CN109299253A - A kind of social text Emotion identification model construction method of Chinese based on depth integration neural network - Google Patents
A kind of social text Emotion identification model construction method of Chinese based on depth integration neural network Download PDFInfo
- Publication number
- CN109299253A CN109299253A CN201811016921.1A CN201811016921A CN109299253A CN 109299253 A CN109299253 A CN 109299253A CN 201811016921 A CN201811016921 A CN 201811016921A CN 109299253 A CN109299253 A CN 109299253A
- Authority
- CN
- China
- Prior art keywords
- text
- layer
- chinese
- mood
- data
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/289—Phrasal analysis, e.g. finite state techniques or chunking
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The social text Emotion identification model construction method of the Chinese that the invention discloses a kind of based on depth integration neural network, includes the following steps: that data acquire, and using Python Scrapy framework establishment social activity text web crawlers, acquires image, text and data；Data prediction pre-processes the Chinese text of data collecting module collected；Data mark, for text to carry out mood mark to treated；Text vector, with Word2Vec tool training term vector；Model construction, design fusion BILSTM-CNN network model；Text after mark is trained by model training by BILSTM-CNN fused neural network model.The present invention constructs a kind of depth integration mood analysis model, it is intended to which the feature extraction ability for making full use of deep neural network model carries out feature representation to Chinese mood text, and constructs the more disaggregated models of mood with this, improves the automation polytypic accuracy rate of mood.
Description
Technical field
The present invention relates to natural language processing technique fields, and in particular to a kind of Chinese based on depth integration neural network
Social text Emotion identification model construction method.
Background technique
Mood analysis belongs to sentiment analysis class problem.Sentiment analysis (SA) is also known as proneness analysis and opinion mining, it
It is that the subjective texts with emotional color are analyzed, are handled, are concluded and the process of reasoning.Sentiment analysis can be applied to electricity
Sub- commercial affairs, the various fields such as brand reputation management, the analysis of public opinion.With popularizing for the social medias such as microblogging, user discusses oneself
The products & services used, or oneself politics and religion viewpoint are expressed, microblogging website has become people's comment and believes with emotion
The valuable source of breath.The extensive concern that sentiment analysis has been subjected to researcher is done to such data now.
So far, the analysis and research of most of microblog emotional are all only focused in how analyzing English text information,
And based on being analyzed with feeling polarities.Lack the analysis Chinese text emotional characteristics more refined in the prior art, analyzes convolution
Therefore the characteristics of neural network and long memory network in short-term, is urgently studied at present how using deep learning Fusion Model, realizes
Preferable Chinese mood classifying quality.
Summary of the invention
The purpose of the present invention is to solve drawbacks described above in the prior art, provide a kind of based on depth integration nerve net
The social text Emotion identification model construction method of the Chinese of network.The characteristics of this method be merged two-way length in short-term memory network with
The characteristics of convolutional neural networks, is indicated using the two-way length global characteristics that memory network completes text in short-term, recycles convolution mind
Local feature through network extracts the emotional characteristics of characterization text, and the method achieves higher standard on mood categorized data set
True rate.
The purpose of the present invention can be reached by adopting the following technical scheme that:
A kind of social text Emotion identification model construction method of Chinese based on depth integration neural network, the construction
Method includes:
Data collection steps, for acquiring Chinese text data from social network data source；
Text Pretreatment step handles collected urtext data；
Text mood annotation step carries out the more classification markers of mood to pretreated data；
Text vector step carries out the training of Chinese term vector by distributed term vector representation method Word2Vec；
Model construction step, initialization model structure construct more taxonomic structures based on neural network fusion model；
Model training step, CNN-BILSTM converged network model of the training for mood analysis of more classifying, obtains final
Mood disaggregated model.
Further, in the data collection steps, using towards multi-threaded crawler capturing network mood text, and
Chinese text therein is stored.
Further, the Text Pretreatment step process is as follows:
Remove the English data in text；
Emoji and hyperlink in text are removed, emoji in text is replaced with into its simple Chinese text, it will be in text
Hyperlink replaces with Chinese " link "；
Text stop words is removed according to the deactivated dictionary of Chinese.
Further, in the text mood annotation step, using disclosed in the data and part partially manually marked
Mood is divided into hobby, fear, indignation, detest, sadness, happiness, surprised seven mood classes in artificial annotation process by data
Not, Various types of data respectively takes 2500, and training set and test set are finally taken to 80% and 20% data, used function respectively
It is 0.2 for train_test_split, parameter test_size.
Further, it in the text vector step, is constructed using distributed term vector representation method Word2Vec
Term vector model sets 350 for output term vector dimension, and training data is by Chinese Wiki corpus and collected mood language
Material is used as training sample together.
Further, in the model construction step, construct Fusion Model, extract text emotional characteristics, using CNN
(convolutional neural networks) and BILSTM (two-way length in short-term memory network) are merged, and extract affective characteristics.
Further, in the model construction step, CNN (convolutional neural networks) and BILSTM are built using Keras
(two-way length in short-term memory network) fusion deep neural network, CNN (convolutional neural networks) and the BILSTM (remember in short-term by two-way length
Recall network) fusion deep neural network structure it is as follows:
First layer is embeding layer, and the input of this layer is trained term vector sequence, and the present invention is by text sequence length (word
Sequence vector number) it is set as 100, being filled with 0 less than 100, the truncation more than 100.Embeding layer is set using pre-training
Term vector；
The second layer be it is LSTM layers two-way, this layer output be 100*200；
Third layer is the first convolutional layer, and the two-dimensional matrix size of input is 100*200, and having size is 32 of 4 × 4 pixels
Filter, step-length 1, activation primitive are set as ReLU function；
4th layer is the first pond layer, uses maximum pool model for MaxPooling2D, and parameter Poolsize is (3,3)；
Layer 5 is the second convolutional layer, uses size for 32 filters of 3 × 3 pixels, and the activation primitive used is
ReLU function；
Layer 6 is the second pond layer, is (2,2) using maximum pond MaxPooling2D, parameter poolsize；
Layer 7 is Dropout layers, and parameter rate is set as 0.3, prevents over-fitting；
8th layer is Flatten layers, the input one-dimensional of multidimensional；
9th layer is the first full articulamentum, the vector after inputting the output expansion of a upper neural net layer, 500 dimension of output
Vector, the activation primitive used are ReLU；
Tenth layer is the second full articulamentum, and input is the input vector of 500 dimensions, this layer is two neurons, that is, exports two
Dimension data, the activation primitive of use are ReLU；
Eleventh floor is Softmax layers of classifier, generates classification results by Softmax classifier.The output of this layer is feelings
The classification number of thread classification is 7.
Further, in the model training step, the loss letter that is used in the Chinese mood text data set of training
Number is categorical_crossentropy, and optimizer adam, batch size batch_size are 100, the number of iterations
Epoch is 15.
The present invention has the following advantages and effects with respect to the prior art:
1, pre-training Word2Vec is used for text distribution term vector by the present invention indicates, training data combines a large amount of feelings
Thread is expected and Chinese Wiki is it is anticipated that preferably indicate the feature of semanteme of text.
2, the present invention has merged two-way length memory network ability sequence indicates in short-term advantage and convolutional neural networks and has existed
Advantage on local shape factor proposes a kind of deep neural network model for Chinese Emotion identification.
Detailed description of the invention
Fig. 1 is the social text Emotion identification Construction of A Model of Chinese disclosed in the present invention based on depth integration neural network
The flow diagram of method；
Fig. 2 is the collecting method logic chart in the present invention；
Fig. 3 is the two-way LSTM figure indicated for global characteristics in the present invention；
Fig. 4 is the CNN model structure that the local feature in the present invention extracts；
Fig. 5 is the Chinese Emotion identification illustraton of model in the present invention.
Specific embodiment
In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with the embodiment of the present invention
In attached drawing, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described embodiment is
A part of the embodiment of the present invention, instead of all the embodiments.Based on the embodiments of the present invention, those of ordinary skill in the art
Every other embodiment obtained without making creative work, shall fall within the protection scope of the present invention.
Embodiment
The method of affection computation is based primarily upon dictionary/rule method and based on statistical learning/deep learning method,
The embodiment of the present invention carries out sentiment analysis by the method based on deep learning, utilizes Word2Vec technique drill term vector, benefit
The analytical calculation of emotion is carried out with BILSTM-CNN converged network.Depth integration mood analysis model is excavated and learning text feelings
The characteristics of thread indicates, and then depth extracts the mood semanteme of text, improves the accuracy rate of Emotion identification.
The social text Emotion identification model construction method of Chinese with reference to the accompanying drawings shown in 1 based on depth integration neural network
Flow diagram, the social text Emotion identification model of the Chinese disclosed by the embodiments of the present invention based on depth integration neural network
Building method the following steps are included:
Data collection steps, for acquiring Chinese text data from the social network datas such as microblogging source；
Text Pretreatment step handles collected urtext data；
Text mood annotation step carries out the more classification markers of mood to pretreated data；
Text vector step carries out the training of Chinese term vector by distributed term vector representation method Word2Vec；
Model construction step, initialization model structure construct more taxonomic structures based on neural network fusion model；
Model training step, CNN-BILSTM converged network model of the training for mood analysis of more classifying, obtains final
Mood disaggregated model.
In data collection steps, concrete scheme is to be made using crawler frame Scrapy, Scrapy under python2.7 version
Network communication is handled with Twisted asynchronous network library.Such as the collecting method logic chart in attached drawing 2, the control of this crawler
Each step of device management crawler processed, url to be crawled is managed with dictionary data structure, for persistence step, is directly stored in
In Mysql database.Query grammar of the resolver using Scrapy, the simpler query grammar of Scrapy internal support,
Help the label that needs are inquired in html and label substance and tag attributes.
In Text Pretreatment step, text mood disclosed in additional downloads subnetwork of the present invention analyzes data (such as
NLPCC2013 data set).Processing for text data, the first English data in removal text.Followed by for emoji
With the processing of hyperlink, the present embodiment specific method be emoji is replaced with its simple Chinese text such as (laugh at and cry, it is awkward
Deng), hyperlink deletion is simply indicated as Chinese " link ".Text stop words is finally removed according to the deactivated dictionary of Chinese.
In text mood annotation step, present invention employs data disclosed in the data partially manually marked and part.
In annotation process, mood is divided into seven mood classifications such as hobby, fear, indignation, detest, sadness, happiness, surprised, all kinds of numbers
According to respectively taking 2500.Training set and test set are finally taken to 80% and 20% data respectively, used function is train_
Test_split, parameter test_size are 0.2.
In text vector step, using Word2Vec method, the present embodiment using collected text data,
Microblogging corpus disclosed in network and Chinese Wiki corpus training term vector, and be arranged the dimension size of output term vector with
The cnn characteristic dimension of front is consistent, is set as 350.Detailed process is segmented first with participle tool, such as jieba points
Word obtains participle corpus corpsw2v.txt then import word2vec, uses word2vec function therein after participle
Text representation in data set is saved at term vector.
BILSTM model in this method, as shown in figure 3, the Chinese text indicated by the distributed vector of training, building
Memory network model filters out the time of 0 vector using embeding layer Embedding before BILSTM network to two-way length in short-term
Step has been processed into equal length by pad_sequences because the sentence length of input is different.After the processing of this layer, to text
This word sequence has done a degree of feature representation, and the sequence vector that this network layer is abstracted both considers text word
Correlation properties, it is also considered that the word order relationship of word, this is also the important advantage of BILSTM neural network.
In convolution process, the present invention uses CNN convolutional neural networks as shown in Figure 4, convolutional layer part of the invention
Totally nine layer network, respectively convolutional layer, maximum pond layer, convolutional layer, maximum pond layer, Dropout layers, Flatten layers, Quan Lian
Layer, full articulamentum and Softmax classification layer is connect to be sequentially connected according to Fig. 4.
By the training dataset training pattern of the data set of downloading and oneself mark, the CNN net of eigen extraction is obtained
Network.It can classify after convolution, this embodiment has selected Softmax activation primitive to classify, this model exports picture and text media
Positive and negative category feature.Here loss function cross entropy loss function categorical_crossentropy, optimization method are used
Adam.By adjusting the value of other hyper parameters, saved after obtaining preferable model, for the mood analysis to unknown data, and
Test etc..
Fig. 5 is the Chinese Emotion identification model that this method is established, in figure, input X etc. for insertion text term vector,
By extracting feature by two layers of convolutional network after BILSTM coding, classify finally by full connection and classifier.?
In this model, two-way BILSTM layers is used to encode urtext sequence, each level in sequence after convolutional layer extraction coding
Feature.In treatment process, Chinese mood text need to be pre-processed by step of the invention, and segmented, vectorization
Deng operation, it is then input to identification mood classification in model.In test process, accuracy rate, recall rate, the parameters such as F1 value can be passed through
Scoring model ability.
In conclusion for the characteristics of picture and text media, this method emphasis passes through in the media such as microblogging, wechat circle of friends now
BILSTM preliminary characterization text feature, two-way length Dependency Specification when memory network preferably can handle long in short-term, to text
During carrying out character representation, past contextual information had both been considered, it is also considered that following contextual information.Then it passes through
CNN convolutional neural networks are crossed by different convolution kernels, preferably take out local feature of the data on each level.Feelings are allowed with this
Thread analysis model more fully mining data feature improves the effect of mood classification.
The above embodiment is a preferred embodiment of the present invention, but embodiments of the present invention are not by above-described embodiment
Limitation, other any changes, modifications, substitutions, combinations, simplifications made without departing from the spirit and principles of the present invention,
It should be equivalent substitute mode, be included within the scope of the present invention.
Claims (8)
1. a kind of social text Emotion identification model construction method of Chinese based on depth integration neural network, which is characterized in that
The building method includes:
Data collection steps, for acquiring Chinese text data from social network data source；
Text Pretreatment step handles collected urtext data；
Text mood annotation step carries out the more classification markers of mood to pretreated data；
Text vector step carries out the training of Chinese term vector by distributed term vector representation method Word2Vec；
Model construction step, initialization model structure construct more taxonomic structures based on neural network fusion model；
Model training step, CNN-BILSTM converged network model of the training for mood analysis of more classifying, obtains final feelings
Thread disaggregated model.
2. the social text Emotion identification model structure of a kind of Chinese based on depth integration neural network according to claim 1
Make method, which is characterized in that in the data collection steps, using towards multi-threaded crawler capturing network mood text,
And Chinese text therein is stored.
3. the social text Emotion identification model structure of a kind of Chinese based on depth integration neural network according to claim 1
Make method, which is characterized in that the Text Pretreatment step process is as follows:
Remove the English data in text；
Emoji and hyperlink in text are removed, emoji in text is replaced with into its simple Chinese text, by hyperlink in text
It takes over and is changed to Chinese " link "；
Text stop words is removed according to the deactivated dictionary of Chinese.
4. the social text Emotion identification model structure of a kind of Chinese based on depth integration neural network according to claim 1
Make method, which is characterized in that in the text mood annotation step, using disclosed in the data and part partially manually marked
Mood is divided into hobby, fear, indignation, detest, sadness, happiness, surprised seven mood classes in artificial annotation process by data
Not, Various types of data respectively takes 2500, and training set and test set are finally taken to 80% and 20% data, used function respectively
It is 0.2 for train_test_split, parameter test_size.
5. the social text Emotion identification model structure of a kind of Chinese based on depth integration neural network according to claim 1
Make method, which is characterized in that in the text vector step, construct using distributed term vector representation method Word2Vec
Term vector model sets 350 for output term vector dimension, and training data is by Chinese Wiki corpus and collected mood language
Material is used as training sample together.
6. the social text Emotion identification model structure of a kind of Chinese based on depth integration neural network according to claim 1
Make method, which is characterized in that in the model construction step, construct Fusion Model, extract text emotional characteristics, using CNN
It is merged with BILSTM, extracts affective characteristics.
7. the social text Emotion identification model structure of a kind of Chinese based on depth integration neural network according to claim 6
Make method, which is characterized in that in the model construction step, CNN is built using Keras and merges depth nerve net with BILSTM
Network, it is as follows which merges deep neural network structure with BILSTM:
First layer is embeding layer, and the input of this layer is trained term vector sequence, text sequence length is set as 100, no
Foot 100 is filled with 0, the truncation more than 100, and sets the term vector that embeding layer uses pre-training；
The second layer be it is LSTM layers two-way, this layer output be 100*200；
Third layer is the first convolutional layer, and the two-dimensional matrix size of input is 100*200, there is 32 filterings that size is 4 × 4 pixels
Device, step-length 1, activation primitive are set as ReLU function；
4th layer is the first pond layer, uses maximum pool model for MaxPooling2D, and parameter Poolsize is (3,3)；
Layer 5 is the second convolutional layer, uses size for 32 filters of 3 × 3 pixels, and the activation primitive used is ReLU letter
Number；
Layer 6 is the second pond layer, is (2,2) using maximum pond MaxPooling2D, parameter poolsize；
Layer 7 is Dropout layers, and parameter rate is set as 0.3；
8th layer is Flatten layers, the input one-dimensional of multidimensional；
9th layer is the first full articulamentum, the vector after inputting the output expansion of a upper neural net layer, output 500 tie up to
Amount, the activation primitive used is ReLU；
Tenth layer is the second full articulamentum, and input is the input vector of 500 dimensions, this layer is two neurons, i.e. output two-dimemsional number
According to the activation primitive of use is ReLU；
Eleventh floor is Softmax layers of classifier, generates classification results by Softmax classifier, the output of this layer is mood point
The classification number of class is 7.
8. the social text Emotion identification model structure of a kind of Chinese based on depth integration neural network according to claim 1
Make method, which is characterized in that in the model training step, the loss letter that is used in the Chinese mood text data set of training
Number is categorical_crossentropy, and optimizer adam, batch size batch_size are 100, the number of iterations
Epoch is 15.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811016921.1A CN109299253A (en) | 2018-09-03 | 2018-09-03 | A kind of social text Emotion identification model construction method of Chinese based on depth integration neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811016921.1A CN109299253A (en) | 2018-09-03 | 2018-09-03 | A kind of social text Emotion identification model construction method of Chinese based on depth integration neural network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109299253A true CN109299253A (en) | 2019-02-01 |
Family
ID=65165915
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811016921.1A Pending CN109299253A (en) | 2018-09-03 | 2018-09-03 | A kind of social text Emotion identification model construction method of Chinese based on depth integration neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109299253A (en) |
Cited By (35)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109885686A (en) * | 2019-02-20 | 2019-06-14 | 延边大学 | A kind of multilingual file classification method merging subject information and BiLSTM-CNN |
| CN110046353A (en) * | 2019-04-22 | 2019-07-23 | 重庆理工大学 | Aspect level emotion analysis method based on multi-language level mechanism |
| CN110096647A (en) * | 2019-05-10 | 2019-08-06 | 腾讯科技（深圳）有限公司 | Optimize method, apparatus, electronic equipment and the computer storage medium of quantitative model |
| CN110119681A (en) * | 2019-04-04 | 2019-08-13 | 平安科技（深圳）有限公司 | A kind of line of text extracting method and device, electronic equipment |
| CN110188933A (en) * | 2019-05-21 | 2019-08-30 | 湖北经济学院 | A kind of School Network public sentiment monitoring and pre-warning method and system |
| CN110209824A (en) * | 2019-06-13 | 2019-09-06 | 中国科学院自动化研究所 | Text emotion analysis method based on built-up pattern, system, device |
| CN110209815A (en) * | 2019-05-23 | 2019-09-06 | 国家计算机网络与信息安全管理中心 | A kind of news Users' Interests Mining method of convolutional neural networks |
| CN110276076A (en) * | 2019-06-25 | 2019-09-24 | 北京奇艺世纪科技有限公司 | A kind of text mood analysis method, device and equipment |
| CN110321563A (en) * | 2019-06-28 | 2019-10-11 | 浙江大学 | Text emotion analysis method based on mixing monitor model |
| CN110442823A (en) * | 2019-08-06 | 2019-11-12 | 北京智游网安科技有限公司 | Website classification method, Type of website judgment method, storage medium and intelligent terminal |
| CN110472245A (en) * | 2019-08-15 | 2019-11-19 | 东北大学 | A kind of multiple labeling emotional intensity prediction technique based on stratification convolutional neural networks |
| CN110473571A (en) * | 2019-07-26 | 2019-11-19 | 北京影谱科技股份有限公司 | Emotion identification method and device based on short video speech |
| CN110472053A (en) * | 2019-08-05 | 2019-11-19 | 广联达科技股份有限公司 | A kind of automatic classification method and its system towards public resource bidding advertisement data |
| CN110502757A (en) * | 2019-08-29 | 2019-11-26 | 西安邮电大学 | A kind of natural language sentiment analysis method |
| CN110532452A (en) * | 2019-07-12 | 2019-12-03 | 西安交通大学 | A kind of general crawler design method of news website based on GRU neural network |
| CN110543560A (en) * | 2019-08-08 | 2019-12-06 | 厦门市美亚柏科信息股份有限公司 | Long text classification and identification method, device and medium based on convolutional neural network |
| CN110957039A (en) * | 2019-12-16 | 2020-04-03 | 河南科技学院 | Campus psychological coaching method and device based on deep learning |
| CN111309859A (en) * | 2020-01-21 | 2020-06-19 | 上饶市中科院云计算中心大数据研究院 | Scenic spot network public praise emotion analysis method and device |
| CN111339251A (en) * | 2020-02-25 | 2020-06-26 | 上海昌投网络科技有限公司 | Method and device for detecting whether WeChat public number has sensitive words or not |
| CN111353313A (en) * | 2020-02-25 | 2020-06-30 | 四川翼飞视科技有限公司 | Emotion analysis model construction method based on evolutionary neural network architecture search |
| CN111414475A (en) * | 2020-03-03 | 2020-07-14 | 北京明略软件系统有限公司 | Text emotion information identification method and device |
| CN111523574A (en) * | 2020-04-13 | 2020-08-11 | 云南大学 | Image emotion recognition method and system based on multi-mode data |
| CN111538835A (en) * | 2020-03-30 | 2020-08-14 | 东南大学 | Social media emotion classification method and device based on knowledge graph |
| CN111651607A (en) * | 2020-07-13 | 2020-09-11 | 深圳市智搜信息技术有限公司 | Information positive and negative emotion analysis method and device, computer equipment and storage medium |
| CN111723198A (en) * | 2019-03-18 | 2020-09-29 | 北京京东尚科信息技术有限公司 | Text emotion recognition method and device and storage medium |
| CN111860981A (en) * | 2020-07-03 | 2020-10-30 | 航天信息(山东)科技有限公司 | Enterprise national industry category prediction method and system based on LSTM deep learning |
| CN111984762A (en) * | 2020-08-05 | 2020-11-24 | 中国科学院重庆绿色智能技术研究院 | Text classification method sensitive to attack resistance |
| CN112183064A (en) * | 2020-10-22 | 2021-01-05 | 福州大学 | Text emotion reason recognition system based on multi-task joint learning |
| CN112597759A (en) * | 2020-11-30 | 2021-04-02 | 深延科技(北京)有限公司 | Text-based emotion detection method and device, computer equipment and medium |
| CN112612878A (en) * | 2020-12-17 | 2021-04-06 | 大唐融合通信股份有限公司 | Customer service information providing method, electronic equipment and device |
| CN113010802A (en) * | 2021-03-25 | 2021-06-22 | 华南理工大学 | Recommendation method facing implicit feedback based on multi-attribute interaction of user and article |
| CN113191212A (en) * | 2021-04-12 | 2021-07-30 | 合肥综合性国家科学中心人工智能研究院(安徽省人工智能实验室) | Driver road rage risk early warning method and system |
| CN113297364A (en) * | 2021-06-07 | 2021-08-24 | 吉林大学 | Natural language understanding method and device for dialog system |
| CN114022909A (en) * | 2022-01-07 | 2022-02-08 | 首都师范大学 | Emotion recognition method and system based on sensor data |
| CN115146059A (en) * | 2022-06-17 | 2022-10-04 | 东方合智数据科技(广东)有限责任公司 | Raw paper market data processing method based on corrugated paper industry and related equipment |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107357889A (en) * | 2017-07-11 | 2017-11-17 | 北京工业大学 | A kind of across social platform picture proposed algorithm based on interior perhaps emotion similitude |
| CN107609009A (en) * | 2017-07-26 | 2018-01-19 | 北京大学深圳研究院 | Text emotion analysis method, device, storage medium and computer equipment |
| CN107832400A (en) * | 2017-11-01 | 2018-03-23 | 山东大学 | A kind of method that location-based LSTM and CNN conjunctive models carry out relation classification |
-
2018
- 2018-09-03 CN CN201811016921.1A patent/CN109299253A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107357889A (en) * | 2017-07-11 | 2017-11-17 | 北京工业大学 | A kind of across social platform picture proposed algorithm based on interior perhaps emotion similitude |
| CN107609009A (en) * | 2017-07-26 | 2018-01-19 | 北京大学深圳研究院 | Text emotion analysis method, device, storage medium and computer equipment |
| CN107832400A (en) * | 2017-11-01 | 2018-03-23 | 山东大学 | A kind of method that location-based LSTM and CNN conjunctive models carry out relation classification |
Cited By (48)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109885686A (en) * | 2019-02-20 | 2019-06-14 | 延边大学 | A kind of multilingual file classification method merging subject information and BiLSTM-CNN |
| CN111723198A (en) * | 2019-03-18 | 2020-09-29 | 北京京东尚科信息技术有限公司 | Text emotion recognition method and device and storage medium |
| CN111723198B (en) * | 2019-03-18 | 2023-09-01 | 北京汇钧科技有限公司 | Text emotion recognition method, device and storage medium |
| CN110119681B (en) * | 2019-04-04 | 2023-11-24 | 平安科技（深圳）有限公司 | Text line extraction method and device and electronic equipment |
| CN110119681A (en) * | 2019-04-04 | 2019-08-13 | 平安科技（深圳）有限公司 | A kind of line of text extracting method and device, electronic equipment |
| CN110046353B (en) * | 2019-04-22 | 2022-05-13 | 重庆理工大学 | Aspect level emotion analysis method based on multi-language level mechanism |
| CN110046353A (en) * | 2019-04-22 | 2019-07-23 | 重庆理工大学 | Aspect level emotion analysis method based on multi-language level mechanism |
| CN110096647A (en) * | 2019-05-10 | 2019-08-06 | 腾讯科技（深圳）有限公司 | Optimize method, apparatus, electronic equipment and the computer storage medium of quantitative model |
| CN110096647B (en) * | 2019-05-10 | 2023-04-07 | 腾讯科技（深圳）有限公司 | Method and device for optimizing quantization model, electronic equipment and computer storage medium |
| CN110188933A (en) * | 2019-05-21 | 2019-08-30 | 湖北经济学院 | A kind of School Network public sentiment monitoring and pre-warning method and system |
| CN110209815A (en) * | 2019-05-23 | 2019-09-06 | 国家计算机网络与信息安全管理中心 | A kind of news Users' Interests Mining method of convolutional neural networks |
| CN110209824A (en) * | 2019-06-13 | 2019-09-06 | 中国科学院自动化研究所 | Text emotion analysis method based on built-up pattern, system, device |
| CN110209824B (en) * | 2019-06-13 | 2021-06-22 | 中国科学院自动化研究所 | Text emotion analysis method, system and device based on combined model |
| CN110276076A (en) * | 2019-06-25 | 2019-09-24 | 北京奇艺世纪科技有限公司 | A kind of text mood analysis method, device and equipment |
| CN110321563A (en) * | 2019-06-28 | 2019-10-11 | 浙江大学 | Text emotion analysis method based on mixing monitor model |
| CN110532452B (en) * | 2019-07-12 | 2022-04-22 | 西安交通大学 | News website universal crawler design method based on GRU neural network |
| CN110532452A (en) * | 2019-07-12 | 2019-12-03 | 西安交通大学 | A kind of general crawler design method of news website based on GRU neural network |
| CN110473571A (en) * | 2019-07-26 | 2019-11-19 | 北京影谱科技股份有限公司 | Emotion identification method and device based on short video speech |
| CN110472053A (en) * | 2019-08-05 | 2019-11-19 | 广联达科技股份有限公司 | A kind of automatic classification method and its system towards public resource bidding advertisement data |
| CN110442823A (en) * | 2019-08-06 | 2019-11-12 | 北京智游网安科技有限公司 | Website classification method, Type of website judgment method, storage medium and intelligent terminal |
| CN110543560A (en) * | 2019-08-08 | 2019-12-06 | 厦门市美亚柏科信息股份有限公司 | Long text classification and identification method, device and medium based on convolutional neural network |
| CN110472245A (en) * | 2019-08-15 | 2019-11-19 | 东北大学 | A kind of multiple labeling emotional intensity prediction technique based on stratification convolutional neural networks |
| CN110502757A (en) * | 2019-08-29 | 2019-11-26 | 西安邮电大学 | A kind of natural language sentiment analysis method |
| CN110502757B (en) * | 2019-08-29 | 2023-01-10 | 西安邮电大学 | Natural language emotion analysis method |
| CN110957039A (en) * | 2019-12-16 | 2020-04-03 | 河南科技学院 | Campus psychological coaching method and device based on deep learning |
| CN111309859A (en) * | 2020-01-21 | 2020-06-19 | 上饶市中科院云计算中心大数据研究院 | Scenic spot network public praise emotion analysis method and device |
| CN111353313A (en) * | 2020-02-25 | 2020-06-30 | 四川翼飞视科技有限公司 | Emotion analysis model construction method based on evolutionary neural network architecture search |
| CN111339251A (en) * | 2020-02-25 | 2020-06-26 | 上海昌投网络科技有限公司 | Method and device for detecting whether WeChat public number has sensitive words or not |
| CN111414475A (en) * | 2020-03-03 | 2020-07-14 | 北京明略软件系统有限公司 | Text emotion information identification method and device |
| CN111538835B (en) * | 2020-03-30 | 2023-05-23 | 东南大学 | Social media emotion classification method and device based on knowledge graph |
| CN111538835A (en) * | 2020-03-30 | 2020-08-14 | 东南大学 | Social media emotion classification method and device based on knowledge graph |
| CN111523574A (en) * | 2020-04-13 | 2020-08-11 | 云南大学 | Image emotion recognition method and system based on multi-mode data |
| CN111860981B (en) * | 2020-07-03 | 2024-01-19 | 航天信息(山东)科技有限公司 | Enterprise national industry category prediction method and system based on LSTM deep learning |
| CN111860981A (en) * | 2020-07-03 | 2020-10-30 | 航天信息(山东)科技有限公司 | Enterprise national industry category prediction method and system based on LSTM deep learning |
| CN111651607A (en) * | 2020-07-13 | 2020-09-11 | 深圳市智搜信息技术有限公司 | Information positive and negative emotion analysis method and device, computer equipment and storage medium |
| CN111984762A (en) * | 2020-08-05 | 2020-11-24 | 中国科学院重庆绿色智能技术研究院 | Text classification method sensitive to attack resistance |
| CN111984762B (en) * | 2020-08-05 | 2022-12-13 | 中国科学院重庆绿色智能技术研究院 | Text classification method sensitive to attack resistance |
| CN112183064B (en) * | 2020-10-22 | 2022-06-03 | 福州大学 | Text emotion reason recognition system based on multi-task joint learning |
| CN112183064A (en) * | 2020-10-22 | 2021-01-05 | 福州大学 | Text emotion reason recognition system based on multi-task joint learning |
| CN112597759B (en) * | 2020-11-30 | 2024-04-09 | 深延科技(北京)有限公司 | Emotion detection method and device based on text, computer equipment and medium |
| CN112597759A (en) * | 2020-11-30 | 2021-04-02 | 深延科技(北京)有限公司 | Text-based emotion detection method and device, computer equipment and medium |
| CN112612878A (en) * | 2020-12-17 | 2021-04-06 | 大唐融合通信股份有限公司 | Customer service information providing method, electronic equipment and device |
| CN113010802B (en) * | 2021-03-25 | 2022-09-20 | 华南理工大学 | Recommendation method facing implicit feedback based on multi-attribute interaction of user and article |
| CN113010802A (en) * | 2021-03-25 | 2021-06-22 | 华南理工大学 | Recommendation method facing implicit feedback based on multi-attribute interaction of user and article |
| CN113191212A (en) * | 2021-04-12 | 2021-07-30 | 合肥综合性国家科学中心人工智能研究院(安徽省人工智能实验室) | Driver road rage risk early warning method and system |
| CN113297364A (en) * | 2021-06-07 | 2021-08-24 | 吉林大学 | Natural language understanding method and device for dialog system |
| CN114022909A (en) * | 2022-01-07 | 2022-02-08 | 首都师范大学 | Emotion recognition method and system based on sensor data |
| CN115146059A (en) * | 2022-06-17 | 2022-10-04 | 东方合智数据科技(广东)有限责任公司 | Raw paper market data processing method based on corrugated paper industry and related equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109299253A (en) | A kind of social text Emotion identification model construction method of Chinese based on depth integration neural network | |
| CN110287320B (en) | Deep learning multi-classification emotion analysis model combining attention mechanism | |
| CN108764268A (en) | A kind of multi-modal emotion identification method of picture and text based on deep learning | |
| CN109543084B (en) | Method for establishing detection model of hidden sensitive text facing network social media | |
| CN106202010B (en) | Method and apparatus based on deep neural network building Law Text syntax tree | |
| CN109492157B (en) | News recommendation method and theme characterization method based on RNN and attention mechanism | |
| CN107066446B (en) | Logic rule embedded cyclic neural network text emotion analysis method | |
| CN109271493B (en) | Language text processing method and device and storage medium | |
| CN106886580B (en) | Image emotion polarity analysis method based on deep learning | |
| CN104298651B (en) | Biomedicine named entity recognition and protein interactive relationship extracting on-line method based on deep learning | |
| CN109543034B (en) | Text clustering method and device based on knowledge graph and readable storage medium | |
| CN111783394B (en) | Training method of event extraction model, event extraction method, system and equipment | |
| CN110175227A (en) | A kind of dialogue auxiliary system based on form a team study and level reasoning | |
| CN110083700A (en) | A kind of enterprise's public sentiment sensibility classification method and system based on convolutional neural networks | |
| CN111368074A (en) | Link prediction method based on network structure and text information | |
| CN107704558A (en) | A kind of consumers' opinions abstracting method and system | |
| CN111985247B (en) | Microblog user interest identification method and system based on multi-granularity text feature representation | |
| CN108614855A (en) | A kind of rumour recognition methods | |
| CN110196945A (en) | A kind of microblog users age prediction technique merged based on LSTM with LeNet | |
| Sheshikala et al. | Natural language processing and machine learning classifier used for detecting the author of the sentence | |
| CN111524593A (en) | Medical question-answering method and system based on context language model and knowledge embedding | |
| CN107463935A (en) | Application class methods and applications sorter | |
| CN109710826A (en) | A kind of internet information artificial intelligence acquisition method and its system | |
| CN107862322A (en) | The method, apparatus and system of picture attribute classification are carried out with reference to picture and text | |
| CN113312478A (en) | Viewpoint mining method and device based on reading understanding |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190201 |