness of subsequent speech recognition. The front-end processing includes signal segmentation, noise suppression, speech enhancement, speech detection, and speech endpoint detection. The feature extraction module includes feature extraction methods such as MFCC, PLP, and LPC. The signal model module establishes an acoustic model in the speech recognition system to describe the characteristics of the speech signal. The language model module processes text data and gives the probability of a certain word appearing in a certain context to help the recognition system perform speech analysis and command recognition. The vocabulary and speech category module establish a vocabulary corresponding to the speech signal to achieve mapping from the phoneme sequence predicted and output by the acoustic model to the corresponding word sequence, thereby improving the recognition accuracy and efficiency of the system. After recognizing the speech command or text in text form, the back-end processing module needs to perform subsequent processing on it, such as semantic understanding, command analysis, and execution operation.

4. The system for analyzing the intention of incoming callers according to claim 1, characterized in that: The intention analysis module includes a classification algorithm module, a feature extraction module, a semantic understanding module, a knowledge graph module, a context analysis module and a dialogue strategy module. The classification algorithm module classifies the text input by the user to determine the user's intention, including rules, decision trees, Bayesian classifiers, support vector machines, and deep learning networks. In the intention analysis module, the feature extraction module needs to extract features from the text input by the user to help the classifier make correct decisions. Common features include text keywords, phrases, intonation, and emotions. The semantic understanding module performs in-depth semantic analysis on the input text, including natural language processing, word vector representation, and topic model. The knowledge graph module helps the machine better understand the user's intention. The context analysis module helps the machine better understand the user's intention and generate more accurate answers based on the context, including dialogue state tracking and dialogue history management. The dialogue strategy module includes dialogue generation, planning, and updating.

5. A method for analyzing the intention of incoming callers, comprising the following steps: a. Record and store the voice signals of incoming customers through the recording module; b. Convert the voice signal of the calling customer into text data through the voice recognition module; c. The intention analysis module uses natural language processing technology and machine learning algorithms to analyze and judge the text data converted from voice to determine the specific intentions and needs of the calling customer; d. Perform corresponding operations based on the needs and intentions of the calling customer.

Images