CN111462752A - Client intention identification method based on attention mechanism, feature embedding and BI-L STM - Google Patents
Client intention identification method based on attention mechanism, feature embedding and BI-L STM Download PDFInfo
- Publication number
- CN111462752A CN111462752A CN202010251934.8A CN202010251934A CN111462752A CN 111462752 A CN111462752 A CN 111462752A CN 202010251934 A CN202010251934 A CN 202010251934A CN 111462752 A CN111462752 A CN 111462752A
- Authority
- CN
- China
- Prior art keywords
- intention
- stm
- attention mechanism
- user intention
- preset
- 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.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/049—Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; 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 OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/08—Speech classification or search
- G10L15/10—Speech classification or search using distance or distortion measures between unknown speech and reference templates
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; 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 OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/08—Speech classification or search
- G10L15/18—Speech classification or search using natural language modelling
- G10L15/1822—Parsing for meaning understanding
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; 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
- G10L2015/0631—Creating reference templates; Clustering
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Abstract
The invention discloses a customer intention recognition method based on an attention mechanism, feature embedding and BI-L STM, which comprises the steps of obtaining customer service conversation contents, cleaning and denoising the customer service conversation contents, screening preset intention worksheet data by using a preset keyword library, and judging user intents corresponding to the preset intention worksheet data by using a pre-established and trained user intention recognition model based on the BI-L STM and the attention mechanism.
Description
Technical Field
The invention relates to the technical field of voice analysis, in particular to a client intention recognition method based on an attention mechanism, feature embedding and BI-L STM.
Background
In the existing algorithm aiming at semantic recognition, the rule-based intention recognition model is complex to construct and poor in generalization capability. Specifically, the rule-based intention classification solution is to classify sentence intentions by heuristic rules of pre-defined category information. The method generally constructs a rule-based classifier according to predefined rules. The keywords of the question are obtained by using the rules defined in advance in the classifier, and the intention of the natural language question is understood according to the keywords, so that the purpose of classification is achieved. However, in order to obtain a good recognition effect in the using process, a large number of rules are often defined, and the rules can be obtained only by manually marking, and when the number of the corpora is large, a large amount of manpower is consumed. Secondly, most rules are constructed without generalization, the rules constructed by analyzing the corpus of one field can only be used on data sets similar to the field, and the effect of the rules on other fields or other data sets is poor. Therefore, it is difficult to construct a universal rule framework with generalization.
Deep neural network based intent recognition models focus on semantic relationships. Deep learning methods have been widely applied to various sub-fields of natural language processing and text mining. The core of the deep learning method is a deep neural network. The current deep neural network mainly comprises three types, namely a convolutional neural network, a cyclic neural network and a mixed network of the convolutional neural network and the cyclic neural network. These networks perform feature learning on texts such as sentences or paragraphs by means of learning with a supervision signal or unsupervised reconstruction, and generate expression vectors capable of effectively representing text semantics. However, whether CNN (Convolutional Neural networks) or RNN (Recurrent Neural networks), the input is generally a word vector or a word vector, so its representation vector is focused on semantics. The rule-based method is also an effective method in text mining, and the idea based on the rule is mostly based on the logic of combination or condition combination.
The deep neural network-based intention recognition model shows good performance in capturing overall semantics to achieve more reasonable intention recognition. But cannot capture text information that is relied upon for a long time when the content is too long. That is, it is actually difficult to capture the dependency between two text words having a large distance at a time in practice. This makes it less than well suited to handle tasks that have a strong dependency on text context. Therefore, the semantics of the whole sentence needs to be fused, but the conventional intention recognition model is difficult to meet the semantic fusion and processing requirements of the long sentence pattern and the complex sentence pattern no matter the input or the output is input.
Existing intent recognition tasks assume that a given text segment does not have too much extraneous or even interfering information, and therefore are generally analyzed directly on the given text segment. But it is highly likely that the original call text message will carry useless or interfering information. The reason is the multi-themes expressed by the user and the liberty of social text. Useless or interfering information is likely to significantly degrade the performance of subsequent intent and semantic analysis. Therefore, in order to make algorithms such as user intention trend analysis and the like really function, how to accurately position the target text segment must be researched.
Disclosure of Invention
Aiming at least one of the problems, the invention provides a customer intention recognition method based on an attention mechanism, a feature embedding and a BI-L STM (Bi-directional L ong Short-Term Memory network), which is characterized in that text conversion is carried out after customer service voice call is cleaned and denoised, irrelevant interference information in voice call content is filtered, the performance of semantic and intention analysis is improved, preset intention worksheet data related to the user intention to be recognized is screened out by utilizing a preset keyword library, the corresponding user intention is judged by utilizing a user intention recognition model which is established in advance and trained and is based on the BI-L STM and the attention mechanism, the attention capability of the model to key feature information and the semantic fusion and processing capability to complex sentence formulas are improved, the accurate understanding of text meaning is realized, and the recognition capability to the user intention is improved.
In order to achieve the purpose, the invention provides a customer intention recognition method based on an attention mechanism, feature embedding and BI-L STM, which comprises the steps of obtaining customer service conversation contents, cleaning and denoising the customer service conversation contents, screening preset intention worksheet data by using a preset keyword library, and judging user intents corresponding to the preset intention worksheet data by using a pre-established and trained user intention recognition model based on the BI-L STM and the attention mechanism.
In the technical scheme, the method for establishing and training the user intention recognition model based on the BI-L STM and the attention mechanism preferably comprises the steps of determining keywords and related synonyms according to the user intention to be recognized, establishing a preset keyword library, screening the customer service call content by using the preset keyword library, manually labeling user intention labels by using worksheet data of the screened keywords as a training set, establishing the user intention recognition model by using the BI-L STM and a network structure of the attention mechanism, inputting the worksheet data in the training set into the user intention recognition model, and training the user intention recognition model by using the manually labeled user intention labels.
In the above technical solution, preferably, the method for determining the user intention corresponding to the preset intention work order data by using the user intention recognition model includes: judging a user intention label hidden in each piece of preset intention work order data through the user intention identification model; and determining the user intention corresponding to each piece of preset intention worksheet data according to the type and the number of the user intention labels in each piece of preset intention worksheet data and through a voting principle and the original state of the user.
In the above technical solution, preferably, a near-sense word model is used to determine a near-sense word related to the keyword, and the near-sense word model is based on a deep learning model based on word2vec, combines semantic similarity, pinyin similarity and editing distance, and provides a semantic near-sense word recognition result exceeding the face of the keyword, so that the keyword and the recognized near-sense word together construct the preset keyword library.
In the above technical solution, preferably, in the user intention recognition model, word vector information and feature information of the work order data are used as input of a BI-L STM neural network, a hidden state output of the BI-L STM neural network is spliced with the feature information and then used as input of an attention mechanism, a weight of each hidden state is calculated through normalization, and a feature representation vector of the work order data is obtained through weighted summation.
In the above technical solution, preferably, the attention mechanism is an Encoder model, and is configured to calculate a weighted summation of a hidden state output vector set of the BI-L STM neural network according to the feature information vector.
In the above technical solution, preferably, the cleaning of the customer service call content specifically includes file merging, duplicate removal, and null data deletion.
In the above technical solution, preferably, the customer intention identification method based on attention mechanism, feature embedding and BI-L STM further includes label fusion of the user intention corresponding to the preset intention worksheet data and worksheet information corresponding to the preset intention worksheet data.
Compared with the prior art, the method has the advantages that text conversion is carried out after the customer service voice call is cleaned and denoised, irrelevant interference information in voice call content is filtered, the performance of semantic and intention analysis is improved, the preset keyword library is utilized to screen out preset intention worksheet data relevant to the user intention to be recognized, the user intention recognition model which is established in advance and trained and based on BI-L STM and attention mechanism is utilized to judge the corresponding user intention, the attention capacity of the model to key characteristic information and the semantic fusion and processing capacity to complex sentence patterns are improved, accurate understanding of text meanings is achieved, and the recognition capacity to the user intention is improved.
Drawings
FIG. 1 is a schematic flow chart diagram of a disclosed customer intent recognition method based on attention mechanism, feature embedding and BI-L STM in accordance with an embodiment of the present invention;
FIG. 2 is a schematic process flow diagram of a client intent recognition model for number portability disclosed in an embodiment of the present invention;
FIG. 3 is a schematic diagram of a process flow of a near word model according to an embodiment of the present disclosure;
FIG. 4 is a schematic diagram of a training process of a user intention recognition model according to an embodiment of the present invention;
FIG. 5 is a diagram illustrating a process of integrating predicted results of a user intent recognition model according to an embodiment of the present invention;
FIG. 6 is a schematic diagram of a network structure of a user intention recognition model according to an embodiment of the present invention;
fig. 7 is an algorithmic schematic of the attention mechanism disclosed in one embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be obtained by a person skilled in the art without any inventive step based on the embodiments of the present invention, are within the scope of the present invention.
The invention is described in further detail below with reference to the attached drawing figures:
as shown in FIG. 1, the customer intention identification method based on the attention mechanism, the feature embedding and the BI-L STM comprises the steps of obtaining customer service conversation contents, cleaning and denoising the customer service conversation contents, screening preset intention worksheet data by using a preset keyword library, and judging user intentions corresponding to the preset intention worksheet data by using a user intention identification model which is established and trained in advance and based on the BI-L STM and the attention mechanism.
In the embodiment, text conversion is carried out after customer service voice call cleaning and denoising, the cleaning specifically comprises file merging, duplicate removal and null data deletion, irrelevant interference information in voice call content is filtered, the performance of semantic and intention analysis is improved, preset intention worksheet data relevant to the user intention to be identified are screened out by utilizing a preset keyword library, the corresponding user intention is judged by utilizing a user intention identification model which is established in advance and trained and based on a BI-L STM and an attention mechanism, the attention capacity of the model to key characteristic information and the semantic fusion and processing capacity to complex sentence patterns are improved, accurate understanding of text meaning is realized, and the identification capacity to the user intention is improved.
Specifically, in the following embodiments, the customer intention identification method based on the attention mechanism, the feature embedding and the BI-L STM is described in detail by taking the example that the unicom judges the customer number portability intention according to the contents of the customer service call.
As shown in fig. 2, according to the above-mentioned method for identifying the client intention based on the attention mechanism, the feature embedding and the BI-L STM, the unicom company can identify the user intention through the user intention identification model according to the historical call data between the user and the customer service, and interface with the market department, so as to provide a relatively scientific data support for the decision, promote the accurate marketing of the number portability client, and reduce the scale of the number portability client by controlling through a big data means.
Specifically, the voice to text of the customer service call content and the original data of the complaint work order are a plurality of small files, the small files are combined, duplication of the data is removed according to the unique identifier (the unique identifier of the customer service voice to text is a contact number, and the complaint work order is a work order serial number), and the work order data in which the unique identifier is not in a uniform format, the telephone number is null, and characters except numbers exist is deleted.
As shown in fig. 3, in the above embodiment, preferably, a near-sense word model is used to determine a near-sense word related to a keyword, and the near-sense word model is based on a deep learning model based on word2vec, combines semantic similarity, pinyin similarity and editing distance, and provides a semantic near-sense word recognition result exceeding the word face of the keyword, so that the keyword and the recognized near-sense word together construct a preset keyword library.
Specifically, firstly, a seed keyword ' number portability and network switching ' is set, content and work order data are crawled on the internet, a number portability and network switching ' synonym word, such as ' slant number portability ', ' business change hall ', ' network switching no-change ', ' operator change ' and the like, is obtained by using a synonym word model, and the work order data of the consulting number portability and network switching service are screened out by using a preset keyword library. The near-meaning word model is based on a deep learning model based on word2vec, multi-dimensional information such as semantic similarity, pinyin similarity and editing distance is fused, a semantic near-meaning word recognition result exceeding the word surface is provided, and the deep near-meaning word model is constructed.
As shown in FIG. 4, in the above embodiment, preferably, the method for establishing and training the user intention recognition model based on the BI-L STM and the attention mechanism includes determining keywords and related synonyms according to the user intention to be recognized, establishing a preset keyword library, performing keyword screening on the content of the customer service call by using the preset keyword library, manually labeling the user intention labels by using the worksheet data of the screened keywords as a training set, establishing the user intention recognition model by using the BI-L STM in combination with the network structure of the attention mechanism, inputting the worksheet data in the training set into the user intention recognition model, and training the user intention recognition model by using the manually labeled user intention labels.
The method comprises the steps of obtaining keywords related to number portability to be identified based on a near-synonym model, constructing a preset keyword library, extracting key information in customer service call content, and accordingly achieving the purpose of filtering irrelevant information in the call content, extracting a plurality of pieces of call content related to number portability service in each customer service worksheet, judging user intents implicit in the key information in each piece of call content according to a user intention identification model, manually labeling important information extracted according to the keywords, using a labeling result as training data of the user intention identification model, obtaining 10000 pieces of labeling data, dividing a training set, a verification set and a test set according to a ratio of 7:2:1, and training by using the training set by using a BI-L STM + Attention network structure of the user intention identification model.
As shown in fig. 5, on the basis that the training of the user intention recognition model is completed in the above embodiment, preferably, the method for determining the user intention corresponding to the preset intention worksheet data by using the trained user intention recognition model includes: judging a user intention label hidden in each preset intention work order data through a user intention recognition model; and determining the user intention corresponding to each piece of preset intention work order data according to the type and the number of the user intention labels in each piece of preset intention work order data and the voting principle and the original state of the user.
Specifically, for a plurality of pieces of key information extracted from preset intention work order data, the user intention labels (turning-in, turning-out and other) hidden in each piece of key information are judged through a user intention identification model, and finally, the final turning-in and turning-out intentions of the user are judged through a voting principle, namely a minority obeying majority, and by combining the network access information of the user corresponding to the preset intention work order data. And when all the tags are other, the final tags of the worksheet data are other, when the number of the transferred-in tags and the transferred-out tags is not equal, according to the voting principle, a minority obeys majority, and when the number of the transferred-in tags and the number of the transferred-out tags are equal, judging according to the network access information of the user that the different network user is transferred in, and the local network user is transferred out.
In the above embodiment, as shown in fig. 6, preferably, in the user intention recognition model, word vector information and feature information of the work order data are used as input of the BI-L STM neural network, the hidden state output of the BI-L STM neural network is spliced with the feature information and used as input of the attention mechanism, the weight of each hidden state is calculated through normalization, and a feature representation vector of the work order data is obtained through weighted summation.
Specifically, the BI-L STM neural network is a bidirectional structure, the dependency relationship between semantics is captured from two directions, then hidden state output of the bidirectional L STM network is fused, feature information in input is spliced again to be used as input of an attention mechanism for weight calculation, important information and important features in semantics are learned by a model, and finally feature information representation of the text is obtained through weighted summation of hidden state information.
In the above embodiment, the attention mechanism is preferably a method for boosting an RNN-based Encoder model for computing a weighted sum of hidden state output vector sets of a BI-L STM neural network from feature information vectors.
As shown in fig. 7, the algorithm process of the attention mechanism is:
1) the encode of the input sequence obtains the state c of the last time step and the output h of each time step, wherein c is used as the initial state z0 of the decode;
2) outputs h and z for each time step0Matching, namely match operation is carried out to obtain a matching vector of each time step
3) Outputs h and z for all time steps0Matching degree of α0Using softmax to carry out normalization processing to obtain z corresponding to each time step0The matching score of (2);
4) and obtaining c0 by weighted summation of the output h of each time step and the matching score.
In the above embodiment, preferably, the customer intention identification method based on attention mechanism, feature embedding and BI-L STM further includes label fusion of the user intention corresponding to the preset intention worksheet data and the worksheet information corresponding to the preset intention worksheet data.
According to the client intention identification method based on the attention mechanism, the feature embedding and the BI-L STM, provided by the embodiment, in the data analysis of the number portability service of the Unicom company, the work orders of the number portability service selected for consultation in Beijing areas are screened and intention judgment is carried out, the verification result after marking the important information extracted from the work order data is shown in table 1, and the verification result of the final transfer-out intention judgment of the user is shown in table 2.
TABLE 1 extraction of key part test results
| Number of test sets | Correct number of | Accuracy rate | |
| Is turned into | 268 | 231 | 86.19％ |
| Is rolled out | 452 | 433 | 95.80％ |
| Others | 281 | 246 | 87.54％ |
| Total of | 1001 | 910 | 91％ |
TABLE 2 post-integration test results
Therefore, the core idea of the rule-based method is fused into the semantic analysis deep neural network, the target text segment is accurately positioned, irrelevant interference information in call content is filtered, and the semantic and intention analysis performance is improved. By introducing the characteristic information embedding, relevant information such as parts of speech, characteristic words, user levels and the like is added into the user intention identification model, and meanwhile, an attention mechanism is introduced, so that the attention capacity of the user intention identification model to key characteristic information and the semantic fusion and processing capacity of a complex sentence pattern are improved, and the identification capacity of the user intention identification model to the user intention is improved.
In the implementation process of the client intention recognition method based on the attention mechanism, the feature embedding and the BI-L STM provided by the embodiment, the method specifically includes:
1. data cleaning: carrying out file merging, duplicate removal, null data deletion and the like;
2. and (3) screening data: expanding service keywords through a near-meaning word model, constructing a preset keyword library, and screening work order data related to the service according to the preset keyword library;
3. and (3) predicting an intention recognition model: extracting key information of a work order to be predicted according to a business keyword lexicon constructed by a near meaning word model, and performing intention prediction by adopting a trained model;
4. integrating the model prediction results: for a plurality of pieces of key information extracted from one work order, the user intention implicit in each piece of information is judged through a user intention identification model, and finally the final intention of the user is judged through a voting principle, namely minority obeying majority and combining with the current state information of the user.
The above is only a preferred embodiment of the present invention, and is not intended to limit the present invention, and various modifications and changes will occur to those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (8)
1. A customer intention recognition method based on attention mechanism, feature embedding and BI-L STM is characterized by comprising the following steps:
acquiring customer service call content, and cleaning and denoising the customer service call content;
screening out preset intention work order data by utilizing a preset keyword library;
and judging the user intention corresponding to the preset intention worksheet data by utilizing a pre-established and trained user intention recognition model based on the BI-L STM and the attention mechanism.
2. The method for client intention recognition based on attention mechanism, feature embedding and BI-L STM as claimed in claim 1, wherein the method for establishing and training the user intention recognition model based on BI-L STM and attention mechanism comprises:
determining keywords and related similar words according to the user intention to be identified, and constructing a preset keyword library;
performing keyword screening on the customer service call content by using the preset keyword library;
taking the work order data with the screened keywords as a training set, and manually marking the user intention label;
constructing a user intention recognition model by a network structure of a BI-L STM combined with an attention mechanism;
and inputting the work order data in the training set into the user intention recognition model, and training the user intention recognition model through the manually marked user intention label.
3. The client intention recognition method based on attention mechanism, feature embedding and BI-L STM as claimed in claim 1, wherein the method for determining the user intention corresponding to the preset intention worksheet data by using the user intention recognition model comprises:
judging a user intention label hidden in each piece of preset intention work order data through the user intention identification model;
and determining the user intention corresponding to each piece of preset intention worksheet data according to the type and the number of the user intention labels in each piece of preset intention worksheet data and through a voting principle and the original state of the user.
4. The method for client intention recognition based on attention mechanism, feature embedding and BI-L STM as claimed in claim 2, wherein a near-meaning word model is used to determine the near-meaning words related to the keywords, the near-meaning word model is based on a deep learning model based on word2vec, and combines semantic similarity, pinyin similarity and editing distance to provide a semantic near-meaning word recognition result exceeding the word surface of the keywords, so as to construct the preset keyword library together with the keywords and the recognized near-meaning words.
5. The customer intention recognition method based on attention mechanism, feature embedding and BI-L STM according to claim 1 or 2, characterized in that in the user intention recognition model, word vector information and feature information of the work order data are used as input of a BI-L STM neural network, hidden state output of the BI-L STM neural network is spliced with the feature information and used as input of the attention mechanism, the weight of each hidden state is calculated through normalization, and feature representation vectors of the work order data are obtained through weighted summation.
6. The method of claim 5, wherein the attention mechanism is an Encoder model for computing a weighted sum of hidden state output vector sets of the BI-L STM neural network from the feature information vectors.
7. The method for customer intent recognition based on attention mechanism, feature embedding and BI-L STM as claimed in claim 1, wherein said cleaning of customer service call content specifically comprises file merging, deduplication and deletion of null data.
8. The method for client intent recognition based on attention mechanism, feature embedding and BI-L STM as claimed in claim 1, further comprising:
and performing label fusion on the user intention corresponding to the preset intention work order data and the work order information corresponding to the preset intention work order data.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010251934.8A CN111462752B (en) | 2020-04-01 | 2020-04-01 | Attention mechanism, feature embedding and BI-LSTM (business-to-business) based customer intention recognition method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010251934.8A CN111462752B (en) | 2020-04-01 | 2020-04-01 | Attention mechanism, feature embedding and BI-LSTM (business-to-business) based customer intention recognition method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111462752A true CN111462752A (en) | 2020-07-28 |
| CN111462752B CN111462752B (en) | 2023-10-13 |
Family
ID=71681173
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010251934.8A Active CN111462752B (en) | 2020-04-01 | 2020-04-01 | Attention mechanism, feature embedding and BI-LSTM (business-to-business) based customer intention recognition method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111462752B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112507696A (en) * | 2021-02-04 | 2021-03-16 | 湖南大学 | Human-computer interaction diagnosis guiding method and system based on global attention intention recognition |
| CN112949317A (en) * | 2021-02-26 | 2021-06-11 | 平安普惠企业管理有限公司 | Text semantic recognition method and device, computer equipment and storage medium |
| CN113190656A (en) * | 2021-05-11 | 2021-07-30 | 南京大学 | Chinese named entity extraction method based on multi-label framework and fusion features |
| CN115114407A (en) * | 2022-07-12 | 2022-09-27 | 平安科技（深圳）有限公司 | Intention recognition method and device, computer equipment and storage medium |
Citations (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108681562A (en) * | 2018-04-26 | 2018-10-19 | 第四范式（北京）技术有限公司 | Category classification method and system and Classification Neural training method and device |
| CN108805088A (en) * | 2018-06-14 | 2018-11-13 | 南京云思创智信息科技有限公司 | Physiological signal analyzing subsystem based on multi-modal Emotion identification system |
| CN108805087A (en) * | 2018-06-14 | 2018-11-13 | 南京云思创智信息科技有限公司 | Semantic temporal fusion association based on multi-modal Emotion identification system judges subsystem |
| CN108829667A (en) * | 2018-05-28 | 2018-11-16 | 南京柯基数据科技有限公司 | It is a kind of based on memory network more wheels dialogue under intension recognizing method |
| CN108874782A (en) * | 2018-06-29 | 2018-11-23 | 北京寻领科技有限公司 | A kind of more wheel dialogue management methods of level attention LSTM and knowledge mapping |
| CN109101537A (en) * | 2018-06-27 | 2018-12-28 | 北京慧闻科技发展有限公司 | More wheel dialogue data classification methods, device and electronic equipment based on deep learning |
| CN109101493A (en) * | 2018-08-01 | 2018-12-28 | 东北大学 | A kind of intelligence house-purchase assistant based on dialogue robot |
| CN109241255A (en) * | 2018-08-20 | 2019-01-18 | 华中师范大学 | A kind of intension recognizing method based on deep learning |
| CN109741751A (en) * | 2018-12-11 | 2019-05-10 | 上海交通大学 | Intension recognizing method and device towards intelligent sound control |
| CN110096570A (en) * | 2019-04-09 | 2019-08-06 | 苏宁易购集团股份有限公司 | A kind of intension recognizing method and device applied to intelligent customer service robot |
| CN110110059A (en) * | 2019-05-20 | 2019-08-09 | 挂号网（杭州）科技有限公司 | A kind of medical conversational system intention assessment classification method based on deep learning |
| CN110162610A (en) * | 2019-04-16 | 2019-08-23 | 平安科技（深圳）有限公司 | Intelligent robot answer method, device, computer equipment and storage medium |
| CN110209791A (en) * | 2019-06-12 | 2019-09-06 | 百融云创科技股份有限公司 | It is a kind of to take turns dialogue intelligent speech interactive system and device more |
| CN110287283A (en) * | 2019-05-22 | 2019-09-27 | 中国平安财产保险股份有限公司 | Intent model training method, intension recognizing method, device, equipment and medium |
| CN110532355A (en) * | 2019-08-27 | 2019-12-03 | 华侨大学 | A kind of intention based on multi-task learning combines recognition methods with slot position |
| CN110570853A (en) * | 2019-08-12 | 2019-12-13 | 阿里巴巴集团控股有限公司 | Intention recognition method and device based on voice data |
| CN110827831A (en) * | 2019-11-15 | 2020-02-21 | 广州洪荒智能科技有限公司 | Voice information processing method, device, equipment and medium based on man-machine interaction |
| CN110866099A (en) * | 2019-10-30 | 2020-03-06 | 南昌众荟智盈信息技术有限公司 | Intelligent steward service method and system based on intelligent sound box voice interaction |
| CN110909543A (en) * | 2019-11-15 | 2020-03-24 | 广州洪荒智能科技有限公司 | Intention recognition method, device, equipment and medium |
| CN110928997A (en) * | 2019-12-04 | 2020-03-27 | 北京文思海辉金信软件有限公司 | Intention recognition method and device, electronic equipment and readable storage medium |
-
2020
- 2020-04-01 CN CN202010251934.8A patent/CN111462752B/en active Active
Patent Citations (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108681562A (en) * | 2018-04-26 | 2018-10-19 | 第四范式（北京）技术有限公司 | Category classification method and system and Classification Neural training method and device |
| CN108829667A (en) * | 2018-05-28 | 2018-11-16 | 南京柯基数据科技有限公司 | It is a kind of based on memory network more wheels dialogue under intension recognizing method |
| CN108805088A (en) * | 2018-06-14 | 2018-11-13 | 南京云思创智信息科技有限公司 | Physiological signal analyzing subsystem based on multi-modal Emotion identification system |
| CN108805087A (en) * | 2018-06-14 | 2018-11-13 | 南京云思创智信息科技有限公司 | Semantic temporal fusion association based on multi-modal Emotion identification system judges subsystem |
| CN109101537A (en) * | 2018-06-27 | 2018-12-28 | 北京慧闻科技发展有限公司 | More wheel dialogue data classification methods, device and electronic equipment based on deep learning |
| CN108874782A (en) * | 2018-06-29 | 2018-11-23 | 北京寻领科技有限公司 | A kind of more wheel dialogue management methods of level attention LSTM and knowledge mapping |
| CN109101493A (en) * | 2018-08-01 | 2018-12-28 | 东北大学 | A kind of intelligence house-purchase assistant based on dialogue robot |
| CN109241255A (en) * | 2018-08-20 | 2019-01-18 | 华中师范大学 | A kind of intension recognizing method based on deep learning |
| CN109741751A (en) * | 2018-12-11 | 2019-05-10 | 上海交通大学 | Intension recognizing method and device towards intelligent sound control |
| CN110096570A (en) * | 2019-04-09 | 2019-08-06 | 苏宁易购集团股份有限公司 | A kind of intension recognizing method and device applied to intelligent customer service robot |
| CN110162610A (en) * | 2019-04-16 | 2019-08-23 | 平安科技（深圳）有限公司 | Intelligent robot answer method, device, computer equipment and storage medium |
| CN110110059A (en) * | 2019-05-20 | 2019-08-09 | 挂号网（杭州）科技有限公司 | A kind of medical conversational system intention assessment classification method based on deep learning |
| CN110287283A (en) * | 2019-05-22 | 2019-09-27 | 中国平安财产保险股份有限公司 | Intent model training method, intension recognizing method, device, equipment and medium |
| CN110209791A (en) * | 2019-06-12 | 2019-09-06 | 百融云创科技股份有限公司 | It is a kind of to take turns dialogue intelligent speech interactive system and device more |
| CN110570853A (en) * | 2019-08-12 | 2019-12-13 | 阿里巴巴集团控股有限公司 | Intention recognition method and device based on voice data |
| CN110532355A (en) * | 2019-08-27 | 2019-12-03 | 华侨大学 | A kind of intention based on multi-task learning combines recognition methods with slot position |
| CN110866099A (en) * | 2019-10-30 | 2020-03-06 | 南昌众荟智盈信息技术有限公司 | Intelligent steward service method and system based on intelligent sound box voice interaction |
| CN110827831A (en) * | 2019-11-15 | 2020-02-21 | 广州洪荒智能科技有限公司 | Voice information processing method, device, equipment and medium based on man-machine interaction |
| CN110909543A (en) * | 2019-11-15 | 2020-03-24 | 广州洪荒智能科技有限公司 | Intention recognition method, device, equipment and medium |
| CN110928997A (en) * | 2019-12-04 | 2020-03-27 | 北京文思海辉金信软件有限公司 | Intention recognition method and device, electronic equipment and readable storage medium |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112507696A (en) * | 2021-02-04 | 2021-03-16 | 湖南大学 | Human-computer interaction diagnosis guiding method and system based on global attention intention recognition |
| CN112507696B (en) * | 2021-02-04 | 2021-04-20 | 湖南大学 | Human-computer interaction diagnosis guiding method and system based on global attention intention recognition |
| CN112949317A (en) * | 2021-02-26 | 2021-06-11 | 平安普惠企业管理有限公司 | Text semantic recognition method and device, computer equipment and storage medium |
| CN113190656A (en) * | 2021-05-11 | 2021-07-30 | 南京大学 | Chinese named entity extraction method based on multi-label framework and fusion features |
| CN113190656B (en) * | 2021-05-11 | 2023-07-14 | 南京大学 | Chinese named entity extraction method based on multi-annotation frame and fusion features |
| CN115114407A (en) * | 2022-07-12 | 2022-09-27 | 平安科技（深圳）有限公司 | Intention recognition method and device, computer equipment and storage medium |
| CN115114407B (en) * | 2022-07-12 | 2024-04-19 | 平安科技（深圳）有限公司 | Intention recognition method, device, computer equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111462752B (en) | 2023-10-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110427623B (en) | Semi-structured document knowledge extraction method and device, electronic equipment and storage medium | |
| CN110096570B (en) | Intention identification method and device applied to intelligent customer service robot | |
| CN109165294B (en) | Short text classification method based on Bayesian classification | |
| CN112784578B (en) | Legal element extraction method and device and electronic equipment | |
| CN111462752B (en) | Attention mechanism, feature embedding and BI-LSTM (business-to-business) based customer intention recognition method | |
| CN111783394A (en) | Training method of event extraction model, event extraction method, system and equipment | |
| CN108549723B (en) | Text concept classification method and device and server | |
| CN113961685A (en) | Information extraction method and device | |
| CN110599324A (en) | Method and device for predicting refund rate | |
| CN112434164B (en) | Network public opinion analysis method and system taking topic discovery and emotion analysis into consideration | |
| CN111160019A (en) | Public opinion monitoring method, device and system | |
| CN115310443A (en) | Model training method, information classification method, device, equipment and storage medium | |
| CN112069312A (en) | Text classification method based on entity recognition and electronic device | |
| CN112287090A (en) | Financial question asking back method and system based on knowledge graph | |
| CN116070632A (en) | Informal text entity tag identification method and device | |
| CN107818173B (en) | Vector space model-based Chinese false comment filtering method | |
| CN116010581A (en) | Knowledge graph question-answering method and system based on power grid hidden trouble shooting scene | |
| CN111782793A (en) | Intelligent customer service processing method, system and equipment | |
| CN112579730A (en) | High-expansibility multi-label text classification method and device | |
| CN114239579A (en) | Electric power searchable document extraction method and device based on regular expression and CRF model | |
| CN113486174A (en) | Model training, reading understanding method and device, electronic equipment and storage medium | |
| Swamy et al. | Nit-agartala-nlp-team at semeval-2020 task 8: Building multimodal classifiers to tackle internet humor | |
| CN110888983A (en) | Positive and negative emotion analysis method, terminal device and storage medium | |
| TW202034207A (en) | Dialogue system using intention detection ensemble learning and method thereof | |
| CN115713085A (en) | Document theme content analysis method and device |
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 |