CN107590134A - Text sentiment classification method, storage medium and computer - Google Patents
Text sentiment classification method, storage medium and computer Download PDFInfo
- Publication number
- CN107590134A CN107590134A CN201711012851.8A CN201711012851A CN107590134A CN 107590134 A CN107590134 A CN 107590134A CN 201711012851 A CN201711012851 A CN 201711012851A CN 107590134 A CN107590134 A CN 107590134A
- Authority
- CN
- China
- Prior art keywords
- vector
- low
- image feature
- level image
- feature vector
- 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
- G06F40/237—Lexical tools
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
A kind of text sentiment classification method, storage medium and computer, wherein method comprise the following steps, sentiment dictionary structure is carried out to input text, and the sentiment dictionary construction step includes part of speech selection expression, low-level image feature vector extraction；Middle level features extract, and with reference to the sentiment dictionary, gather the term vector of training sample, it is vectorial that the term vector progress Chi Huahou to training sample obtains middle level features；Fusion is weighted to low-level image feature vector, middle level features vector, obtains fusion feature vector, low-level image feature vector disaggregated model, middle level features vector disaggregated model, fusion feature vector disaggregated model is based respectively on and calculates classification results.Solve the problems, such as that prior art emotional semantic classification is not efficient enough, stable.
Description
Technical field
The present invention relates to machine learning field, more particularly to the method and storage medium of a kind of classification of text emotion.
Background technology
Emotional semantic classification, it is mainly used in analyzing or predicts the emotional category belonging to the text with Sentiment orientation.General point
For positive, negative sense or forward direction, negative sense and neutrality., can be roughly by feelings according to the difference of the size granularity of research object
Sense analytical technology is divided into following three level：The sentiment analysis of word-level, Sentence-level and chapter level.
Emotional semantic classification based on word-level can be divided into the sentiment classification model based on dictionary and the feelings based on corpus again
Feel disaggregated model.Sentiment classification model based on dictionary relies on the synonymous, antonymy in existing dictionary to judge in text
The Sentiment orientation of word.Have scholar by " good " and " bad " this kind of word being substantially inclined to as benchmark word, then calculate again posting term with
The difference of mutual information between benchmark word.There is scholar to detect the adjectival fuzzy emotional category in text using HowNet marks,
The core adjective that the uncertain adjective of emotional category and emotional category determine is distinguished by calculating net covering score.It is based on
The sentiment classification model of corpus to existing corpus mainly by carrying out statistical analysis, to identify the Sentiment orientation of word
Property.There is scholar to propose a kind of method based on emotion congruity theory, they think that different conjunctions contains potential language
Adopted relation, so the semantic emotion of unregistered word can be excavated using the conjunction in corpus.There is scholar to propose one kind
Solve emotion word and lead domain-dependent method, extract the emotion word and emotion object in text with existing corpus first, so
They are formed into an emotion collocation pair afterwards, the emotion of each emotion collocation pair is calculated using heuritic approach, will be last
As a result an emotion dictionary of collocations is configured to, this way solves the Context-dependent of emotion word to a certain extent.
Emotional semantic classification based on Sentence-level can be divided into two sub- directions again：Based on semantic emotional semantic classification and based on statistics
Emotional semantic classification.Need to match sentiment dictionary based on the emotional semantic classification of semanteme to find out the emotion word in sentence, then pass through emotion
The emotion intensity or polarity of word calculate the overall emotion of sentence.There is scholar to attempt solve sentence using Rhetorical Structure Theory
Sentiment orientation sex chromosome mosaicism, sentence is divided into according to the theory by different text element blocks first, and according to document entirety feelings
The significance level of sense distributes each element blocks different weights, and trying to achieve the overall emotion score of sentence finally by weighting is carried out
Emotion is predicted.The method that sentiment analysis method based on statistics is namely based on machine learning, is passed through using the data marked
Machine learning algorithm trains a model, then carries out the prediction of Sentiment orientation to unknown text data with the model.Have
Scholar attempts using the number of positive negative sense emotion word, negative word, special keyword, part of speech label and emoticon and waited
Carry out construction feature vector, Sentiment orientation classification is carried out to pushing away special data using the method for machine learning, it is big with deep learning
Heat, some scholars are combined by the use of recurrent neural network to phrase vector sum term vector and it is sent into grader as feature
Sentiment orientation analysis is carried out, experiment demonstrates the validity of such method.
The overall emotion of the chapter level text as news, blog etc. is mainly studied based on chapter level emotional semantic classification.Research
The semantic information that emphasis is placed upon text on.The method for having scholar to propose analyzes the evaluation phrase occurred in chapter level text
Phrase, by analyzing the emotion tendency of these evaluation phrase phrases, a sentiment dictionary is semi-automatically built, then utilizes feelings
Dictionary is felt to analyze the overall emotion of chapter.And sentiment analysis is carried out then more to chapter level text based on the method for machine learning
To be universal.Such method using the various resources such as emotion word, phrase, by SVMs this classical machine learning algorithm come
Build the sentiment classification model of chapter level text.In addition, also a kind of method is that chapter level text first is divided into multiple sentences,
And sentiment analysis is carried out to each sentence using maximum entropy algorithm；Then by the Sentiment orientation of sentence and its position, clause etc.
Feature combines, and the feature for forming chapter is sent into SVMs, trains the emotion classifiers of chapter level text, also achieves not
Wrong result.
The content of the invention
For this reason, it may be necessary to provide a kind of text sentiment classification method, it is not efficient enough, stabilization to solve prior art emotional semantic classification
Problem；
To achieve the above object, a kind of text sentiment classification method is inventor provided, is comprised the following steps, to input text
This progress sentiment dictionary structure, the sentiment dictionary construction step include part of speech selection expression, low-level image feature vector extraction；Middle level
Feature extraction, with reference to the sentiment dictionary, the term vector of training sample is gathered, the term vector of training sample is carried out behind pond
To middle level features vector；Fusion is weighted to low-level image feature vector, middle level features vector, obtains fusion feature vector,
It is based respectively on low-level image feature vector disaggregated model, middle level features vector disaggregated model, fusion feature vector disaggregated model and calculates and divides
Class result.
Specifically, the extraction of bottom vector is using vector space model specifically, expressed low-level image feature, wherein
It is the TF-TDF weights after normalization per one-dimensional feature.
Further, low-level image feature vector, middle level features vector Weighted Fusion are expressed as,
Wherein, L represents low-level image feature vector, and M is expressed as middle level features vector,For the weight of low-level image feature, | | expression
It is the symbol of series connection.
Preferably, include after the progress pond specific steps to term vector, by the number of dimensions decile of low-level image feature vector
For several pieces, the term vector in every a dimension is summed, then summed result is sequentially carried out to summed result
Merge.
A kind of text emotion classification storage medium, is stored with computer program, and the computer program is held by processor
Following steps are realized during row, sentiment dictionary structure are carried out to input text, the sentiment dictionary construction step selects including part of speech
Expression, low-level image feature vector extraction；Middle level features extract, and with reference to the sentiment dictionary, the term vector of training sample are gathered, to instruction
The term vector progress Chi Huahou for practicing sample obtains middle level features vector；Low-level image feature vector, middle level features vector are carried out
Weighted Fusion, obtain fusion feature vector, be based respectively on low-level image feature vector disaggregated model, middle level features vector disaggregated model,
Fusion feature vector disaggregated model calculates classification results.
Further, the extraction of bottom vector is using vector space model specifically, expressed low-level image feature, its
In per one-dimensional feature for normalization after TF-TDF weights.
Specifically, low-level image feature vector, middle level features vector Weighted Fusion are expressed as,
Wherein, L represents low-level image feature vector, and M is expressed as middle level features vector,For the weight of low-level image feature, | | expression
It is the symbol of series connection.
Preferably, it is described that term vector progress pond specific steps are also included, by the number of dimensions decile of low-level image feature vector
For several pieces, the term vector in every a dimension is summed, then summed result is sequentially carried out to summed result
Merge.
A kind of computer, the computer include above-mentioned storage medium.
Prior art is different from, the present invention can establish the relatively low emotion of a dimension efficiently, stable by learning
Dictionary, continue to use sentiment dictionary, in combination with Fusion Features and the mode of Multiple Classifier Fusion, to effectively improve nicety of grading,
Generation classification results are removed by bottom, middle level, fusion feature vector, then by three graders, enable to final classification
As a result it is more stable, with more robustness.Also reduce the amount of calculation of the inventive method by careful pond process, to sum up institute
State, the present invention solves in the prior art the problem of text emotion classification effectiveness is not high, nicety of grading deficiency.
Brief description of the drawings
Fig. 1 is the flow chart for the text sentiment classification method that an embodiment of the present invention is related to；
Fig. 2 is the text sentiment classification method whole process schematic diagram that an embodiment of the present invention is related to；
Fig. 3 is the pond procedure chart that an embodiment of the present invention is related to；
Fig. 4 is the Fusion Features figure that an embodiment of the present invention is related to；
Embodiment
To describe the technology contents of technical scheme, construction feature, the objects and the effects in detail, below in conjunction with specific reality
Apply example and coordinate accompanying drawing to be explained in detail.
Referring to Fig. 1, being a kind of text sentiment classification method, this method is the emotional semantic classification mould based on extreme learning machine
Type.Extreme learning machine is a kind of feedforward neural network of single hidden layer (Single-hidden Layer Feedforward
Neural Networks, SLFNs), the network is made up of input layer, hidden layer, output layer three parts, while input layer is to hidden
Layer, hidden layer are hidden to being all to connect entirely between output layer.The inventive method may begin at step,
S100 to input text carry out sentiment dictionary structure, the sentiment dictionary construction step include part of speech selection expression,
Low-level image feature vector extraction.In certain embodiments, as shown in Fig. 2 sentiment dictionary construction step includes part of speech selection and bottom
Two processes of feature selecting.Part of speech selection expression selects noun, verb, adjective and adverbial word collectively as benchmark in the present invention
Word, sentiment dictionary can be the benchmark set of words of the four kinds of parts of speech occurred in all selection materials.And by the word of different parts of speech
The combined potential applications information for forming a document, so can at utmost ensure that the coverage rate of sentiment dictionary, protect simultaneously
The semantic information of document is stayed.Stratum characteristic vector extraction is further selected using the low-level image feature choosing principles based on chi
The Feature Words of text feeling polarities can most be represented by selecting.Wherein low-level image feature selection vector space model is expressed, wherein vector
In often tie up be characterized in normalization after TF-IDF weight.
Step S102 middle level features extract, and with reference to the sentiment dictionary, the term vector of training sample are gathered, to training sample
Term vector carry out Chi Huahou obtain middle level features vector；Specifically, we can train Skip-gram using unsupervised mode
Model, and with the mode input training sample trained, produce training sample term vector.Specific pond step as shown in figure 3,
The number of dimensions of term vector can be divided into several pieces by us, and the term vector in every a dimension is summed, then will summation
As a result sequentially summed result is merged.Assuming that text includes x word, it is left t after low-level image feature extracts
Individual word, this text are expressed as T=(w1,w2,...wt), wherein the term vector of each word is, each term vector has k Wei Te
Sign；
(2) term vector in text T is divided into N parts, forms N number of term vector group, each group the inside is corresponding with t/N word
Vector；
(3) following operate is carried out for each term vector group：All term vectors in group are added up, final each word to
Amount group can all form a characteristic vector v (z), and the dimension of this feature vector is also k；
(4) characteristic vector of N number of term vector group is together in series and just obtains one brand-new vector of feature of whole document,
As shown by the equation：v(z1)||v(z2)||...||v(zN).Wherein | | represent the symbol of series connection.
As shown in Figure 1 and Figure 4, the present invention also progress step S104 enters to low-level image feature vector, middle level features vector
Row Weighted Fusion, obtains fusion feature vector, and S106 is based respectively on low-level image feature vector disaggregated model, middle level features vector classification
Model, fusion feature vector disaggregated model calculate classification results.In some embodiments it is possible to referring to Fig. 2, to input sample institute
The detailed process of the emotional semantic classification of category is：Respectively by low-level image feature, middle level features and the assemblage characteristic feeding pair of sample to be determined
In the sentiment classification model based on extreme learning machine trained answered, then the output result vector progress by three disaggregated models
It is added, obtains final discriminant vector, the vectorial intermediate value maximum corresponding label is exactly final emotional category.
By above-mentioned steps, the present invention can establish the relatively low emotion word of a dimension efficiently, stable by learning
Allusion quotation, continue to use sentiment dictionary, in combination with Fusion Features and the mode of Multiple Classifier Fusion, to effectively improve nicety of grading, lead to
Bottom, middle level, fusion feature vector are crossed, then generation classification results are removed by three graders, enables to final classification knot
Fruit is more stable, with more robustness.Also reduce the amount of calculation of the inventive method by careful pond process, in summary,
The present invention solves in the prior art the problem of text emotion classification effectiveness is not high, nicety of grading deficiency.
In some other further embodiment, low-level image feature vector, middle level features vector Weighted Fusion are expressed as,
Wherein, L represents low-level image feature vector, and M is expressed as middle level features vector,For the weight of low-level image feature, | | expression
It is the symbol of series connection.By the above-mentioned means, enable to the combination ratio of low-level image feature vector and middle level features vector being capable of root
Need to be adjusted exactly according to user.Adjusted by carrying out preferably fitting to combination, preferably reach raising model
The effect of nicety of grading.
In embodiment preferably, step can also be first carried out before step S100, text is pre-processed, removed
With the incoherent information of this task, such as canonical code form, removal forbidden character, participle and part-of-speech tagging processing and deactivation
Word processing.Canonical code form is used for unified text code and operated, such as content of text is unified for UTF-8 coded format；
Filtration treatment can be carried out by the way of matching regular expressions to forbidden character by removing forbidden character；Segment at part-of-speech tagging
Reason is segmented using ICTCLAS Chinese lexical analysis systems and part-of-speech tagging；Stop words processing is using deactivation vocabulary to text
In often occur but itself the word that sentiment analysis has little significance is filtered.Pass through preprocessing process, it is possible to increase text
The specific aim and adaptability for grader are segmented, greatly accelerates recognition efficiency of the inventive method to text.
A kind of text emotion classification storage medium, is stored with computer program, and the computer program is held by processor
Following steps are realized during row, sentiment dictionary structure are carried out to input text, the sentiment dictionary construction step selects including part of speech
Expression, low-level image feature vector extraction；Middle level features extract, and with reference to the sentiment dictionary, the term vector of training sample are gathered, to instruction
The term vector progress Chi Huahou for practicing sample obtains middle level features vector；Low-level image feature vector, middle level features vector are carried out
Weighted Fusion, obtain fusion feature vector, be based respectively on low-level image feature vector disaggregated model, middle level features vector disaggregated model,
Fusion feature vector disaggregated model calculates classification results.
Further, the extraction of bottom vector is using vector space model specifically, expressed low-level image feature, its
In per one-dimensional feature for normalization after TF-TDF weights.
Specifically, low-level image feature vector, middle level features vector Weighted Fusion are expressed as,
Wherein, L represents low-level image feature vector, and M is expressed as middle level features vector,For the weight of low-level image feature, | | expression
It is the symbol of series connection.
Preferably, it is described that term vector progress pond specific steps are also included, by the number of dimensions decile of low-level image feature vector
For several pieces, the term vector in every a dimension is summed, then summed result is sequentially carried out to summed result
Merge.
A kind of computer, the computer include above-mentioned storage medium.By designing above-mentioned storage medium and computer,
The present invention solves in the prior art the problem of text emotion classification effectiveness is not high, nicety of grading deficiency.
It should be noted that although the various embodiments described above have been described herein, but not thereby limit
The scope of patent protection of the present invention.Therefore, based on the present invention innovative idea, to embodiment described herein carry out change and repair
Change, or the equivalent structure or equivalent flow conversion made using description of the invention and accompanying drawing content, directly or indirectly will be with
Upper technical scheme is used in other related technical areas, is included within the scope of patent protection of the present invention.
Claims (9)
1. a kind of text sentiment classification method, it is characterised in that comprise the following steps, sentiment dictionary structure is carried out to input text
Build, the sentiment dictionary construction step includes part of speech selection expression, low-level image feature vector extraction；Middle level features extract, with reference to institute
Sentiment dictionary is stated, gathers the term vector of training sample, the term vector progress Chi Huahou to training sample obtains middle level features vector；
Fusion is weighted to low-level image feature vector, middle level features vector, fusion feature vector is obtained, is based respectively on low-level image feature
Vectorial disaggregated model, middle level features vector disaggregated model, fusion feature vector disaggregated model calculate classification results.
2. text sentiment classification method according to claim 1, it is characterised in that the extraction of bottom vector specifically,
Low-level image feature is expressed using vector space model, the feature of each of which dimension is the TF-TDF weights after normalization.
3. text sentiment classification method according to claim 1, it is characterised in that low-level image feature vector, middle level features to
Amount Weighted Fusion is expressed as,
<mrow>
<mo>&part;</mo>
<mi>L</mi>
<mo>|</mo>
<mo>|</mo>
<mrow>
<mo>(</mo>
<mn>1</mn>
<mo>-</mo>
<mo>&part;</mo>
<mo>)</mo>
</mrow>
<mi>M</mi>
</mrow>
Wherein, L represents low-level image feature vector, and M is expressed as middle level features vector,For the weight of low-level image feature, | | expression is string
The symbol of connection.
4. text sentiment classification method according to claim 1, it is characterised in that described that pondization is carried out to term vector specifically
Include after step, the number of dimensions of low-level image feature vector be divided into several pieces, the term vector in every a dimension is summed,
Summed result is sequentially merged to summed result again.
5. a kind of text emotion classification storage medium, is stored with computer program, it is characterised in that the computer program is in quilt
Following steps are realized during computing device, sentiment dictionary structure are carried out to input text, the sentiment dictionary construction step includes
Part of speech selection expression, low-level image feature vector extraction；Middle level features extract, and with reference to the sentiment dictionary, gather the word of training sample
Vector, the term vector progress Chi Huahou to training sample obtain middle level features vector；To low-level image feature vector, middle level features
Vector is weighted fusion, obtains fusion feature vector, is based respectively on low-level image feature vector disaggregated model, middle level features vector point
Class model, fusion feature vector disaggregated model calculate classification results.
6. text emotion classification storage medium according to claim 5, it is characterised in that the bottom vector extraction is specific
To be expressed using vector space model low-level image feature, the feature of each of which dimension is the TF-TDF weights after normalization.
7. text emotion classification storage medium according to claim 5, it is characterised in that low-level image feature vector, middle level are special
Vectorial Weighted Fusion is levied to be expressed as,
<mrow>
<mo>&part;</mo>
<mi>L</mi>
<mo>|</mo>
<mo>|</mo>
<mrow>
<mo>(</mo>
<mn>1</mn>
<mo>-</mo>
<mo>&part;</mo>
<mo>)</mo>
</mrow>
<mi>M</mi>
</mrow>
Wherein, L represents low-level image feature vector, and M is expressed as middle level features vector,For the weight of low-level image feature, | | expression is string
The symbol of connection.
8. text emotion classification storage medium according to claim 5, it is characterised in that described that pond is carried out to term vector
Specific steps also include, and the number of dimensions of low-level image feature vector is divided into several pieces, the term vector in every a dimension is carried out
Summation, then summed result is sequentially merged to summed result.
9. a kind of computer, it is characterised in that the computer includes the storage medium described in claim any one of 5-8.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711012851.8A CN107590134A (en) | 2017-10-26 | 2017-10-26 | Text sentiment classification method, storage medium and computer |
| PCT/CN2018/111607 WO2019080863A1 (en) | 2017-10-26 | 2018-10-24 | Text sentiment classification method, storage medium and computer |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711012851.8A CN107590134A (en) | 2017-10-26 | 2017-10-26 | Text sentiment classification method, storage medium and computer |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107590134A true CN107590134A (en) | 2018-01-16 |
Family
ID=61043372
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711012851.8A Pending CN107590134A (en) | 2017-10-26 | 2017-10-26 | Text sentiment classification method, storage medium and computer |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN107590134A (en) |
| WO (1) | WO2019080863A1 (en) |
Cited By (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108363753A (en) * | 2018-01-30 | 2018-08-03 | 南京邮电大学 | Comment text sentiment classification model is trained and sensibility classification method, device and equipment |
| CN108804417A (en) * | 2018-05-21 | 2018-11-13 | 山东科技大学 | A kind of documentation level sentiment analysis method based on specific area emotion word |
| CN109033089A (en) * | 2018-09-06 | 2018-12-18 | 北京京东尚科信息技术有限公司 | Sentiment analysis method and apparatus |
| CN109582963A (en) * | 2018-11-29 | 2019-04-05 | 福建南威软件有限公司 | A kind of archives automatic classification method based on extreme learning machine |
| WO2019080863A1 (en) * | 2017-10-26 | 2019-05-02 | 福建亿榕信息技术有限公司 | Text sentiment classification method, storage medium and computer |
| CN109783800A (en) * | 2018-12-13 | 2019-05-21 | 北京百度网讯科技有限公司 | Acquisition methods, device, equipment and the storage medium of emotion keyword |
| CN110362734A (en) * | 2019-06-24 | 2019-10-22 | 北京百度网讯科技有限公司 | Text recognition method, device, equipment and computer readable storage medium |
| CN110795564A (en) * | 2019-11-01 | 2020-02-14 | 南京稷图数据科技有限公司 | Text classification method lacking negative cases |
| CN110909167A (en) * | 2019-11-29 | 2020-03-24 | 重庆邮电大学 | Microblog text classification system |
| CN111159410A (en) * | 2019-12-31 | 2020-05-15 | 广州广电运通信息科技有限公司 | Text emotion classification method, system and device and storage medium |
| CN111324734A (en) * | 2020-02-17 | 2020-06-23 | 昆明理工大学 | Case microblog comment emotion classification method integrating emotion knowledge |
| CN111930938A (en) * | 2020-07-06 | 2020-11-13 | 武汉卓尔数字传媒科技有限公司 | Text classification method and device, electronic equipment and storage medium |
| CN113222772A (en) * | 2021-04-08 | 2021-08-06 | 合肥工业大学 | Native personality dictionary construction method, system, storage medium and electronic device |
| CN113344121A (en) * | 2021-06-29 | 2021-09-03 | 北京百度网讯科技有限公司 | Method for training signboard classification model and signboard classification |
| CN115545573A (en) * | 2022-11-30 | 2022-12-30 | 北京清众神州大数据有限公司 | Risk early warning method, device, equipment and storage medium based on social event |
Families Citing this family (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110968697B (en) * | 2019-12-20 | 2023-06-16 | 合肥讯飞数码科技有限公司 | Text classification method, apparatus, device and readable storage medium |
| CN111339305B (en) * | 2020-03-20 | 2023-04-14 | 北京中科模识科技有限公司 | Text classification method and device, electronic equipment and storage medium |
| CN111708886A (en) * | 2020-06-11 | 2020-09-25 | 国网天津市电力公司 | Public opinion analysis terminal and public opinion text analysis method based on data driving |
| CN112163607A (en) * | 2020-09-17 | 2021-01-01 | 平顶山学院 | Network social media emotion classification method based on multi-dimension and multi-level combined modeling |
| CN113239685B (en) * | 2021-01-13 | 2023-10-31 | 中国科学院计算技术研究所 | Public opinion detection method and system based on double emotions |
| CN112836051B (en) * | 2021-02-19 | 2024-03-26 | 太极计算机股份有限公司 | Online self-learning court electronic file text classification method |
| CN113011192B (en) * | 2021-03-16 | 2023-09-15 | 广东工业大学 | Text emotion feature extraction method based on attention causal interpretation |
| CN113312481A (en) * | 2021-05-27 | 2021-08-27 | 中国平安人寿保险股份有限公司 | Text classification method, device and equipment based on block chain and storage medium |
| CN117521639B (en) * | 2024-01-05 | 2024-04-02 | 湖南工商大学 | Text detection method combined with academic text structure |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103020712A (en) * | 2012-12-28 | 2013-04-03 | 东北大学 | Distributed classification device and distributed classification method for massive micro-blog data |
| CN103729431A (en) * | 2013-12-26 | 2014-04-16 | 东北大学 | Massive microblog data distributed classification device and method with increment and decrement function |
| CN104820997A (en) * | 2015-05-14 | 2015-08-05 | 北京理工大学 | Target tracking method based on block sparse expression and HSV feature fusion |
| CN105824922A (en) * | 2016-03-16 | 2016-08-03 | 重庆邮电大学 | Emotion classifying method fusing intrinsic feature and shallow feature |
| CN106503049A (en) * | 2016-09-22 | 2017-03-15 | 南京理工大学 | A kind of microblog emotional sorting technique for merging multiple affection resources based on SVM |
| CN107092596A (en) * | 2017-04-24 | 2017-08-25 | 重庆邮电大学 | Text emotion analysis method based on attention CNNs and CCR |
| CN107247702A (en) * | 2017-05-05 | 2017-10-13 | 桂林电子科技大学 | A kind of text emotion analysis and processing method and system |
| CN107291696A (en) * | 2017-06-28 | 2017-10-24 | 达而观信息科技（上海）有限公司 | A kind of comment word sentiment analysis method and system based on deep learning |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101377769A (en) * | 2007-08-29 | 2009-03-04 | 中国科学院自动化研究所 | Method for representing multiple graininess of text message |
| CN102789498B (en) * | 2012-07-16 | 2014-08-06 | 钱钢 | Method and system for carrying out sentiment classification on Chinese comment text on basis of ensemble learning |
| CN106326212B (en) * | 2016-08-26 | 2019-04-16 | 北京理工大学 | A kind of implicit chapter relationship analysis method based on level deep semantic |
| CN107590134A (en) * | 2017-10-26 | 2018-01-16 | 福建亿榕信息技术有限公司 | Text sentiment classification method, storage medium and computer |
-
2017
- 2017-10-26 CN CN201711012851.8A patent/CN107590134A/en active Pending
-
2018
- 2018-10-24 WO PCT/CN2018/111607 patent/WO2019080863A1/en active Application Filing
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103020712A (en) * | 2012-12-28 | 2013-04-03 | 东北大学 | Distributed classification device and distributed classification method for massive micro-blog data |
| CN103729431A (en) * | 2013-12-26 | 2014-04-16 | 东北大学 | Massive microblog data distributed classification device and method with increment and decrement function |
| CN104820997A (en) * | 2015-05-14 | 2015-08-05 | 北京理工大学 | Target tracking method based on block sparse expression and HSV feature fusion |
| CN105824922A (en) * | 2016-03-16 | 2016-08-03 | 重庆邮电大学 | Emotion classifying method fusing intrinsic feature and shallow feature |
| CN106503049A (en) * | 2016-09-22 | 2017-03-15 | 南京理工大学 | A kind of microblog emotional sorting technique for merging multiple affection resources based on SVM |
| CN107092596A (en) * | 2017-04-24 | 2017-08-25 | 重庆邮电大学 | Text emotion analysis method based on attention CNNs and CCR |
| CN107247702A (en) * | 2017-05-05 | 2017-10-13 | 桂林电子科技大学 | A kind of text emotion analysis and processing method and system |
| CN107291696A (en) * | 2017-06-28 | 2017-10-24 | 达而观信息科技（上海）有限公司 | A kind of comment word sentiment analysis method and system based on deep learning |
Cited By (23)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2019080863A1 (en) * | 2017-10-26 | 2019-05-02 | 福建亿榕信息技术有限公司 | Text sentiment classification method, storage medium and computer |
| CN108363753B (en) * | 2018-01-30 | 2020-05-19 | 南京邮电大学 | Comment text emotion classification model training and emotion classification method, device and equipment |
| CN108363753A (en) * | 2018-01-30 | 2018-08-03 | 南京邮电大学 | Comment text sentiment classification model is trained and sensibility classification method, device and equipment |
| CN108804417A (en) * | 2018-05-21 | 2018-11-13 | 山东科技大学 | A kind of documentation level sentiment analysis method based on specific area emotion word |
| CN108804417B (en) * | 2018-05-21 | 2022-03-15 | 山东科技大学 | Document-level emotion analysis method based on specific field emotion words |
| CN109033089A (en) * | 2018-09-06 | 2018-12-18 | 北京京东尚科信息技术有限公司 | Sentiment analysis method and apparatus |
| CN109033089B (en) * | 2018-09-06 | 2021-01-26 | 北京京东尚科信息技术有限公司 | Emotion analysis method and device |
| CN109582963A (en) * | 2018-11-29 | 2019-04-05 | 福建南威软件有限公司 | A kind of archives automatic classification method based on extreme learning machine |
| CN109783800A (en) * | 2018-12-13 | 2019-05-21 | 北京百度网讯科技有限公司 | Acquisition methods, device, equipment and the storage medium of emotion keyword |
| CN109783800B (en) * | 2018-12-13 | 2024-04-12 | 北京百度网讯科技有限公司 | Emotion keyword acquisition method, device, equipment and storage medium |
| CN110362734A (en) * | 2019-06-24 | 2019-10-22 | 北京百度网讯科技有限公司 | Text recognition method, device, equipment and computer readable storage medium |
| CN110795564A (en) * | 2019-11-01 | 2020-02-14 | 南京稷图数据科技有限公司 | Text classification method lacking negative cases |
| CN110795564B (en) * | 2019-11-01 | 2022-02-22 | 南京稷图数据科技有限公司 | Text classification method lacking negative cases |
| CN110909167A (en) * | 2019-11-29 | 2020-03-24 | 重庆邮电大学 | Microblog text classification system |
| CN110909167B (en) * | 2019-11-29 | 2022-07-01 | 重庆邮电大学 | Microblog text classification system |
| CN111159410A (en) * | 2019-12-31 | 2020-05-15 | 广州广电运通信息科技有限公司 | Text emotion classification method, system and device and storage medium |
| CN111324734A (en) * | 2020-02-17 | 2020-06-23 | 昆明理工大学 | Case microblog comment emotion classification method integrating emotion knowledge |
| CN111930938A (en) * | 2020-07-06 | 2020-11-13 | 武汉卓尔数字传媒科技有限公司 | Text classification method and device, electronic equipment and storage medium |
| CN113222772B (en) * | 2021-04-08 | 2023-10-31 | 合肥工业大学 | Native personality dictionary construction method, native personality dictionary construction system, storage medium and electronic equipment |
| CN113222772A (en) * | 2021-04-08 | 2021-08-06 | 合肥工业大学 | Native personality dictionary construction method, system, storage medium and electronic device |
| CN113344121A (en) * | 2021-06-29 | 2021-09-03 | 北京百度网讯科技有限公司 | Method for training signboard classification model and signboard classification |
| CN113344121B (en) * | 2021-06-29 | 2023-10-27 | 北京百度网讯科技有限公司 | Method for training a sign classification model and sign classification |
| CN115545573A (en) * | 2022-11-30 | 2022-12-30 | 北京清众神州大数据有限公司 | Risk early warning method, device, equipment and storage medium based on social event |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2019080863A1 (en) | 2019-05-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107590134A (en) | Text sentiment classification method, storage medium and computer | |
| Lu et al. | Neural baby talk | |
| Mathews et al. | Semstyle: Learning to generate stylised image captions using unaligned text | |
| CN106844346A (en) | Short text Semantic Similarity method of discrimination and system based on deep learning model Word2Vec | |
| CN107679580A (en) | A kind of isomery shift image feeling polarities analysis method based on the potential association of multi-modal depth | |
| CN107025284A (en) | The recognition methods of network comment text emotion tendency and convolutional neural networks model | |
| CN107133214A (en) | A kind of product demand preference profiles based on comment information are excavated and its method for evaluating quality | |
| CN105279495A (en) | Video description method based on deep learning and text summarization | |
| CN104965822A (en) | Emotion analysis method for Chinese texts based on computer information processing technology | |
| CN110796160A (en) | Text classification method, device and storage medium | |
| Lin et al. | Deep structured scene parsing by learning with image descriptions | |
| CN106997341A (en) | A kind of innovation scheme matching process, device, server and system | |
| CN110750648A (en) | Text emotion classification method based on deep learning and feature fusion | |
| CN108154156B (en) | Image set classification method and device based on neural topic model | |
| CN108647191A (en) | It is a kind of based on have supervision emotion text and term vector sentiment dictionary construction method | |
| CN106446147A (en) | Emotion analysis method based on structuring features | |
| Huber et al. | Predicting discourse structure using distant supervision from sentiment | |
| Liu et al. | Idiom-aware compositional distributed semantics | |
| Huang et al. | C-Rnn: a fine-grained language model for image captioning | |
| CN111144410A (en) | Cross-modal image semantic extraction method, system, device and medium | |
| CN110472245A (en) | A kind of multiple labeling emotional intensity prediction technique based on stratification convolutional neural networks | |
| CN109033073A (en) | Text contains recognition methods and device | |
| Errami et al. | Sentiment Analysis onMoroccan Dialect based on ML and Social Media Content Detection | |
| CN109670169B (en) | Deep learning emotion classification method based on feature extraction | |
| Mahdin et al. | Sentiment analysis on covid-19 vaccine tweets using machine learning and deep learning algorithms |
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: 20180116 |