CN111182162B - Telephone quality inspection method, device, equipment and storage medium based on artificial intelligence - Google Patents
Telephone quality inspection method, device, equipment and storage medium based on artificial intelligence Download PDFInfo
- Publication number
- CN111182162B CN111182162B CN201911367527.7A CN201911367527A CN111182162B CN 111182162 B CN111182162 B CN 111182162B CN 201911367527 A CN201911367527 A CN 201911367527A CN 111182162 B CN111182162 B CN 111182162B
- Authority
- CN
- China
- Prior art keywords
- processed
- neural network
- call
- input
- call information
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 63
- 238000007689 inspection Methods 0.000 title claims abstract description 48
- 238000013473 artificial intelligence Methods 0.000 title claims abstract description 34
- 238000013528 artificial neural network Methods 0.000 claims abstract description 83
- 238000005516 engineering process Methods 0.000 claims abstract description 44
- 238000003062 neural network model Methods 0.000 claims abstract description 32
- 230000011218 segmentation Effects 0.000 claims abstract description 17
- 238000012549 training Methods 0.000 claims description 29
- 210000002569 neuron Anatomy 0.000 claims description 25
- 230000006870 function Effects 0.000 claims description 23
- 238000012545 processing Methods 0.000 claims description 21
- 238000003058 natural language processing Methods 0.000 claims description 17
- 238000004364 calculation method Methods 0.000 claims description 14
- 230000001537 neural effect Effects 0.000 claims description 6
- 238000012546 transfer Methods 0.000 claims description 6
- 238000004422 calculation algorithm Methods 0.000 claims description 5
- 238000011156 evaluation Methods 0.000 claims description 5
- 230000008569 process Effects 0.000 description 14
- 238000013461 design Methods 0.000 description 12
- 238000004590 computer program Methods 0.000 description 7
- 238000004458 analytical method Methods 0.000 description 5
- 238000012372 quality testing Methods 0.000 description 4
- 238000005070 sampling Methods 0.000 description 4
- 238000006243 chemical reaction Methods 0.000 description 3
- 238000007726 management method Methods 0.000 description 3
- 238000012544 monitoring process Methods 0.000 description 3
- 230000004913 activation Effects 0.000 description 2
- 125000004122 cyclic group Chemical group 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000012886 linear function Methods 0.000 description 2
- 238000007477 logistic regression Methods 0.000 description 2
- 238000007781 pre-processing Methods 0.000 description 2
- 238000000638 solvent extraction Methods 0.000 description 2
- 238000012550 audit Methods 0.000 description 1
- 230000001413 cellular effect Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000006835 compression Effects 0.000 description 1
- 238000007906 compression Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000004806 packaging method and process Methods 0.000 description 1
- 238000012797 qualification Methods 0.000 description 1
- 238000001303 quality assessment method Methods 0.000 description 1
- 230000000306 recurrent effect Effects 0.000 description 1
- 238000000714 time series forecasting Methods 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04M—TELEPHONIC COMMUNICATION
- H04M3/00—Automatic or semi-automatic exchanges
- H04M3/42—Systems providing special services or facilities to subscribers
- H04M3/50—Centralised arrangements for answering calls; Centralised arrangements for recording messages for absent or busy subscribers ; Centralised arrangements for recording messages
- H04M3/51—Centralised call answering arrangements requiring operator intervention, e.g. call or contact centers for telemarketing
- H04M3/5175—Call or contact centers supervision arrangements
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/06—Creation of reference templates; Training of speech recognition systems, e.g. adaptation to the characteristics of the speaker's voice
- G10L15/063—Training
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/08—Speech classification or search
- G10L15/16—Speech classification or search using artificial neural networks
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
- G10L25/48—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 specially adapted for particular use
- G10L25/51—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 specially adapted for particular use for comparison or discrimination
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04M—TELEPHONIC COMMUNICATION
- H04M3/00—Automatic or semi-automatic exchanges
- H04M3/42—Systems providing special services or facilities to subscribers
- H04M3/42221—Conversation recording systems
Landscapes
- Engineering & Computer Science (AREA)
- Human Computer Interaction (AREA)
- Acoustics & Sound (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Signal Processing (AREA)
- Physics & Mathematics (AREA)
- Multimedia (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Business, Economics & Management (AREA)
- Marketing (AREA)
- Telephonic Communication Services (AREA)
- Machine Translation (AREA)
Abstract
The invention relates to the field of artificial intelligence and provides a telephone quality inspection method, device, equipment and storage medium based on artificial intelligence. The method comprises the following steps: acquiring a plurality of call records to be processed; marking a corresponding score for each call record to be processed; converting a plurality of call records to be processed into a plurality of text messages to be processed through a preset sequence neural network model; word segmentation is carried out on text information to be processed through a language technology platform LTP technology, so that a plurality of core keywords and frequencies corresponding to the core keywords are obtained; taking the frequency of a core keyword as input x i The marked call score is taken as ideal output y i Inputting the target neural network into the neural network to obtain the target neural network; deploying the target neural network to the client; receiving call information input by a user, calculating to obtain a score of the call information through an ideal weight, and returning the score of the call information to the user. Improving the efficiency of telephone quality inspection.
Description
Technical Field
The present invention relates to the field of monitoring, and in particular, to a method, apparatus, device, and storage medium for quality inspection of a phone based on artificial intelligence.
Background
Various index data such as call completing rate, call duration, complaint rate, customer satisfaction and the like are very important in the operation of a call center. The data directly corresponds to the business quality gap among different personnel, and at present, a lot of monitoring and quality inspection are performed in a manual sampling inspection mode, the quality inspector is responsible for quality monitoring, flow management, index management, site follow-up, report writing and cultural construction, and is also responsible for quality inspection of a plurality of links such as knowledge base management, business training, employee feedback guidance and the like, so that time and effort are consumed, only about 30% of telephone traffic quality inspection can be completed, and the efficiency is low. The standards can be the same, the index settings can be the same, the results can be different, the key is that the purposes of using the standards by quality inspection personnel are different, the methods are different, the thought is different, and the accuracy of the quality inspection results is lower.
Disclosure of Invention
The invention provides a telephone quality inspection method based on artificial intelligence, which improves the efficiency of telephone quality inspection.
In a first aspect, the present invention provides a phone quality inspection method based on artificial intelligence, including:
acquiring a plurality of call records to be processed;
marking a corresponding score for each call record to be processed;
Converting the plurality of call records to be processed into a plurality of text messages to be processed through a preset sequence neural network model;
word segmentation is carried out on the text information to be processed through a language technology platform LTP technology, so that a plurality of core keywords and frequencies corresponding to the core keywords are obtained;
taking the frequency corresponding to the core keyword as input x i The score is noted as an ideal output y i Inputting the neural values into an initial neural network, and training the weights of the neurons in the initial neural network through a loss function to obtain a target neural network;
deploying the target neural network to a client;
receiving call information input by a user, and inputting the call information to the target neural network through the client;
and calculating the score of the call information through the neuron weight in the target neural network, and returning the score of the call information to the user.
In some possible designs, before the text information to be processed is segmented by the language technology platform LTP technology to obtain a plurality of core keywords and frequencies corresponding to the core keywords, the method further includes:
generating a unique identification ID number for each text message to be processed;
Constructing a hypertext transfer protocol (HTTP) request according to the API parameters of the text message to be processed;
extracting Chinese segmentation from text information to be processed by using an HTTP request and LTP technology;
and matching the words recorded in the database, and deleting the words with the matching degree lower than the threshold value to obtain words with the matching degree higher than the threshold value.
In some possible designs, the converting the plurality of call recordings to be processed into the plurality of text messages to be processed through the preset sequential neural network model includes:
acquiring a plurality of training voice samples;
inputting the training voice sample into the preset sequence neural network model, and updating the neuron weight of the preset sequence neural network model through a natural language processing NLP algorithm, the voice information and a text label corresponding to the voice information to obtain a target model;
adjusting the weight of the neuron of the target model, and updating the preset sequence neural network model;
and converting the plurality of call records to be processed into a plurality of text messages to be processed through the updated preset sequence neural network model.
In some possible designs, receiving call information input by a user, inputting the call information to a target neural network through a client, including:
When prompting a user to input call information, starting a microphone application through the client;
after receiving the indication message, prompting a user to input call information through the client;
when call information is obtained, noise is removed from the call information, background sound is removed, and the call information is compressed through the client side, so that the call information after preprocessing is obtained;
judging whether the call information subjected to pretreatment accords with an input preset rule or not through a client;
if the rule does not accord with the preset rule, commanding the client to prompt the user to input again;
if the preset rule is met, uploading the call information which is input by the user and is subjected to pretreatment to a server through the client;
and inputting the pretreated call information into a target neural network.
In some possible designs, before obtaining the plurality of call records to be processed, the method further includes:
rejecting the call record to be processed with the call duration smaller than a threshold value;
adding a plurality of characteristic items for the call record to be processed, wherein the characteristic items at least comprise: whether to make a first call, industry record average duration, and satisfaction of user phone evaluation.
In some possible designs, word segmentation is performed on text information to be processed through a language technology platform LTP technology to obtain a plurality of core keywords and frequencies corresponding to the core keywords, including:
The frequency corresponding to the core keyword passes
Calculation of F w The frequency of the core keywords is indicated, N is the number of sentences in which the core keywords appear, and M is the number of sentences of the text information to be processed.
In some possible designs, the method uses the frequency corresponding to the core keyword as input x i The score is noted as an ideal output y i Before inputting the text information to be processed into an initial neural network and training the neuron weight in the initial neural network through a loss function to obtain a target neural network, the method further comprises the steps of:
if the output y of the neural network j With n valid feature inputs x 1 ，x 2 ，…，x n The weight value of the corresponding connection is w 1 ，w 2 ，…，w n The neural network passes through
And initializing the weight.
In a second aspect, the present invention provides an artificial intelligence based phone quality inspection device having a function of implementing a method corresponding to the artificial intelligence based phone quality inspection platform provided in the first aspect. The functions may be implemented by hardware, or may be implemented by hardware executing corresponding software. The hardware or software includes one or more modules corresponding to the functions described above, which may be software and/or hardware.
The phone quality inspection device based on artificial intelligence comprises:
the input/output module is used for acquiring a plurality of call records to be processed;
the processing module is used for marking corresponding scores for each call record to be processed; converting the plurality of call records to be processed into a plurality of text messages to be processed through a preset sequence neural network model; word segmentation is carried out on the text information to be processed through a language technology platform LTP technology, so that a plurality of core keywords and frequencies corresponding to the core keywords are obtained; the frequency corresponding to the core keyword is used as input x through the input/output module i The score is noted as an ideal output y i Inputting the neural values into an initial neural network, and training the weights of the neurons in the initial neural network through a loss function to obtain a target neural network; deploying the target neural network to a client; receiving call information input by a user through the input/output module, and inputting the call information to the target neural network through the client; and calculating the score of the call information through the neuron weight in the target neural network, and returning the score of the call information to the user.
In some possible designs, the processing module is further to:
generating a unique identification ID number for each piece of text information to be processed;
constructing a hypertext transfer protocol (HTTP) request according to the API parameters of the text information to be processed;
extracting Chinese segmentation from text information to be processed by the HTTP request and the LTP technology;
and matching the words recorded in the database, and deleting the words with the matching degree lower than the threshold value to obtain words with the matching degree higher than the threshold value.
In some possible designs, the processing module is further to:
acquiring a plurality of training voice samples;
inputting the training voice sample into the preset sequence neural network model, and updating the neuron weight of the preset sequence neural network model through a natural language processing NLP technology, the voice information and a text label corresponding to the voice information to obtain a target model;
adjusting the weight of the neuron of the target model, and updating the preset sequence neural network model;
and converting the plurality of call records to be processed into a plurality of text messages to be processed through the updated preset sequence neural network model.
In some possible designs, the processing module is further to:
when prompting a user to input the call information, starting a microphone application through the client;
after receiving the indication message, prompting a user to input call information through the client;
when call information is obtained, noise is removed from the call information, background sound is removed, and the call information is compressed through the client side, so that the call information which is preprocessed is obtained;
judging whether the pretreated call information accords with an input preset rule or not through the client;
if the rule does not accord with the preset rule, the client is instructed to prompt the user to input again;
if the pre-processed call information meets the preset rule, uploading the pre-processed call information input by the user to a server through the client;
and inputting the pretreated call information to the target neural network.
In some possible designs, the processing module is further to:
rejecting the call record to be processed with the call duration smaller than a threshold value;
adding a plurality of characteristic items for the call record to be processed, wherein the characteristic items at least comprise: whether to make a first call, industry record average duration, and satisfaction of user phone evaluation.
In some possible designs, the processing module is further to:
the frequency corresponding to the core keyword passes through
Calculation of F w The frequency of the core keywords is indicated, N is the number of sentences in which the core keywords appear, and M is the number of sentences in the text information to be processed.
In some possible designs, the processing module is further to:
if the output y of the neural network j With n of said valid characteristic inputs x 1 ，x 2 ，…，x n The weight value of the corresponding connection is w 1 ，w 2 ，…，w n The neural network passes through
And initializing the weight.
In yet another aspect, the present invention provides an artificial intelligence based phone quality inspection device, which includes at least one connected processor, a memory, and an input/output unit, where the memory is configured to store program code, and the processor is configured to invoke the program code in the memory to perform the method described in the above aspects.
In yet another aspect, the invention provides a computer storage medium comprising instructions which, when run on a computer, cause the computer to perform the method of the above aspects.
According to the invention, the full quantity of call voice is converted into the text through the voice conversion technology, then the text is converted into the key core words through the LTP technology, the key core words are obtained, the key core words are counted, the score of the call record is obtained through inputting the counted key core words into the neural network, the rule is detected more objectively through intelligent quality inspection, the quality inspection is more accurate, the time cost is effectively saved by the intelligent quality inspection system, and the configuration of optimized personnel is completed.
Drawings
FIG. 1 is a flow chart of a phone quality inspection method based on artificial intelligence in an embodiment of the invention;
FIG. 2 is a schematic diagram of a phone quality inspection device based on artificial intelligence in an embodiment of the invention;
fig. 3 is a schematic structural diagram of an artificial intelligence based telephone quality inspection device in an embodiment of the present invention.
The achievement of the objects, functional features and advantages of the present invention will be further described with reference to the accompanying drawings, in conjunction with the embodiments.
Detailed Description
It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention. The terms first, second and the like in the description and in the claims and in the above-described figures, are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that the embodiments described herein may be implemented in other sequences than those illustrated or otherwise described herein. Furthermore, the terms "comprises," "comprising," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or modules is not necessarily limited to those listed or explicitly listed or inherent to such process, method, article, or apparatus, but may include other steps or modules that may not be listed or inherent to such process, method, article, or apparatus, the partitioning of such modules by the present invention may be by one logical partitioning, and may be implemented by other means, such as a plurality of modules may be combined or integrated in another system, or some features may be omitted, or not implemented.
In order to solve the technical problems, the invention mainly provides the following technical proposal
The invention converts the whole amount of call voice into text by voice conversion technology, then obtains key core words after converting the text into text by LTP technology, counts the key core words, and obtains the score of the call record by inputting the counted key words into the neural network. Through intelligent quality testing, realize rule objectivity, quality testing is accurate. The intelligent quality inspection system effectively saves time and cost and completes configuration of optimizing personnel. Through the real-time quality inspection of artificial intelligence, the time of the artificial quality inspection is saved, statistics and analysis can be well carried out, specific demands of customers and complaint points are well calculated, relevant rules are summarized, and the development of business and the resolution of complaints are assisted.
Referring to fig. 1, the following illustrates a phone quality inspection method based on artificial intelligence, which includes:
101. and obtaining a plurality of call records to be processed.
Call recording refers to a technique or method by which voice communication signals on a telephone line are monitored and converted into a medium that can be saved and played back. The sampling index of the call record includes format, sampling frequency, sampling accuracy, sound channel, compression rate and data quantity per second.
102. And marking the corresponding score for each call record to be processed.
The score scores the quality of the call record and is assessed by user input. Taking the example of telephone scoring, such as customer service telephone quality assessment, there may be several considerations for it, such as its call duration, call voice size, call words, etc., called the attribute or feature of the telephone. And for the talk time period: 20 minutes, sound size: moderate, words for conversation: the word "you" appears 40 times such a set of data we call an example or sample, and when each attribute of the phone is so expanded, the set formed can be called a data set. Where specific values for those attributes are referred to as attribute values. The space of attributes is referred to as the "attribute space", "sample space", or "input space". For example: for phone scoring, he has three attributes: call duration, call sound size, call words. And generating a three-dimensional coordinate space by taking each attribute as a coordinate. Each call, we can find their respective corresponding locations in this resulting three-dimensional space, and therefore each example is also referred to as a feature vector. The process of learning from data to a model is called "learning" or "training", and the entire process is implemented by executing some learning algorithm. The data used in the training process is referred to as "training data", wherein each sample is referred to as a "training sample", and the set of all training samples is referred to as a "training set". The learned model corresponds to some potential law about the data and is therefore referred to as a "hypothesis"; this underlying law itself is called "true phase", and the learning process is to find or approximate the true phase. Since we eventually need to do something similar to 'predictive', i.e. we are helped to judge whether the phone in front of us is acceptable. We need to label the previous sample with a point, i.e. satisfying the qualification here is called label, and the labeled sample we refer to as sample.
103. And converting the plurality of call records to be processed into a plurality of text messages to be processed through a preset sequence neural network model.
The preset sequential neural network model refers to converting the lexical content in human speech into computer readable input. The sequence neural network is a type of recurrent neural network which takes sequence data as input, performs recursion in the evolution direction of the sequence and all nodes (circulation units) are connected in a chained mode. The sequence neural network has memory, parameter sharing and complete figure, so that the sequence neural network has certain advantages in learning the nonlinear characteristics of the sequence. The cyclic neural network has application in the fields of natural language processing, such as speech recognition, language modeling, machine translation and the like, and is also used for various time series forecasting. The cyclic neural network constructed by introducing the convolutional neural network can process the computer vision problem containing the sequence input.
104. And segmenting the text information to be processed through a language technology platform LTP technology to obtain a plurality of core keywords and frequencies corresponding to the core keywords.
The language technology platform provides rich, efficient and accurate natural language processing technologies including Chinese word segmentation, part-of-speech tagging, named entity recognition, dependency syntactic analysis, semantic role tagging and the like. LTP provides a series of chinese natural language processing tools that users can use to perform word segmentation, part-of-speech tagging, syntactic analysis, etc. on chinese text. From an application perspective, LTP provides the following components for a user: means for generating a statistical machine learning model for a single natural language processing task. And calling a programming interface for analyzing the model aiming at a single natural language processing task. And combining all analysis tools in a pipeline mode to form a unified Chinese natural language processing system. The system can call the model file for Chinese language processing. For a single natural language processing task, a cloud-based programming interface.
105. Taking the frequency corresponding to the core keyword as input x i The noted score is taken as ideal output y i Inputting the target neural network into the initial neural network, and training the neuron weight in the initial neural network through the loss function to obtain the target neural network.
The loss function is
Where m is the number of call voices to be processed, b and lambda are constants and are used to determine, i w i 1 Is the L1 norm of w.
Neural networks refer to an algorithm that replicates such dense neuronal networks. By processing multiple data streams at a time, the computer can significantly reduce the time required to process the data. Application of this technique to deep learning has resulted in artificial neural networks. These artificial neural networks are composed of input nodes, output nodes and node layers.
An input node for receiving data.
And the output node is used for outputting the result data.
And the node layer is used for converting the data input from the input node into content which can be used by the output node. The node layer refers to a plurality of hidden nodes between the input node and the output node, and may be a hidden layer. As data progresses through these hidden nodes, the neural network uses logic to decide to pass the data to the next hidden node.
106. And deploying the target neural network to the client.
Deployment refers to the process of collecting, packaging, installing, configuring, publishing configuration files, user manuals, help documents, and the like of the neural network. The main characteristics of the software deployment process are: process coverage, process variability, inter-process coordination, and model abstraction. Some abstract software deployment models have been proposed for efficiently guiding the deployment process, including application models, enterprise models, site models, product models, policy models, and deployment models.
107. And receiving call information input by a user, and inputting the call information to a target neural network through a client.
Each node of the input layer needs to perform point-to-point calculation with each node of the hidden layer, and the calculation method is weighted summation and activation. Each value calculated using the hidden layer is calculated using the same method, and the output layer. The hidden layer uses a logistic regression function as the activation function, while the output layer uses a linear function. This is because the linear function can maintain a numerical scale of any range previously, facilitating comparison with sample values, whereas the numerical range of logistic regression can only be between 0 and 1. The values of the input layer are firstly transmitted to the hidden layer through network calculation and then transmitted to the output layer in the same way, and finally the output value and the sample value are compared to calculate the error.
108, calculating the score of the call information through the neuron weight value in the target neural network, and returning the score of the call information to the user.
The invention converts the whole amount of call voice into text by voice conversion technology, then obtains key core words after converting the text into text by LTP technology, counts the key core words, and obtains the score of the call record by inputting the counted key words into the neural network. Through intelligent quality testing, realize rule objectivity, quality testing is accurate. The intelligent quality inspection system effectively saves time and cost and completes configuration of optimizing personnel. Through the real-time quality inspection of artificial intelligence, the time of the artificial quality inspection is saved, statistics and analysis can be well carried out, specific demands of customers and complaint points are well calculated, relevant rules are summarized, and the development of business and the resolution of complaints are assisted.
In some embodiments, before the text information to be processed is segmented by the language technology platform LTP technology to obtain a plurality of core keywords and frequencies corresponding to the core keywords, the method further includes:
generating a unique identification ID number for each text message to be processed;
constructing a hypertext transfer protocol (HTTP) request according to the API parameters of the text message to be processed;
Extracting Chinese segmentation from text information to be processed by using an HTTP request and LTP technology;
and matching the words recorded in the database, and deleting the words with the matching degree lower than the threshold value to obtain words with the matching degree higher than the threshold value.
In the above embodiment, the language technical platform is connected through the hypertext transfer protocol, and the text is segmented through the language technical platform.
In some embodiments, the converting, by a preset sequential neural network model, the plurality of call recordings to be processed into the plurality of text messages to be processed includes:
acquiring a plurality of training voice samples;
inputting a training voice sample into a preset sequence neural network model, and updating neuron weights of the preset sequence neural network model through a natural language processing NLP algorithm, voice information and text labels corresponding to the voice information to obtain a target model;
adjusting the weight of the neuron of the target model, and updating a preset sequence neural network model;
and converting the plurality of call records to be processed into a plurality of text messages to be processed through the updated preset sequence neural network model.
In the above embodiment, through the trained target model, the call record may be converted into the corresponding text sequence by the computer.
In some embodiments, receiving call information input by a user, and inputting the call information to a target neural network through a client, includes:
when prompting a user to input call information, starting a microphone application through a client;
after receiving the indication message, prompting a user to input call information through the client;
when call information is obtained, noise is removed from the call information, background sound is removed, and the call information is compressed through a client side, so that the call information after preprocessing is obtained;
judging whether the call information subjected to pretreatment accords with an input preset rule or not through a client;
if the rule does not accord with the preset rule, the client side is instructed to prompt the user to input again;
if the preset rule is met, uploading the call information which is input by the user and is subjected to pretreatment to a server through the client;
and inputting the pretreated call information into a target neural network.
In the above embodiment, the call information input by the user is received by turning on the microphone and the client.
In some embodiments, before obtaining the plurality of call records to be processed, the method further includes:
rejecting call records to be processed with call duration smaller than a threshold value;
adding a plurality of characteristic items for the call record to be processed, wherein the characteristic items at least comprise: whether to make a first call, industry record average duration, and satisfaction of user phone evaluation.
In the above embodiment, invalid voice information input by the user is audited. The original data should be mainly audited in terms of both integrity and accuracy. The integrity check is mainly to check whether units or individuals to be investigated are missing or not, and whether all investigation items or indexes are completely filled or not. Accuracy audit mainly includes two aspects: firstly, checking whether the data information truly reflects objective actual conditions or not, and whether the content accords with the actual conditions or not; and secondly, checking whether the data has errors, whether the calculation is correct, and the like. The method for checking the accuracy of the data mainly comprises logic checking and calculation checking. The logic check is mainly to check whether the data accords with the logic or not, whether the content is reasonable or not, and whether the mutual contradiction exists between each item or number or not. The calculation check is to check whether each item of data in the questionnaire has errors on the calculation result and the calculation method, and is mainly used for checking quantitative data.
In some embodiments, word segmentation is performed on text information to be processed through a language technology platform LTP technology to obtain a plurality of core keywords and frequencies corresponding to the core keywords, including:
The frequency corresponding to the core keyword passes
Calculation of F w The frequency of the core keywords is indicated, N is the number of sentences in which the core keywords appear, and M is the number of sentences of the text information to be processed.
In the above embodiment, the ratio of the number of times each object appears to the total number of times is the frequency.
In some embodiments, the frequency corresponding to the core keyword is used as the input x i The noted score is taken as ideal output y i Inputting the text information to be processed through a language technology platform LTP technology before training neuron weights in an initial neural network through a loss function to obtain a target neural network, and obtaining a plurality of core keywords and coresAfter the frequencies corresponding to the keywords, the method further comprises the following steps:
if the output y of the neural network j With n valid feature inputs x 1 ，x 2 ，…，x n The weight value of the corresponding connection is w 1 ，w 2 ，…，w n Then the neural network passes through
And initializing the weight.
In the above embodiment, training of the neural network is expedited by initializing weights for the neural network.
An artificial intelligence based telephone quality inspection device 20 is shown in fig. 2 and is applicable to artificial intelligence based telephone quality inspection. The phone quality inspection device based on artificial intelligence in the embodiment of the invention can realize the steps corresponding to the phone quality inspection method based on artificial intelligence executed in the embodiment corresponding to the figure 1. The functions implemented by the phone quality inspection device 20 based on artificial intelligence can be implemented by hardware, or can be implemented by executing corresponding software by hardware. The hardware or software includes one or more modules corresponding to the functions described above, which may be software and/or hardware. The phone quality inspection device based on artificial intelligence may include an input/output module 201 and a processing module 202, and the functional implementation of the processing module 202 and the input/output module 201 may refer to the operations performed in the embodiment corresponding to fig. 1, which are not described herein. The input-output module 201 may be used to control the input, output and acquisition operations of the input-output module 201.
In some embodiments, the input/output module 201 may be configured to obtain a plurality of call records to be processed;
the processing module 202 may be configured to annotate the score corresponding to each call record to be processed; converting the plurality of call records to be processed into a plurality of text messages to be processed through a preset sequence neural network model; word segmentation is carried out on the text information to be processed through a language technology platform LTP technology, so that a plurality of core keywords and frequencies corresponding to the core keywords are obtained;the frequency corresponding to the core keyword is used as input x through the input/output module i The score is noted as an ideal output y i Inputting the neural values into an initial neural network, and training the weights of the neurons in the initial neural network through a loss function to obtain a target neural network; deploying the target neural network to a client; receiving call information input by a user through the input/output module, and inputting the call information to the target neural network through the client; and calculating the score of the call information through the neuron weight in the target neural network, and returning the score of the call information to the user.
In some embodiments, the processing module 202 is further configured to:
converting the plurality of call records to be processed into text information to be processed through a preset sequence neural network model;
generating a unique identification ID number for each piece of text information to be processed;
constructing a hypertext transfer protocol (HTTP) request according to the API parameters of the text information to be processed;
extracting Chinese segmentation from text information to be processed by the HTTP request and the LTP technology;
and matching the words recorded in the database, and deleting the words with the matching degree lower than the threshold value to obtain words with the matching degree higher than the threshold value.
In some embodiments, the processing module 202 is further configured to:
acquiring a plurality of training voice samples;
inputting the training voice sample into the preset sequence neural network model, and updating the neuron weight of the preset sequence neural network model through a natural language processing NLP technology, the voice information and a text label corresponding to the voice information to obtain a target model;
adjusting the weight of the neuron of the target model, and updating the preset sequence neural network model;
And converting the plurality of call records to be processed into a plurality of text messages to be processed through the updated preset sequence neural network model.
In some embodiments, the processing module 202 is further configured to:
when prompting a user to input the call information, starting a microphone application through the client;
after receiving the indication message, prompting a user to input call information through the client;
when call information is obtained, noise is removed from the call information, background sound is removed, and the call information is compressed through the client side, so that the call information which is preprocessed is obtained;
judging whether the pretreated call information accords with an input preset rule or not through the client;
if the rule does not accord with the preset rule, the client is instructed to prompt the user to input again;
if the pre-processed call information meets the preset rule, uploading the pre-processed call information input by the user to a server through the client;
and inputting the pretreated call information to the target neural network.
In some embodiments, the processing module 202 is further configured to:
rejecting the call record to be processed with the call duration smaller than a threshold value;
Adding a plurality of characteristic items for the call record to be processed, wherein the characteristic items at least comprise: whether to make a first call, industry record average duration, and satisfaction of user phone evaluation.
In some embodiments, the processing module 202 is further configured to:
the frequency corresponding to the core keyword passes through
Calculation of F w Refers to the frequency of the core keywords, N refers to the number of sentences in which the core keywords appear, M refers to the text information to be processedIs a sentence number of (a) in the sentence number.
In some embodiments, the processing module 202 is further configured to:
if the output y of the neural network j With n of said valid characteristic inputs x 1 ，x 2 ，…，x n The weight value of the corresponding connection is w 1 ，w 2 ，…，w n The neural network passes through
And initializing the weight.
The phone quality inspection device based on artificial intelligence in the embodiment of the invention is described above from the point of view of modularized functional entities, and the phone quality inspection device based on artificial intelligence is described below from the point of view of hardware, as shown in fig. 3, and includes: a processor, a memory, an input output unit (which may also be a transceiver, not identified in fig. 3) and a computer program stored in the memory and executable on the processor. For example, the computer program may be a program corresponding to the phone quality inspection method based on artificial intelligence in the embodiment corresponding to fig. 1. For example, when the computer device implements the functionality of the artificial intelligence based phone quality inspection apparatus 20 as shown in fig. 2, the processor, when executing the computer program, implements the steps of the artificial intelligence based phone quality inspection method performed by the artificial intelligence based phone quality inspection apparatus 20 in the embodiment corresponding to fig. 2 described above. Alternatively, the processor may implement the functions of the modules in the artificial intelligence based phone quality inspection device 20 according to the embodiment of fig. 2 when executing the computer program. For another example, the computer program may be a program corresponding to the phone quality inspection method based on artificial intelligence in the embodiment corresponding to fig. 1.
The processor may be a central processing unit (central processing unit, CPU), or other general purpose processor, digital signal processor (digital signal processor, DSP), application specific integrated circuit (application specific integrated circuit, ASIC), off-the-shelf programmable gate array (field-programmable gate array, FPGA) or other programmable logic device, discrete gate or transistor logic device, discrete hardware components, or the like. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like that is a control center of the computer device, connecting various parts of the overall computer device using various interfaces and lines.
The memory may be used to store the computer program and/or modules, and the processor may implement various functions of the computer device by running or executing the computer program and/or modules stored in the memory, and invoking data stored in the memory. The memory may mainly include a storage program area and a storage data area, wherein the storage program area may store an operating system, an application program (such as a sound playing function, an image playing function, etc.) required for at least one function, and the like; the storage data area may store data (such as audio data, video data, etc.) created according to the use of the cellular phone, etc. In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as a hard disk, memory, plug-in hard disk, smart Media Card (SMC), secure Digital (SD) Card, flash Card (Flash Card), at least one disk storage device, flash memory device, or other volatile solid-state storage device.
The input-output unit may be replaced by a receiver and a transmitter, and may be the same or different physical entities. Are the same physical entities and may be collectively referred to as input/output units. The input and output may be a transceiver.
The memory may be integrated in the processor or may be provided separately from the processor.
From the above description of the embodiments, it will be clear to those skilled in the art that the above-described embodiment method may be implemented by means of software plus a necessary general hardware platform, but of course may also be implemented by means of hardware, but in many cases the former is a preferred embodiment. Based on such understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art in the form of a software product stored in a storage medium (e.g. ROM/RAM), comprising instructions for causing a terminal (which may be a mobile phone, a computer, a server or a network device, etc.) to perform the method according to the embodiments of the present invention.
While the embodiments of the present invention have been described above with reference to the drawings, the present invention is not limited to the above-described embodiments, which are merely illustrative and not restrictive, and many modifications may be made thereto by those of ordinary skill in the art without departing from the spirit of the present invention and the scope of the appended claims, which are to be accorded the full scope of the present invention as defined by the following description and drawings, or by any equivalent structures or equivalent flow changes, or by direct or indirect application to other relevant technical fields.
Claims (9)
1. A method for phone quality inspection based on artificial intelligence, the method comprising:
acquiring a plurality of call records to be processed;
marking a corresponding score for each call record to be processed;
converting the plurality of call records to be processed into a plurality of text messages to be processed through a preset sequence neural network model;
word segmentation is carried out on the text information to be processed through a language technology platform LTP technology, so that a plurality of core keywords and frequencies corresponding to the core keywords are obtained;
taking the frequency corresponding to the core keyword as input x i The score is noted as an ideal output y i Inputting the neural values into an initial neural network, and training the weights of the neurons in the initial neural network through a loss function to obtain a target neural network;
deploying the target neural network to a client;
receiving call information input by a user, and inputting the call information to the target neural network through the client;
calculating the score of the call information through the neuron weight in the target neural network, and returning the score of the call information to a user;
the word segmentation is carried out on the text information to be processed through a language technology platform LTP technology to obtain a plurality of core keywords and frequencies corresponding to the core keywords, and the method comprises the following steps:
The frequency corresponding to the core keyword passes through
Calculation of F w The frequency of the core keywords is indicated, N is the number of sentences in which the core keywords appear, and M is the number of sentences in the text information to be processed.
2. The method according to claim 1, wherein before the text information to be processed is segmented by the language technical platform LTP technology to obtain a plurality of core keywords and frequencies corresponding to the core keywords, the method further comprises:
generating a unique identification ID number for each piece of text information to be processed;
constructing a hypertext transfer protocol (HTTP) request according to the API parameters of the text information to be processed;
extracting Chinese segmentation from text information to be processed by the HTTP request and the LTP technology;
and matching the words recorded in the database, and deleting the words with the matching degree lower than the threshold value to obtain words with the matching degree higher than the threshold value.
3. The method of claim 1, wherein converting the plurality of call recordings to be processed into a plurality of text messages to be processed by a preset sequential neural network model, comprises:
Acquiring a plurality of training voice samples;
inputting the training voice sample into the preset sequence neural network model, and updating the neuron weight of the preset sequence neural network model through a natural language processing NLP algorithm, the training voice sample and a text label corresponding to the training voice sample to obtain a target model;
adjusting the weight of the neuron of the target model, and updating the preset sequence neural network model;
and converting the plurality of call records to be processed into a plurality of text messages to be processed through the updated preset sequence neural network model.
4. The method of claim 1, wherein receiving call information entered by a user, entering the call information into the target neural network by the client, comprises:
when prompting a user to input the call information, starting a microphone application through the client;
after receiving the indication message, prompting a user to input call information through the client;
when call information is obtained, noise is removed from the call information, background sound is removed, and the call information is compressed through the client side, so that the call information which is preprocessed is obtained;
Judging whether the pretreated call information accords with an input preset rule or not through the client;
if the rule does not accord with the preset rule, the client is instructed to prompt the user to input again;
if the pre-processed call information meets the preset rule, uploading the pre-processed call information input by the user to a server through the client;
and inputting the pretreated call information to the target neural network.
5. The method of claim 1, wherein prior to the obtaining the plurality of pending call records, the method further comprises:
rejecting the call record to be processed with the call duration smaller than a threshold value;
adding a plurality of characteristic items for the call record to be processed, wherein the characteristic items at least comprise: whether to make a first call, industry record average duration, and satisfaction of user phone evaluation.
6. The method according to any one of claims 1-5, wherein the step of inputting x is performed by taking as input x a frequency corresponding to the core keyword i The score is noted as an ideal output y i Before inputting the text information to be processed into an initial neural network and training the neuron weight in the initial neural network through a loss function to obtain a target neural network, the method further comprises the steps of:
If the output y of the neural network j With n valid feature inputs x 1 ，x 2 ，…，x n The weight value of the corresponding connection is w 1 ,w 2 ,…，w n The neural network passes through
b=0 initializing weight, said +.>
For representing the sum of the products of 2 of the n valid feature inputs and the corresponding weights, the b=0 for representing the value of the constant b in the loss function as 0, the ≡>
Including the constant b and the sum number of valid feature inputs.
7. An artificial intelligence based telephone quality inspection device, the device comprising:
the input/output module is used for acquiring a plurality of call records to be processed;
the processing module is used for marking corresponding scores for each call record to be processed; converting the plurality of call records to be processed into a plurality of text messages to be processed through a preset sequence neural network model; word segmentation is carried out on the text information to be processed through a language technology platform LTP technology, so that a plurality of core keywords and frequencies corresponding to the core keywords are obtained; the frequency corresponding to the core keyword is used as input x through the input/output module i The score is noted as an ideal output y i Inputting the neural values into an initial neural network, and training the weights of the neurons in the initial neural network through a loss function to obtain a target neural network; deploying the target neural network to a client; receiving call information input by a user through the input/output module, and inputting the call information to the target neural network through the client; calculating the score of the call information through the neuron weight in the target neural network, and returning the score of the call information to a user;
The word segmentation is carried out on the text information to be processed through a language technology platform LTP technology to obtain a plurality of core keywords and frequencies corresponding to the core keywords, and the method comprises the following steps:
the frequency corresponding to the core keyword passes through
Calculation of F w The frequency of the core keywords is indicated, N is the number of sentences in which the core keywords appear, and M is the number of sentences in the text information to be processed.
8. An artificial intelligence based telephone quality inspection device, the device comprising:
at least one processor, memory, and input output unit;
wherein the memory is for storing program code and the processor is for invoking the program code stored in the memory to perform the method of any of claims 1-6.
9. A computer storage medium comprising instructions which, when run on a computer, cause the computer to perform the method of any of claims 1-6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911367527.7A CN111182162B (en) | 2019-12-26 | 2019-12-26 | Telephone quality inspection method, device, equipment and storage medium based on artificial intelligence |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911367527.7A CN111182162B (en) | 2019-12-26 | 2019-12-26 | Telephone quality inspection method, device, equipment and storage medium based on artificial intelligence |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111182162A CN111182162A (en) | 2020-05-19 |
| CN111182162B true CN111182162B (en) | 2023-04-25 |
Family
ID=70652188
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911367527.7A Active CN111182162B (en) | 2019-12-26 | 2019-12-26 | Telephone quality inspection method, device, equipment and storage medium based on artificial intelligence |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111182162B (en) |
Families Citing this family (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111723204B (en) * | 2020-06-15 | 2021-04-02 | 龙马智芯（珠海横琴）科技有限公司 | Method and device for correcting voice quality inspection area, correction equipment and storage medium |
| CN111753994B (en) * | 2020-06-22 | 2023-11-03 | 深圳鲲云信息科技有限公司 | Data processing method and device of AI chip and computer equipment |
| CN112163434B (en) * | 2020-10-20 | 2024-03-08 | 腾讯科技（深圳）有限公司 | Text translation method, device, medium and electronic equipment based on artificial intelligence |
| CN112911072A (en) * | 2021-01-28 | 2021-06-04 | 携程旅游网络技术（上海）有限公司 | Call center volume identification method and device, electronic equipment and storage medium |
| CN113132523B (en) * | 2021-04-19 | 2023-05-26 | 广州绿怡信息科技有限公司 | Call detection model training method and call detection method |
| CN113179352B (en) * | 2021-05-25 | 2023-04-07 | 北京捷通华声科技股份有限公司 | Call quality determination method, call quality determination device, and computer-readable storage medium |
| CN113724708A (en) * | 2021-07-23 | 2021-11-30 | 山东环维通讯科技有限公司 | Quality control system for converting voice into text intelligent checklist |
| CN115499545A (en) * | 2022-09-13 | 2022-12-20 | 上海浦东发展银行股份有限公司 | Telephone customer service performance scoring method, device, equipment and medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107885726A (en) * | 2017-11-06 | 2018-04-06 | 广州杰赛科技股份有限公司 | Customer service quality evaluating method and device |
| CN109327632A (en) * | 2018-11-23 | 2019-02-12 | 深圳前海微众银行股份有限公司 | Intelligent quality inspection system, method and the computer readable storage medium of customer service recording |
| CN110378587A (en) * | 2019-07-08 | 2019-10-25 | 上海辰尚信息科技有限公司 | Intelligent quality detecting method, system, medium and equipment |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP5855290B2 (en) * | 2014-06-16 | 2016-02-09 | パナソニックＩｐマネジメント株式会社 | Service evaluation device, service evaluation system, and service evaluation method |
| CN106776806A (en) * | 2016-11-22 | 2017-05-31 | 广东电网有限责任公司佛山供电局 | The methods of marking and system of call center's quality inspection voice |
| CN108806667B (en) * | 2018-05-29 | 2020-04-17 | 重庆大学 | Synchronous recognition method of voice and emotion based on neural network |
| CN108962282B (en) * | 2018-06-19 | 2021-07-13 | 京北方信息技术股份有限公司 | Voice detection analysis method and device, computer equipment and storage medium |
| CN109961801A (en) * | 2019-01-16 | 2019-07-02 | 深圳壹账通智能科技有限公司 | Intelligent Service evaluation method, computer readable storage medium and terminal device |
| CN110310663A (en) * | 2019-05-16 | 2019-10-08 | 平安科技（深圳）有限公司 | Words art detection method, device, equipment and computer readable storage medium in violation of rules and regulations |
-
2019
- 2019-12-26 CN CN201911367527.7A patent/CN111182162B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107885726A (en) * | 2017-11-06 | 2018-04-06 | 广州杰赛科技股份有限公司 | Customer service quality evaluating method and device |
| CN109327632A (en) * | 2018-11-23 | 2019-02-12 | 深圳前海微众银行股份有限公司 | Intelligent quality inspection system, method and the computer readable storage medium of customer service recording |
| CN110378587A (en) * | 2019-07-08 | 2019-10-25 | 上海辰尚信息科技有限公司 | Intelligent quality detecting method, system, medium and equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111182162A (en) | 2020-05-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111182162B (en) | Telephone quality inspection method, device, equipment and storage medium based on artificial intelligence | |
| CN111026842B (en) | Natural language processing method, natural language processing device and intelligent question-answering system | |
| CN111046132B (en) | Customer service question-answering processing method and system for searching multiple rounds of conversations | |
| KR102288249B1 (en) | Information processing method, terminal, and computer storage medium | |
| CN110688454A (en) | Method, device, equipment and storage medium for processing consultation conversation | |
| CN110232109A (en) | A kind of Internet public opinion analysis method and system | |
| CN110910283A (en) | Method, device, equipment and storage medium for generating legal document | |
| CN112036842B (en) | Intelligent matching device for scientific and technological service | |
| CN109857846B (en) | Method and device for matching user question and knowledge point | |
| CN111680147A (en) | Data processing method, device, equipment and readable storage medium | |
| CN110287292B (en) | Judgment criminal measuring deviation degree prediction method and device | |
| CN113297365B (en) | User intention judging method, device, equipment and storage medium | |
| CN112288584A (en) | Insurance application processing method and device, computer readable medium and electronic equipment | |
| CN114282498B (en) | Data knowledge processing system applied to electric power transaction | |
| CN117520522B (en) | Intelligent dialogue method and device based on combination of RPA and AI and electronic equipment | |
| CN110209804B (en) | Target corpus determining method and device, storage medium and electronic device | |
| CN117592778A (en) | Risk early warning system | |
| CN111552785A (en) | Method and device for updating database of human-computer interaction system, computer equipment and medium | |
| CN116452212A (en) | Intelligent customer service commodity knowledge base information management method and system | |
| CN113010664B (en) | Data processing method and device and computer equipment | |
| CN115510192A (en) | News event context relationship detection method and device | |
| CN115080732A (en) | Complaint work order processing method and device, electronic equipment and storage medium | |
| CN111178068A (en) | Conversation emotion detection-based urge tendency evaluation method and apparatus | |
| CN117453895B (en) | Intelligent customer service response method, device, equipment and readable storage medium | |
| CN118573776A (en) | Behavior intention recognition method and device and electronic equipment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |