CN108595601A - A kind of long text sentiment analysis method incorporating Attention mechanism - Google Patents
A kind of long text sentiment analysis method incorporating Attention mechanism Download PDFInfo
- Publication number
- CN108595601A CN108595601A CN201810357279.7A CN201810357279A CN108595601A CN 108595601 A CN108595601 A CN 108595601A CN 201810357279 A CN201810357279 A CN 201810357279A CN 108595601 A CN108595601 A CN 108595601A
- Authority
- CN
- China
- Prior art keywords
- hidden layer
- hidden
- text
- weight matrix
- state
- 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
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/40—Processing or translation of natural language
- G06F40/58—Use of machine translation, e.g. for multi-lingual retrieval, for server-side translation for client devices or for real-time translation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computational Linguistics (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Biomedical Technology (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Machine Translation (AREA)
Abstract
The present invention relates to a kind of long text sentiment analysis methods incorporating Attention mechanism, and text emotion disaggregated model is established using the two-way thresholding Recognition with Recurrent Neural Network of Attention mechanism is combined（Bi‑Attention model）.Attention can allow neural network that can be concerned about the important information in text, while ignore or reducing the influence of the secondary information in text, to reduce the complexity of long article present treatment.On the other hand context vector is generated by two-way thresholding cycling element, updates memory state, to fully consider historical information with Future Information to semantic influence.
Description
Technical field
The present invention relates to a kind of long text sentiment analysis methods incorporating Attention mechanism
Background technology
Text emotion analysis (also referred to as opinion mining) is to use natural language processing, text mining and Computational Linguistics
The methods of identify and extract the subjective information in essence material.These subjective texts are increased with exponential speed daily, are adopted
The emotion that the expression of these subjective texts is automatically analyzed with computer, becomes a hot spot of academic circles at present.Society at present
It is mostly short text data to hand over the comment data in network, because longer sequence data or chapter data, may contain abundant
Emotion information, it is also possible to include the information unrelated to current sentiment analysis.
The short text comment data in social networks is analyzed usually using unidirectional neural network, due to unidirectional neural network
Consider history and current information, Future Information can not also be modeled, with the continuous growth of text sequence, tradition is unidirectional
LSTM network trainings can not efficiently solve Long-range dependence problem, and the ability of the contextual information of capture is limited.Thus draw
Shen goes out bidirectional circulating neural network (Bi-directional Recurrent Neural Network, Bi-RNN).
Attention mechanism (Attention Mechanism) can be concentrated on image specific part by human vision and be inspired,
It is applied to visual pattern field earliest, the neural network machine translation being subsequently applied in natural language processing task
In (Neural Machine Translation, NMT).Application in translation is that attention mechanism is introduced into Encoder-
In Decoder frames, source language and the target language are effectively associated by perceptron formula, are aligned one by one, obtained general
Rate is distributed.Attention mechanism not only obtains effect outstanding, while the mark in terms of image/video, reading reason in machine translation
Solution etc. has obtained preferable application.
Invention content
In view of this, the purpose of the present invention is to provide a kind of long text sentiment analysis sides incorporating Attention mechanism
Method is used to analyze the emotional color of long article notebook data in social networks.
To achieve the above object, the present invention adopts the following technical scheme that：
A kind of long text sentiment analysis method incorporating Attention mechanism, it is characterised in that：
Step S1:The word in text is initialized with term vector, text is mapped to term vector set, and by word to
Input of the duration set as network, and it is input to hidden layer；
Step S2:GRU units are introduced in hidden layer and calculate hidden state, are increased reversed network, are recycled using bidirectional gate
Modelon Modeling obtains each term vector in term vector set the information of its context, obtains the current hidden state of hidden layer
ht；
Step S3：Give hidden layer current hidden state htAddition attention mechanism, determined by way of weighting automatically
Input text needs the part paid close attention to, and obtains the probability distribution of sentence vector S；
Step S4：According to the probability distribution of sentence vector S, the general of emotional category is judged with full articulamentum and softmax functions
Rate is distributed, and is enable model preferably to characterize long text by the probability distribution of emotional category, is captured text key message.
Further, the step S2 is specifically included：
It is modeled using bidirectional gate cycling element, then the more new formula of GRU units is as follows：
Update door ztCalculation formula：
zt=σ (Wzxt+Uzht-1) (1)
Wherein, xtFor the term vector of the input at current time, σ is logistics functions, WZIt is defeated for current time hidden layer
Enter to update door ztWeight matrix, UZIt is input to update door z for last moment hidden layertWeight matrix, ht-1It is hidden layer
The hidden state of last moment；
Reset door rtUpdate mode：
rt=σ (Wrxt+Urht-1) (2)
Wherein WrIt is input to resetting door z for current time hidden layerrWeight matrix, UrIt is inputted for last moment hidden layer
To resetting door zrWeight matrix, by the historical information before ignoring, do not influence following output when it is 1 to reset gate value；
Node state of the memory reset cell at current time
Wherein, tanh is hyperbolic tangent function,It indicates by element multiplication；W, U are the weight matrix to be trained；
The current hidden state h of hidden layertBy update door zt, resetting door rtWith the node shape at memory reset cell current time
StateIt codetermines, formula is：
Wherein, when updating door close to 1, new hidden state almost depends on last state.
Further, the step S3 is specifically included following：
According to the current hidden state h of hidden layertIt can obtain hidden layer expression：
μt=tanh (Wwht+bw) (5)
Wherein, WwIt is the weight matrix of hidden layer, bwIt is bias
Different weight matrix α is distributed for the output of each of last layert：
Wherein, μwIt is the context vector of word rank；
The hidden state h current to hidden layertWith weight matrix αtWeighting is averaging, and obtains sentence vector S probability distribution, formula
For：
Further, the step S2 can also introduce LSTM units in hidden layer and calculate hidden state, add reversed net
Network obtains each word in sentence the information of its context.
The present invention has the advantages that compared with prior art：
The present invention introduces context measurement and attention mechanism, is obtained from long text on the basis of Recognition with Recurrent Neural Network
Take more sufficient emotional expression.Text emotion is established using the two-way thresholding Recognition with Recurrent Neural Network of Attention mechanism is combined
Disaggregated model (Bi-Attention model).Attention can allow neural network that can be concerned about the important information in text,
The influence for ignoring or reducing the secondary information in text simultaneously, to reduce the complexity of long article present treatment.On the other hand pass through
Two-way thresholding cycling element generates context vector, updates memory state, to fully consider historical information with Future Information to language
The influence of justice.
Description of the drawings
Fig. 1 is the two-way GRU schematic diagrames of the present invention
Fig. 2 is GRU structure charts of the present invention
Fig. 3 is LSTM structure charts of the present invention
Specific implementation mode
The present invention will be further described with reference to the accompanying drawings and embodiments.
Referring to Fig.1 and 2, the present invention provides a kind of long text sentiment analysis method incorporating Attention mechanism,
It is characterized in that：
Step S1:The word in text is initialized with term vector, text is mapped to term vector set, and by word to
The input as network is measured, and is input to hidden layer；
Step S2:GRU units are introduced in hidden layer and calculate hidden state, increase reversed network, to each of sentence
Word obtains the information of its context；It specifically includes：
It is modeled using bidirectional gate cycling element, then the more new formula of GRU units is as follows：
Update door ztCalculation formula：
zt=σ (Wzxt+Uzht-1) (1)
Wherein, xtFor the term vector of the input at current time, σ is logistics functions, WZIt is defeated for current time hidden layer
Enter to update door ztWeight matrix, UZIt is input to update door z for last moment hidden layertWeight matrix, ht-1It is hidden layer
The hidden state of last moment；
Reset door rtUpdate mode：
rt=σ (Wrxt+Urht-1) (2)
Wherein WrIt is input to resetting door z for current time hidden layerrWeight matrix, UrIt is inputted for last moment hidden layer
To resetting door zrWeight matrix, by the historical information before ignoring, do not influence following output when it is 1 to reset gate value；
Node state of the memory reset cell of GRU at current time
Wherein, tanh is hyperbolic tangent function,It indicates by element multiplication；W, U are the weight matrix to be trained；
The current hidden state h of hidden layertBy update door zt, resetting door rtWith the node shape at memory reset cell current time
StateIt codetermines, formula is：
Wherein, when updating door close to 1, new hidden state almost depends on last state.
Step S3：Give hidden layer current hidden state htAddition attention mechanism, determined by way of weighting automatically
Input text needs the part paid close attention to, and is that the output of each of last layer distributes different weight αstWeighting is averaging, and obtains probability
Distribution, calculation formula are as follows：
μt=tanh (Wwht+bw) (5)
Wherein, μtIt is that hidden layer indicates, WwIt is the weight matrix of hidden layer, htIt is the hidden state of t moment, bwIt is bias,
μwIt is the context vector of word rank, s is sentence vector；
Step S4：The probability that emotional category is finally judged with full articulamentum and softmax functions, passes through the general of emotional category
Rate distribution enables model preferably to characterize long text, captures text key message.
With reference to Fig. 3, in an embodiment of the present invention, further, the step S2 can also be introduced in hidden layer
LSTM units calculate hidden state, add reversed network, the information of its context is obtained to each word in sentence.
The foregoing is merely presently preferred embodiments of the present invention, all equivalent changes done according to scope of the present invention patent with
Modification should all belong to the covering scope of the present invention.
Claims (4)
1. a kind of long text sentiment analysis method incorporating Attention mechanism, it is characterised in that：
Step S1:The word in text is initialized with term vector, text is mapped to term vector set, and by term vector collection
Cooperation is the input of network, and is input to hidden layer；
Step S2:GRU units are introduced in hidden layer and calculate hidden state, increase reversed network, using bidirectional gate cycling element
Modeling, obtains each term vector in term vector set the information of its context, obtains the current hidden state h of hidden layert；
Step S3：Give hidden layer current hidden state htAddition attention mechanism, determined by way of weighting automatically input text
The part that this needs is paid close attention to obtains the probability distribution of sentence vector S；
Step S4：According to the probability distribution of sentence vector S, the probability of emotional category is judged with full articulamentum and softmax functions, is led to
Crossing probability distribution enables model preferably to characterize long text, captures text key message.
2. a kind of long text sentiment analysis method incorporating Attention mechanism according to claim 1, feature exist
In：The step S2 is specifically included：
It is modeled using bidirectional gate cycling element, then the more new formula of GRU units is as follows：
Update door ztCalculation formula：
zt=σ (Wzxt+Uzht-1) (1)
Wherein, xtFor the term vector of the input at current time, σ is logistics functions, WZIt is input to for current time hidden layer
Update door ztWeight matrix, UZIt is input to update door z for last moment hidden layertWeight matrix, ht-1It is one on hidden layer
The hidden state at moment；
Reset door rtCalculation：
rt=σ (Wrxt+Urht-1) (2)
Wherein WrIt is input to resetting door z for current time hidden layerrWeight matrix, UrIt is input to weight for last moment hidden layer
Set a zrWeight matrix, by the historical information before ignoring, do not influence following output when it is 1 to reset gate value；
Node state of the memory reset cell at current time
Wherein, tanh is hyperbolic tangent function,It indicates by element multiplication；W, U are the weight matrix to be trained；
The current hidden state h of hidden layertBy update door zt, resetting door rtWith the node state at memory reset cell current timeAltogether
With decision, formula is：
Wherein, when updating door close to 1, new hidden state almost depends on last state.
3. a kind of long text sentiment analysis method incorporating Attention mechanism according to claim 1, feature exist
In：The step S3 specifically includes following：
According to the current hidden state h of hidden layertIt can obtain hidden layer expression：
μt=tanh (Wwht+bw) (5)
Wherein, WwIt is the weight matrix of hidden layer, bwIt is bias
Different weight matrix α is distributed for the output of each of last layert：
Wherein, μwIt is the context vector of word rank；
The hidden state h current to hidden layertWith weight matrix αtWeighting is averaging, and obtains sentence vector S probability distribution, formula is：
4. a kind of long text sentiment analysis method incorporating Attention mechanism according to claim 1, feature exist
In：The step S2 can also introduce LSTM units in hidden layer and calculate hidden state, add reversed network, to every in sentence
A word obtains the information of its context.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810357279.7A CN108595601A (en) | 2018-04-20 | 2018-04-20 | A kind of long text sentiment analysis method incorporating Attention mechanism |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810357279.7A CN108595601A (en) | 2018-04-20 | 2018-04-20 | A kind of long text sentiment analysis method incorporating Attention mechanism |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108595601A true CN108595601A (en) | 2018-09-28 |
Family
ID=63613403
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810357279.7A Pending CN108595601A (en) | 2018-04-20 | 2018-04-20 | A kind of long text sentiment analysis method incorporating Attention mechanism |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108595601A (en) |
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109447129A (en) * | 2018-09-29 | 2019-03-08 | 平安科技（深圳）有限公司 | A kind of multi-mode Emotion identification method, apparatus and computer readable storage medium |
| CN109508642A (en) * | 2018-10-17 | 2019-03-22 | 杭州电子科技大学 | Ship monitor video key frame extracting method based on two-way GRU and attention mechanism |
| CN109543180A (en) * | 2018-11-08 | 2019-03-29 | 中山大学 | A kind of text emotion analysis method based on attention mechanism |
| CN109726745A (en) * | 2018-12-19 | 2019-05-07 | 北京理工大学 | A kind of sensibility classification method based on target incorporating description knowledge |
| CN109902174A (en) * | 2019-02-18 | 2019-06-18 | 山东科技大学 | A kind of feeling polarities detection method of the memory network relied on based on aspect |
| CN110134757A (en) * | 2019-04-19 | 2019-08-16 | 杭州电子科技大学 | A kind of event argument roles abstracting method based on bull attention mechanism |
| CN110147446A (en) * | 2019-04-19 | 2019-08-20 | 中国地质大学（武汉） | A kind of word embedding grammar based on the double-deck attention mechanism, equipment and storage equipment |
| CN110223429A (en) * | 2019-06-19 | 2019-09-10 | 上海应用技术大学 | Voice access control system |
| CN110378335A (en) * | 2019-06-17 | 2019-10-25 | 杭州电子科技大学 | A kind of information analysis method neural network based and model |
| CN110457480A (en) * | 2019-08-16 | 2019-11-15 | 国网天津市电力公司 | The construction method of fine granularity sentiment classification model based on interactive attention mechanism |
| CN111027313A (en) * | 2018-10-08 | 2020-04-17 | 中国科学院沈阳计算技术研究所有限公司 | BiGRU judgment result tendency analysis method based on attention mechanism |
| CN111353040A (en) * | 2019-05-29 | 2020-06-30 | 北京工业大学 | GRU-based attribute level emotion analysis method |
| CN111368524A (en) * | 2020-03-05 | 2020-07-03 | 昆明理工大学 | Microblog viewpoint sentence recognition method based on self-attention bidirectional GRU and SVM |
| WO2020147409A1 (en) * | 2019-01-14 | 2020-07-23 | 平安科技（深圳）有限公司 | Text classification method and apparatus, computer device, and storage medium |
| CN111671426A (en) * | 2020-05-13 | 2020-09-18 | 北京航空航天大学 | Human body respiration state monitoring system and method based on flexible sensing and deep learning |
| WO2020211611A1 (en) * | 2019-04-17 | 2020-10-22 | 腾讯科技（深圳）有限公司 | Method and device for generating hidden state in recurrent neural network for language processing |
| WO2021057424A1 (en) * | 2019-09-23 | 2021-04-01 | 腾讯科技（深圳）有限公司 | Virtual image behavior control method and device based on text, and medium |
| CN112782762A (en) * | 2021-01-29 | 2021-05-11 | 东北大学 | Earthquake magnitude determination method based on deep learning |
| CN117688974A (en) * | 2024-02-01 | 2024-03-12 | 中国人民解放军总医院 | Knowledge graph-based generation type large model modeling method, system and equipment |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107092596A (en) * | 2017-04-24 | 2017-08-25 | 重庆邮电大学 | Text emotion analysis method based on attention CNNs and CCR |
| CN107239446A (en) * | 2017-05-27 | 2017-10-10 | 中国矿业大学 | A kind of intelligence relationship extracting method based on neutral net Yu notice mechanism |
| WO2018057809A1 (en) * | 2016-09-22 | 2018-03-29 | Salesforce.Com, Inc. | Pointer sentinel mixture architecture |
| CN107909421A (en) * | 2017-09-29 | 2018-04-13 | 中国船舶重工集团公司第七0九研究所 | A kind of implicit feedback of more GRU layers of neutral net based on user's space recommends method and system |
-
2018
- 2018-04-20 CN CN201810357279.7A patent/CN108595601A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2018057809A1 (en) * | 2016-09-22 | 2018-03-29 | Salesforce.Com, Inc. | Pointer sentinel mixture architecture |
| CN107092596A (en) * | 2017-04-24 | 2017-08-25 | 重庆邮电大学 | Text emotion analysis method based on attention CNNs and CCR |
| CN107239446A (en) * | 2017-05-27 | 2017-10-10 | 中国矿业大学 | A kind of intelligence relationship extracting method based on neutral net Yu notice mechanism |
| CN107909421A (en) * | 2017-09-29 | 2018-04-13 | 中国船舶重工集团公司第七0九研究所 | A kind of implicit feedback of more GRU layers of neutral net based on user's space recommends method and system |
Non-Patent Citations (1)
| Title |
|---|
| 郑雄风等: "基于用户和产品Attention机制的层次BGRU模型", 《计算机工程与应用》 * |
Cited By (28)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109447129B (en) * | 2018-09-29 | 2023-04-18 | 平安科技（深圳）有限公司 | Multi-mode emotion recognition method and device and computer readable storage medium |
| CN109447129A (en) * | 2018-09-29 | 2019-03-08 | 平安科技（深圳）有限公司 | A kind of multi-mode Emotion identification method, apparatus and computer readable storage medium |
| CN111027313A (en) * | 2018-10-08 | 2020-04-17 | 中国科学院沈阳计算技术研究所有限公司 | BiGRU judgment result tendency analysis method based on attention mechanism |
| CN109508642A (en) * | 2018-10-17 | 2019-03-22 | 杭州电子科技大学 | Ship monitor video key frame extracting method based on two-way GRU and attention mechanism |
| CN109508642B (en) * | 2018-10-17 | 2021-08-17 | 杭州电子科技大学 | Ship monitoring video key frame extraction method based on bidirectional GRU and attention mechanism |
| CN109543180A (en) * | 2018-11-08 | 2019-03-29 | 中山大学 | A kind of text emotion analysis method based on attention mechanism |
| CN109543180B (en) * | 2018-11-08 | 2020-12-04 | 中山大学 | Text emotion analysis method based on attention mechanism |
| CN109726745A (en) * | 2018-12-19 | 2019-05-07 | 北京理工大学 | A kind of sensibility classification method based on target incorporating description knowledge |
| CN109726745B (en) * | 2018-12-19 | 2020-10-09 | 北京理工大学 | Target-based emotion classification method integrating description knowledge |
| WO2020147409A1 (en) * | 2019-01-14 | 2020-07-23 | 平安科技（深圳）有限公司 | Text classification method and apparatus, computer device, and storage medium |
| CN109902174B (en) * | 2019-02-18 | 2023-06-20 | 山东科技大学 | Emotion polarity detection method based on aspect-dependent memory network |
| CN109902174A (en) * | 2019-02-18 | 2019-06-18 | 山东科技大学 | A kind of feeling polarities detection method of the memory network relied on based on aspect |
| WO2020211611A1 (en) * | 2019-04-17 | 2020-10-22 | 腾讯科技（深圳）有限公司 | Method and device for generating hidden state in recurrent neural network for language processing |
| CN110134757B (en) * | 2019-04-19 | 2020-04-07 | 杭州电子科技大学 | Event argument role extraction method based on multi-head attention mechanism |
| CN110134757A (en) * | 2019-04-19 | 2019-08-16 | 杭州电子科技大学 | A kind of event argument roles abstracting method based on bull attention mechanism |
| CN110147446A (en) * | 2019-04-19 | 2019-08-20 | 中国地质大学（武汉） | A kind of word embedding grammar based on the double-deck attention mechanism, equipment and storage equipment |
| CN111353040A (en) * | 2019-05-29 | 2020-06-30 | 北京工业大学 | GRU-based attribute level emotion analysis method |
| CN110378335A (en) * | 2019-06-17 | 2019-10-25 | 杭州电子科技大学 | A kind of information analysis method neural network based and model |
| CN110223429A (en) * | 2019-06-19 | 2019-09-10 | 上海应用技术大学 | Voice access control system |
| CN110457480A (en) * | 2019-08-16 | 2019-11-15 | 国网天津市电力公司 | The construction method of fine granularity sentiment classification model based on interactive attention mechanism |
| WO2021057424A1 (en) * | 2019-09-23 | 2021-04-01 | 腾讯科技（深圳）有限公司 | Virtual image behavior control method and device based on text, and medium |
| US11714879B2 (en) | 2019-09-23 | 2023-08-01 | Tencent Technology (Shenzhen) Company Limited | Method and device for behavior control of virtual image based on text, and medium |
| CN111368524A (en) * | 2020-03-05 | 2020-07-03 | 昆明理工大学 | Microblog viewpoint sentence recognition method based on self-attention bidirectional GRU and SVM |
| CN111671426A (en) * | 2020-05-13 | 2020-09-18 | 北京航空航天大学 | Human body respiration state monitoring system and method based on flexible sensing and deep learning |
| CN111671426B (en) * | 2020-05-13 | 2022-07-12 | 北京航空航天大学 | Human body respiration state monitoring system and method based on flexible sensing and deep learning |
| CN112782762A (en) * | 2021-01-29 | 2021-05-11 | 东北大学 | Earthquake magnitude determination method based on deep learning |
| CN117688974A (en) * | 2024-02-01 | 2024-03-12 | 中国人民解放军总医院 | Knowledge graph-based generation type large model modeling method, system and equipment |
| CN117688974B (en) * | 2024-02-01 | 2024-04-26 | 中国人民解放军总医院 | Knowledge graph-based generation type large model modeling method, system and equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108595601A (en) | A kind of long text sentiment analysis method incorporating Attention mechanism | |
| Wang et al. | Learning visual relationship and context-aware attention for image captioning | |
| Si et al. | Skeleton-based action recognition with hierarchical spatial reasoning and temporal stack learning network | |
| CN108984530A (en) | A kind of detection method and detection system of network sensitive content | |
| Ke et al. | Leveraging structural context models and ranking score fusion for human interaction prediction | |
| Zhang et al. | Object semantics sentiment correlation analysis enhanced image sentiment classification | |
| Wang et al. | Discovering attractive segments in the user-generated video streams | |
| WO2020238353A1 (en) | Data processing method and apparatus, storage medium, and electronic apparatus | |
| CN108427740B (en) | Image emotion classification and retrieval algorithm based on depth metric learning | |
| Yan et al. | Cross-domain facial expression recognition based on transductive deep transfer learning | |
| CN110210358A (en) | A kind of video presentation generation method and device based on two-way timing diagram | |
| Wang et al. | Exploiting topic-based adversarial neural network for cross-domain keyphrase extraction | |
| Xiao et al. | Chinese sentiment analysis using bidirectional LSTM with word embedding | |
| CN112733764A (en) | Method for recognizing video emotion information based on multiple modes | |
| Shen et al. | Emotion analysis of ideological and political education using a GRU deep neural network | |
| CN115934951A (en) | Network hot topic user emotion prediction method | |
| CN113244627B (en) | Method and device for identifying plug-in, electronic equipment and storage medium | |
| CN117313709B (en) | Method for detecting generated text based on statistical information and pre-training language model | |
| Zhu et al. | NAGNet: A novel framework for real‐time students' sentiment analysis in the wisdom classroom | |
| Yan et al. | Image captioning based on a hierarchical attention mechanism and policy gradient optimization | |
| Ke et al. | Spatial, structural and temporal feature learning for human interaction prediction | |
| Peng et al. | Unsupervised visual–textual correlation learning with fine-grained semantic alignment | |
| Jasmir et al. | Feature Extraction for Improvement Text Classification of Spam YouTube Video Comment using Deep Learning | |
| Li et al. | Frame aggregation and multi-modal fusion framework for video-based person recognition | |
| Li et al. | [Retracted] Human Sports Action and Ideological and PoliticalEvaluation by Lightweight Deep Learning Model |
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 |
Application publication date: 20180928 |
|
| RJ01 | Rejection of invention patent application after publication |