CN109325238A - A kind of method of multiple entity sentiment analysis in long text - Google Patents
A kind of method of multiple entity sentiment analysis in long text Download PDFInfo
- Publication number
- CN109325238A CN109325238A CN201811283252.4A CN201811283252A CN109325238A CN 109325238 A CN109325238 A CN 109325238A CN 201811283252 A CN201811283252 A CN 201811283252A CN 109325238 A CN109325238 A CN 109325238A
- Authority
- CN
- China
- Prior art keywords
- entity
- text
- sentence
- neural network
- emotion
- 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
- 238000004458 analytical method Methods 0.000 title claims abstract description 53
- 238000000034 method Methods 0.000 title claims abstract description 28
- 230000008451 emotion Effects 0.000 claims abstract description 73
- 238000013528 artificial neural network Methods 0.000 claims description 52
- 238000006116 polymerization reaction Methods 0.000 claims description 17
- 210000005036 nerve Anatomy 0.000 claims description 4
- 239000000654 additive Substances 0.000 claims description 3
- 230000000996 additive effect Effects 0.000 claims description 3
- 230000002996 emotional effect Effects 0.000 claims description 3
- 238000001914 filtration Methods 0.000 claims description 3
- 210000002569 neuron Anatomy 0.000 claims description 3
- 210000004218 nerve net Anatomy 0.000 claims 1
- 238000010586 diagram Methods 0.000 description 3
- 208000027534 Emotional disease Diseases 0.000 description 2
- 230000012010 growth Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 239000002360 explosive Substances 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000003058 natural language processing Methods 0.000 description 1
- 230000007935 neutral effect Effects 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 230000007306 turnover Effects 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/30—Semantic analysis
-
- 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
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)
- Document Processing Apparatus (AREA)
Abstract
The present invention provides a kind of method of multiple entity sentiment analysis in long text, is related to information technology technical field.A kind of method of multiple entity sentiment analysis in long text, the entity in name entity algorithm identification text is used to the long article for needing to carry out multiple entity sentiment analysis first, and text is split, it is analyzed by syntactic structure, obtains text information relevant to entity in each text sentence;Then according to text information relevant to entity in each text sentence, the emotion of each entity in each text sentence is obtained;Finally according to the emotion of entity each in each text sentence, emotion of each entity in entire full text sheet is obtained.The method of multiple entity sentiment analysis in long text provided by the invention can take into account the identification of entity in text, the correlation between the sentiment analysis of entity while analyzing multiple entities, considering entity;And effectively the entity emotion being dispersed in the multiple sentences of article is polymerize and has the entity result of inclusiveness to judge.
Description
Technical field
The present invention relates to a kind of methods of multiple entity sentiment analysis in information technology field more particularly to long text.
Background technique
With the fast development of internet, the information on network shows explosive growth.Text information is the network information
One important channel of exchange, content all rapid growths of the forms such as report, comment, microblogging, blog are corresponding, are
It is difficult to excavate the abundant content in text to the slow development of content detection and digging technology.
Sentiment analysis is the one long-term burning hot topic in current natural language processing.Currently, the development of sentiment analysis
It is very fast.There is the analysis method based on dictionary and WordNet in analysis method, based on machine learning and based on deep learning
Analysis method;There is the sentiment analysis of chapter rank in analysis content, there are also the sentiment analysis of sentence level and phrase rank;Analysis
As a result there are front, neutral, negative polarity check and the emotion degree analyzing by scoring on.Most of sentiment analysis methods are all
It is to be built upon under single name entity or only comprising being carried out under the premise of single emotion in the text of required analysis, it can
It is that most of content of text are not write according to this premise, passage includes multiple entities or a variety of emotions are all
A possibility that being very likely to, especially occurring in long text is very high, and the existing sentiment analysis method of such case is also not
It is able to carry out and handles well.
Name entity refers to name, mechanism name, place name etc. with entitled mark ground entity, this usually has specific in the text
Reference meaning, also occur in a large amount of text, it is especially in the majority with long report.And in long report, it will usually have multiple names
Entity occurs, and the single emotion for naming entity can also change, and front is presented in introductory song, end presents negative;Or it is opening
A piece and end are presented negatively, but many in the intermediate situation that front is presented.It is right there are these complexity
Long text carries out quickly and effectively sentiment analysis and is exactly one and is difficult to solve the problems, such as.
At present in the method for the more emotional problems of multiple entity, has by the way of dividing text, text is cut,
Retain the name entity for needing to carry out sentiment analysis, only to be compared comprehensive Judgment by emotion.This method can destroy text
Relevance between content, although certain words in do not mention some name entity, the entity referred to want to be judged
Name entity contacted, the name entity for wanting to be judged can also be had an impact, therefore this mode is not ideal enough.
And other some most of methods for the more emotional problems of multiple entity focus on short text, the information statement of short text does not have
Long text is so complicated, and discrimination degree is more preferable, and the case where occurring the susceptible sense of multiple entity in short text is less, does not have pervasive
Property.Defect on both modes also effective percentage, the target article of analysis only have one, but to carry out repeatedly traversing analysis, when
Between waste it is serious.
Summary of the invention
The technical problem to be solved by the present invention is in view of the above shortcomings of the prior art, provide multiple entity in a kind of long text
The method of sentiment analysis is realized and all carries out global sentiment analysis to the name entity in a long text.
In order to solve the above technical problems, the technical solution used in the present invention is: multiple entity emotion is divided in a kind of long text
The method of analysis, comprising the following steps:
Step 1 uses the reality in name entity algorithm identification text to the long article for needing to carry out multiple entity sentiment analysis
Body;
Step 2 is split text according to punctuation mark, obtains complete text sentence;
Step 3 carries out syntactic structure analysis to each text sentence, obtains text relevant to entity in each text sentence
This information;
Step 4, according to text information relevant to entity in each text sentence, obtain each reality in each text sentence
The emotion of body;
Step 5, according to the emotion of entity each in each text sentence, obtain feelings of each entity in entire full text in this
Sense, method particularly includes:
Step 5.1 carries out vectorization to the related emotion information of each entity, converts term vector for emotion information;
Step 5.2, by external emotional term database to the related emotion information additive phrase polarity of each entity, obtain
To the feeling polarities vector of each term vector, term vector and feeling polarities vector are then combined into an entirety, then every
All related emotion informations of each entity are all integrated into an entirety in a sentence, and emotion belonging to each entity is believed
Condensate is ceased as an input vector;
Step 5.3, building LSTM neural network structure;The LSTM neural network structure includes two layers, and first layer is real
Body layer neural network, the second layer are sentence layer neural network;The physical layer neural network receives the input of entity emotion information,
Timing memory step-length is sentence quantity * physical quantities, and can receive emotion information of the same entity under different sentences;It is described
The output of sentence layer neural network receiving entity layer neural network, every physical quantities timing memory step in physical layer neural network
An input of the result as sentence layer neural network after length, sentence layer neural network can finally export result;
The input vector for each entity that step 5.2 is constituted is input to the LSTM nerve that step 5.3 constructs by step 5.4
In the neuron of network structure, the polymerization of emotion information is carried out using LSTM neural network structure;
It is described to carry out emotion information polymerization using LSTM neural network structure method particularly includes:
Emotion information condensate belonging to each entity is defeated as the related emotion information of some entity under some sentence
Enter into LSTM neural network structure, the timing memory step-length of each LSTM is made to remember the emotion of certain entity in some sentence
Information;The hidden state of the last one entity can be input to the LSTM unit in sentence layer neural network, without being enter into down
First entity LSTM unit in one sentence;By LSTM unit in sentence layer neural network again to occurring in this sentence
Entity and relevant information are polymerize, and are adjusted conducive to polymerization result;And the LSTM unit in sentence layer neural network is
Connect with the LSTM unit in sentence layer neural network, thus carry out be all sentence surface entity information filtering, often
The information of a entity in the text is effectively polymerize;The related emotion information of each entity after finally output has traversed full text originally
Polymerization result;
Step 5.5, each reality exported according to the parallel judgment step 5.4 of related emotion information polymerization result of each entity
The feeling polarities of body.
The beneficial effects of adopting the technical scheme are that multiple entity feelings in a kind of long text provided by the invention
The method for feeling analysis can take into account the identification of entity in text, to the sentiment analysis of entity while analyze multiple entities, consideration
Correlation between entity;According to the difference of entity, different emotion informations can be extracted to divide for the emotion of corresponding entity
Analysis, and can the text emotion information to the different location of same entity be associated, polymerize, thus realize to a certain entity
Emotion carries out comprehensive descision.The final sentiment analysis of different entities independently carries out, and maintains and generates the effective of sentiment analysis result
Property, it reduces when generating analysis result, other entities influence the sentiment analysis of this entity.Organically combine different entities point
Analysis and the related emotion information extracting method for entity, with to long text content carry out efficiently, comprehensive extract.And it is effectively right
The entity emotion being dispersed in the multiple sentences of article is polymerize and has the entity result of inclusiveness to judge.
Detailed description of the invention
Fig. 1 is the architecture diagram of the method for multiple entity sentiment analysis in a kind of long text provided in an embodiment of the present invention;
Fig. 2 is the flow chart of the method for multiple entity sentiment analysis in a kind of long text provided in an embodiment of the present invention;
Fig. 3 is the signal that Entity recognition and interdependent syntactic analysis are named in long text provided in an embodiment of the present invention
Figure;
Fig. 4 is the simple sentence emotion information input provided in an embodiment of the present invention that multiple entity sentiment analysis is carried out in long text
Schematic diagram;
Fig. 5 carries out emotion information to entity each in text using LSTM neural network structure to be provided in an embodiment of the present invention
The structural schematic diagram of polymerization;
Fig. 6 is the method figure provided in an embodiment of the present invention judged entity emotion.
Specific embodiment
With reference to the accompanying drawings and examples, specific embodiments of the present invention will be described in further detail.Implement below
Example is not intended to limit the scope of the invention for illustrating the present invention.
The present embodiment with text " Zhang San has applied for a loan to Wang Dai company, but he is overdue, and with Wang Dai company
Negotiate to delay to refund.Wang Dai company still has collected the information of Zhang San correlation kith and kin, and affects the related parent of Zhang San using means
The normal life of friend.Zhang San is a common employee of company, often applies for loan to Wang Dai company to be had enough to meet the need.Because
Three refund in time, therefore credit line increases 8000 yuan all the way.Once after writing debt-credit, Zhang San suggests in Wang Dai company
Payment beforehand, Zhang San agree to.But Zhang San's capital turnover difficulty, it can not pay the bill.Zhang San and net borrow corporate negotiations and delay
It refunds, Wang Dai company is also agreed to, but Wang Dai company still urges third company to dun.The thing of Zhang San allows Zhang San's
Kith and kin feel very big pressure, and third company often phones them and harassed to urge Zhang San to refund.Zhang San returns
Recall and once provide the service password of oneself address list, therefore Wang Dai company has obtained the message registration of Zhang San." for, it uses
The method of multiple entity sentiment analysis in long text of the invention carries out sentiment analysis to the text.
A kind of method of multiple entity sentiment analysis in long text, as depicted in figs. 1 and 2, comprising the following steps:
Step 1 uses the reality in name entity algorithm identification text to the long article for needing to carry out multiple entity sentiment analysis
Body;
The present embodiment identifies the entity in the long text by BiLSTM-CRF model, mainly includes inside the long text
" Zhang San ", " Wang Dai company ", " kith and kin of Zhang San ", " third company " four entities, clear in order to enumerate, the present embodiment is only right
" Zhang San " and " Wang Dai company " two entity objects carry out sentiment analysis.
Step 2 is split text according to punctuation mark, obtains complete text sentence;
Step 3 carries out syntactic structure analysis to each text sentence, obtains text relevant to entity in each text sentence
This information, as shown in Figure 3;
In the present embodiment, the related sentiment analysis of " Zhang San " to " Wang Dai company " the two entity objects under each sentence
Are as follows:
Sentence 1:
Zhang San: loan, it is overdue, delay to refund
Wang Dai company: delay to refund
Sentence 2:
Zhang San: (without correlation emotion information)
Wang Dai company: information is collected, normal life is influenced
Sentence 3:
Zhang San: common employee of company, application loan
Wang Dai company: (without correlation emotion information)
Sentence 4:
Zhang San: in time, credit line increases
Wang Dai company: (without correlation emotion information)
Sentence 5:
Zhang San: agree to
Wang Dai company: it is recommended that payment beforehand
Sentence 6:
Zhang San: it is difficult, it can not pay the bill
Wang Dai company: (without correlation emotion information)
Sentence 7:
Zhang San: delay to refund
Wang Dai company: delay to refund, agree to, urge, dun
Sentence 8:
Zhang San: (without correlation emotion information)
Wang Dai company: (without correlation emotion information)
Sentence 9:
Zhang San: recall, password is provided
Wang Dai company: message registration is obtained
Step 4, according to text information relevant to entity in each text sentence, obtain each reality in each text sentence
The emotion of body;
Step 5, according to the emotion of entity each in each text sentence, obtain feelings of each entity in entire full text in this
Sense, method particularly includes:
Step 5.1 carries out vectorization to the related emotion information of each entity, converts term vector for emotion information, such as
Shown in Fig. 4;
Step 5.2, by external emotional term database to the related emotion information additive phrase polarity of each entity, obtain
To the feeling polarities vector of each term vector, term vector and feeling polarities vector are then combined into an entirety, then every
All related emotion informations of each entity are all integrated into an entirety in a sentence, and emotion belonging to each entity is believed
Cease condensate as an input vector, as shown in Figure 4;
In the present embodiment, in sentence 1 the related emotion information " overdue " of entity " Zhang San " can be converted to term vector [1,0,
0,0], " overdue " word for negative feeling polarities, vector form are [- 1], then become [1,0,0,0] and [- 1] splicing
[1,0,0,0, -1] then splices to other related emotion informations of the entity " Zhang San " again, as with " delaying to refund "
Term vector and feeling polarities vector splicing result [0,1,0,0, -1] spliced, obtain new variable [1,0,0,0, -1,0,
1,0,0, -1] as emotion information condensate belonging to the entity " Zhang San ";
Step 5.3, building LSTM neural network structure;The LSTM neural network structure includes two layers, and first layer is real
Body layer neural network, the second layer are sentence layer neural network;The physical layer neural network receives the input of entity emotion information,
Timing memory step-length is sentence quantity * physical quantities, and can receive emotion information of the same entity under different sentences;It is described
The output of sentence layer neural network receiving entity layer neural network, every physical quantities timing memory step in physical layer neural network
Result after length can finally export result as an input of sentence layer neural network, sentence layer neural network;
The input vector for each entity that step 5.2 is constituted is input to the LSTM nerve that step 5.3 constructs by step 5.4
In the neuron of network structure, the polymerization of emotion information is carried out to entity each in text using LSTM neural network structure;
It is described that emotion information polymerization is carried out to entity each in text using LSTM neural network structure, as shown in figure 5, specifically
Method are as follows:
Emotion information condensate belonging to each entity is defeated as the related emotion information of some entity under some sentence
Enter into LSTM neural network structure, the timing memory step-length of each LSTM is made to remember the emotion of certain entity in some sentence
Information;The hidden state of the last one entity can be input to the LSTM unit in sentence layer neural network, without being enter into down
First entity LSTM unit in one sentence;By LSTM unit in sentence layer neural network again to occurring in this sentence
Entity and relevant information are polymerize, and are adjusted conducive to polymerization result;And the LSTM unit in sentence layer neural network is
Connect with the LSTM unit in sentence layer neural network, thus carry out be all sentence surface entity information filtering, often
The information of a entity in the text is effectively polymerize;The related emotion information of each entity after finally output has traversed full text originally
Polymerization result;
Step 5.5, each reality exported according to the parallel judgment step 5.4 of related emotion information polymerization result of each entity
The feeling polarities of body maintain result judgement as shown in fig. 6, each entity individually judges in entire deterministic process
Independence.
In the present embodiment, by LSTM neural network structure to the emotion letter of the two entities of " Zhang San " and " Wang Dai company "
Breath condensate is analyzed, and the result finally exported is [- 0.347, -0.765];In the present embodiment, without as common classification
Neural network equally the operation such as is normalized using functions such as softmax, but each output valve is made to represent one
The full text emotion polymerization result of entity;
In the present embodiment, judged by judging the polymerization result to the related emotion information of each entity parallel, it can
To judge the feeling polarities for obtaining " Zhang San " as neutrality, the feeling polarities of " Wang Dai company " are negative.
Finally, it should be noted that the above embodiments are merely illustrative of the technical solutions of the present invention, rather than its limitations;Although
Present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that: it still may be used
To modify to technical solution documented by previous embodiment, or some or all of the technical features are equal
Replacement;And these are modified or replaceed, model defined by the claims in the present invention that it does not separate the essence of the corresponding technical solution
It encloses.
Claims (3)
1. a kind of method of multiple entity sentiment analysis in long text, it is characterised in that: the following steps are included:
Step 1 uses the entity in name entity algorithm identification text to the long article for needing to carry out multiple entity sentiment analysis;
Step 2 is split text according to punctuation mark, obtains complete text sentence;
Step 3 carries out syntactic structure analysis to each text sentence, obtains text envelope relevant to entity in each text sentence
Breath;
Step 4, according to text information relevant to entity in each text sentence, obtain each entity in each text sentence
Emotion;
Step 5, according to the emotion of entity each in each text sentence, obtain emotion of each entity in entire full text in this,
Method particularly includes:
Step 5.1 carries out vectorization to the related emotion information of each entity, converts term vector for emotion information;
Step 5.2, by external emotional term database to the related emotion information additive phrase polarity of each entity, obtain every
Then term vector and feeling polarities vector are combined into an entirety by the feeling polarities vector of a term vector, then each sentence
All related emotion informations of each entity are all integrated into an entirety in son, and emotion information belonging to each entity is gathered
Zoarium is used as an input vector;
Step 5.3, building LSTM neural network structure;
The input vector for each entity that step 5.2 is constituted is input to the LSTM neural network that step 5.3 constructs by step 5.4
In the neuron of structure, the polymerization of emotion information is carried out using LSTM neural network structure;
Step 5.5, each entity for being exported according to the parallel judgment step 5.4 of the related emotion information polymerization result of each entity
Feeling polarities.
2. the method for multiple entity sentiment analysis in a kind of long text according to claim 1, it is characterised in that: step 5.3
The LSTM neural network structure of the building includes two layers, and first layer is physical layer neural network, and the second layer is sentence layer nerve
Network;The physical layer neural network receives the input of entity emotion information, and timing memory step-length is sentence quantity * entity number
Amount, and emotion information of the same entity under different sentences can be received;The sentence layer neural network receiving entity layer nerve
The output of network, the result in physical layer neural network after every physical quantities timing memory step-length is as sentence layer nerve net
One input of network, sentence layer neural network can finally export result.
3. the method for multiple entity sentiment analysis in a kind of long text according to claim 2, it is characterised in that: step 5.4
It is described to carry out emotion information polymerization using LSTM neural network structure method particularly includes:
It is input to emotion information condensate belonging to each entity as the related emotion information of some entity under some sentence
In LSTM neural network structure, the timing memory step-length of each LSTM is made to remember the emotion letter of certain entity in some sentence
Breath;The hidden state of the last one entity can be input to the LSTM unit in sentence layer neural network, next without being enter into
First entity LSTM unit in a sentence;By the LSTM unit in sentence layer neural network again to the reality occurred in this sentence
Body and relevant information are polymerize, and are adjusted conducive to polymerization result;And the LSTM unit in sentence layer neural network be with
In sentence layer neural network LSTM unit connection, therefore carry out be all sentence surface entity information filtering, each
The information of entity in the text is effectively polymerize;The related emotion information of each entity is poly- after finally output has traversed full text originally
Close result.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811283252.4A CN109325238A (en) | 2018-10-31 | 2018-10-31 | A kind of method of multiple entity sentiment analysis in long text |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811283252.4A CN109325238A (en) | 2018-10-31 | 2018-10-31 | A kind of method of multiple entity sentiment analysis in long text |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109325238A true CN109325238A (en) | 2019-02-12 |
Family
ID=65259343
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811283252.4A Pending CN109325238A (en) | 2018-10-31 | 2018-10-31 | A kind of method of multiple entity sentiment analysis in long text |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109325238A (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110287405A (en) * | 2019-05-21 | 2019-09-27 | 百度在线网络技术(北京)有限公司 | The method, apparatus and storage medium of sentiment analysis |
CN111027322A (en) * | 2019-12-13 | 2020-04-17 | 新华智云科技有限公司 | Sentiment dictionary-based sentiment analysis method for fine-grained entities in financial news |
CN111241234A (en) * | 2019-12-27 | 2020-06-05 | 北京百度网讯科技有限公司 | Text classification method and device |
CN113065331A (en) * | 2021-04-15 | 2021-07-02 | 上海金融期货信息技术有限公司 | Entity emotion recognition method and system based on entity context discrimination |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103631961A (en) * | 2013-12-17 | 2014-03-12 | 苏州大学张家港工业技术研究院 | Method for identifying relationship between sentiment words and evaluation objects |
KR101628956B1 (en) * | 2015-03-09 | 2016-06-09 | 서울대학교산학협력단 | Method of generating emotion expressing text and apparatus thereof |
CN108170681A (en) * | 2018-01-15 | 2018-06-15 | 中南大学 | Text emotion analysis method, system and computer readable storage medium |
CN108446275A (en) * | 2018-03-21 | 2018-08-24 | 北京理工大学 | Long text emotional orientation analytical method based on attention bilayer LSTM |
-
2018
- 2018-10-31 CN CN201811283252.4A patent/CN109325238A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103631961A (en) * | 2013-12-17 | 2014-03-12 | 苏州大学张家港工业技术研究院 | Method for identifying relationship between sentiment words and evaluation objects |
KR101628956B1 (en) * | 2015-03-09 | 2016-06-09 | 서울대학교산학협력단 | Method of generating emotion expressing text and apparatus thereof |
CN108170681A (en) * | 2018-01-15 | 2018-06-15 | 中南大学 | Text emotion analysis method, system and computer readable storage medium |
CN108446275A (en) * | 2018-03-21 | 2018-08-24 | 北京理工大学 | Long text emotional orientation analytical method based on attention bilayer LSTM |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110287405A (en) * | 2019-05-21 | 2019-09-27 | 百度在线网络技术(北京)有限公司 | The method, apparatus and storage medium of sentiment analysis |
CN110287405B (en) * | 2019-05-21 | 2021-06-01 | 百度在线网络技术(北京)有限公司 | Emotion analysis method, emotion analysis device and storage medium |
CN111027322A (en) * | 2019-12-13 | 2020-04-17 | 新华智云科技有限公司 | Sentiment dictionary-based sentiment analysis method for fine-grained entities in financial news |
CN111241234A (en) * | 2019-12-27 | 2020-06-05 | 北京百度网讯科技有限公司 | Text classification method and device |
CN111241234B (en) * | 2019-12-27 | 2023-07-18 | 北京百度网讯科技有限公司 | Text classification method and device |
CN113065331A (en) * | 2021-04-15 | 2021-07-02 | 上海金融期货信息技术有限公司 | Entity emotion recognition method and system based on entity context discrimination |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109325238A (en) | A kind of method of multiple entity sentiment analysis in long text | |
Fabbri et al. | ConvoSumm: Conversation summarization benchmark and improved abstractive summarization with argument mining | |
CN111143576A (en) | Event-oriented dynamic knowledge graph construction method and device | |
CN106886567B (en) | Microblogging incident detection method and device based on semantic extension | |
CN108628828A (en) | A kind of joint abstracting method of viewpoint and its holder based on from attention | |
CN107766371A (en) | A kind of text message sorting technique and its device | |
CN107688576B (en) | Construction and tendency classification method of CNN-SVM model | |
CN109344234A (en) | Machine reads understanding method, device, computer equipment and storage medium | |
CN104268160A (en) | Evaluation object extraction method based on domain dictionary and semantic roles | |
CN107463553A (en) | For the text semantic extraction, expression and modeling method and system of elementary mathematics topic | |
CN105975478A (en) | Word vector analysis-based online article belonging event detection method and device | |
CN109558492A (en) | A kind of listed company's knowledge mapping construction method and device suitable for event attribution | |
CN107357785A (en) | Theme feature word abstracting method and system, feeling polarities determination methods and system | |
CN109101489A (en) | A kind of text automatic abstracting method, device and a kind of electronic equipment | |
Fan et al. | Humor detection via an internal and external neural network | |
CN116561592B (en) | Training method of text emotion recognition model, text emotion recognition method and device | |
Van Hee et al. | Monday mornings are my fave:)# not exploring the automatic recognition of irony in english tweets | |
CN106503256A (en) | A kind of hot information method for digging based on social networkies document | |
Ashraf et al. | Author profiling on bi-lingual tweets | |
CN117114112B (en) | Vertical field data integration method, device, equipment and medium based on large model | |
Xu et al. | Gar-net: A graph attention reasoning network for conversation understanding | |
CN106445914A (en) | Microblog emotion classifier establishing method and device | |
Kuila et al. | A Neural Network based Event Extraction System for Indian Languages. | |
Kavitha et al. | Study of Natural Language Processing for Sentiment Analysis | |
CN107818078B (en) | Semantic association and matching method for Chinese natural language dialogue |
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: 20190212 |