CN115906824A - Text fine-grained emotion analysis method, system, medium and computing equipment - Google Patents
Text fine-grained emotion analysis method, system, medium and computing equipment Download PDFInfo
- Publication number
- CN115906824A CN115906824A CN202211441031.1A CN202211441031A CN115906824A CN 115906824 A CN115906824 A CN 115906824A CN 202211441031 A CN202211441031 A CN 202211441031A CN 115906824 A CN115906824 A CN 115906824A
- Authority
- CN
- China
- Prior art keywords
- text
- fine
- grained
- data set
- emotion analysis
- 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
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Abstract
The invention discloses a text fine-grained emotion analysis method, a text fine-grained emotion analysis system, a text fine-grained emotion analysis medium and a text fine-grained emotion analysis computing device, wherein the text fine-grained emotion analysis method comprises the following steps: obtaining a comment text data set, and preprocessing the comment text data set; performing word vectorization on the comment text in the preprocessed data set by adopting a BERT model; inputting the comment text word vector into a BilSTM + Attention model for coarse-grained emotion analysis, and training to obtain a neural network model for emotion analysis; adopting an LDA topic model to perform topic extraction on the preprocessed data set to obtain a topic-attribute word; short sentences containing attribute words in the data set are screened, and corresponding topics are marked; and inputting the short sentence sets marked with the themes into the trained neural network model for fine-grained emotional analysis to obtain the emotional tendency of each theme. The emotion analysis method based on Bert + BiLSTM + LDA can effectively improve the accuracy of text fine-grained emotion analysis.
Description
Technical Field
The invention belongs to the technical field of natural language processing, relates to a text fine-grained sentiment analysis method, and particularly relates to a text fine-grained sentiment analysis method, a text fine-grained sentiment analysis system, a text fine-grained sentiment analysis medium and a text fine-grained sentiment analysis computing device based on BERT + BilSTM + LDA.
Background
With the rapid development and popularization of the internet, social networks permeate into the aspects of life of people, and more users make comments and opinions on the social networks; the development of e-commerce websites also enables more and more people to purchase goods on the internet and publish their own evaluations and opinions on the goods. People leave a great amount of text comments on commodities, movies, books and the like on the internet, and the text comments contain emotional tendency information of people. The text comments generally contain emotional tendencies in multiple aspects, for example, books may contain evaluations on multiple aspects such as scenarios and pens, and the reviewers may be positive on the emotional tendencies of the scenarios and negative on the pens, and analyze the emotional tendencies of the text comments on the various aspects, namely, fine-grained emotional analysis of the text, which is a difficulty in the field of natural language processing.
Emotional analysis is a popular direction in the field of natural language processing, and mainly analyzes viewpoints, attitudes and preferences expressed in texts. According to the granularity division of the analysis, the method can be divided into chapter level emotion analysis, sentence level emotion analysis and attribute level emotion analysis, wherein the attribute level emotion analysis is also called fine-grained emotion analysis. The fine-grained sentiment analysis can be mainly divided into two steps: extracting and identifying themes (aspects) described in the text, and performing sentiment analysis on sentiment tendency of each theme; wherein, a machine learning method is generally adopted in theme extraction, such as a PageRank algorithm, an LDA theme model, a text clustering method based on HowNet and the like; the method for analyzing the emotion of the emotional tendency of each theme mainly comprises a method based on an emotion dictionary and a method based on traditional machine learning. The method based on the emotion dictionary obtains the emotion scores of the emotion words in the text by searching the emotion dictionary, and obtains the emotion scores of the text through a certain emotion value calculation rule so as to judge the emotion tendency of the text; the method based on the emotion dictionary mostly needs to manually construct the emotion dictionary, and the emotion classification accuracy is poor. The method based on traditional machine learning mainly carries out emotion analysis through classifier models such as naive Bayes, support vector machines and K nearest neighbors.
With the development of deep learning, the method based on deep learning is also widely applied to the emotion analysis problem, and compared with the method based on the emotion dictionary and the method based on the traditional machine learning, the accuracy is greatly improved; however, on the aspect of fine-grained emotion analysis, the application of a deep learning method is less, and the accuracy rate further improves the space.
Disclosure of Invention
Aiming at the defects in the prior art, the invention provides a text fine-grained emotion analysis method, a text fine-grained emotion analysis system, a text fine-grained emotion analysis medium and a text fine-grained emotion analysis computing device.
The invention discloses a text fine-grained emotion analysis method, which comprises the following steps:
acquiring a comment text data set;
preprocessing the comment text data set, wherein the preprocessing comprises data cleaning and data labeling;
performing word vectorization on the comment text in the preprocessed data set by adopting a BERT model to obtain a comment text word vector;
inputting the comment text word vector into a BilSTM + Attention model for coarse-grained emotion analysis, and training to obtain a neural network model for emotion analysis;
adopting an LDA theme model to perform theme extraction on the preprocessed data set to obtain theme-attribute words; short sentences containing attribute words in the data set are screened, and corresponding topics are marked;
and inputting the short sentence sets marked with the themes into the trained neural network model for fine-grained emotional analysis to obtain the emotional tendency of each theme.
As a further improvement of the invention, a web crawler technology is adopted to obtain the comment text data set.
As a further improvement of the invention, the data cleaning comprises removing irrelevant texts and repeated comment texts, wherein the irrelevant texts comprise but are not limited to abbreviations, emoticons, repeated punctuation marks and sentences with unknown meanings;
the data labeling method comprises but is not limited to labeling the comments by using the rating information of the comments, wherein the scores of 0-2 are labeled as negative, the scores of 3 are labeled as neutral, and the scores of 4-5 are labeled as positive.
As a further improvement of the invention, the comment text word vector is input into a BilSTM + Attention model for coarse grain emotion analysis, and a neural network model for emotion analysis is obtained through training; the method comprises the following steps:
extracting implicit information of each word in the comment text sentence by using the BilSTM;
carrying out weighted fusion on the hidden information of each word by adopting an attention mechanism to obtain the hidden information of the whole sentence;
implicit information of the whole sentence is output through a full connection layer and a softmax activation function, emotional tendency of the whole sentence is predicted, and finally a neural network model for emotion analysis is obtained through training.
As a further improvement of the invention, the LDA topic model is adopted to perform topic extraction on the preprocessed data set to obtain a topic-attribute word; short sentences containing attribute words in the data set are screened, and corresponding topics are marked; the method comprises the following steps:
performing word segmentation on the preprocessed data set;
inputting the text after word segmentation into an LDA topic model, and extracting topics according to the number of preset topics to finally obtain N topics and M attribute words related to the N topics;
and screening short sentences with the attribute words according to the obtained attribute words, and labeling the short sentences with corresponding topics according to the attribute words.
As a further improvement of the invention, when a plurality of attribute words are screened out from a sentence of comment, the comment needs to be segmented to obtain a plurality of short sentences with the attribute words.
The invention also discloses a text fine-grained sentiment analysis system, which comprises the following steps:
the obtaining module is used for obtaining a comment text data set;
the preprocessing module is used for preprocessing the comment text data set, and the preprocessing comprises data cleaning and data labeling;
the word vectorization module is used for carrying out word vectorization on the comment text in the preprocessed data set by adopting a BERT model to obtain a comment text word vector;
the training module is used for inputting the comment text word vectors into a BilSTM + Attention model for coarse-grained emotion analysis, and training to obtain a neural network model for emotion analysis;
the theme extraction module is used for extracting themes of the preprocessed data set by adopting an LDA theme model to obtain theme-attribute words; short sentences containing attribute words in the data set are screened, and corresponding topics are marked;
and the analysis module is used for inputting the short sentence sets marked with the themes into the trained neural network model for fine-grained sentiment analysis to obtain the sentiment tendency of each theme.
The invention also discloses a computer readable storage medium, which stores executable instructions, and the instructions can cause a processor to execute the text fine-grained emotion analysis method when being executed by the processor.
The invention also discloses a computing device, comprising: one or more memories storing executable instructions; and one or more processors executing the executable instructions to implement the text fine-grained emotion analysis method.
Compared with the prior art, the invention has the following beneficial effects:
the invention introduces the related technology in deep learning to carry out fine-grained emotion analysis, for example, a BERT model is adopted to carry out text word vectorization, and the generated word vector contains deep semantic information and also integrates context information; an attention mechanism is added behind the BilSTM model, and when the information of each word is fused, the more important word has higher weight and more information is fused; the finally constructed fine-grained emotion analysis method based on the Bert + BiLSTM + LDA can effectively improve the accuracy of text fine-grained emotion analysis.
Drawings
FIG. 1 is a flowchart of a method for fine-grained sentiment analysis of a text according to an embodiment of the present invention;
FIG. 2 is a block diagram of a Bert model according to an embodiment of the present invention.
FIG. 3 is a block diagram of an LSTM single neuron as disclosed in one embodiment of the present invention.
Fig. 4 is a block diagram of BiLSTM disclosed in an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be obtained by a person skilled in the art without any inventive step based on the embodiments of the present invention, are within the scope of the present invention.
The invention is described in further detail below with reference to the following drawings:
as shown in fig. 1, the present invention provides a text fine-grained emotion analysis method, including:
specifically, the method comprises the following steps:
and crawling the comments of a certain book in the bean book by adopting a web crawler technology to construct a comment text data set, wherein each comment text is used as a sentence or a short sentence.
Step 2, preprocessing the comment text data set, wherein the preprocessing comprises data cleaning and data labeling;
specifically, the method comprises the following steps:
the data cleaning comprises the steps of removing irrelevant texts and repeated comment texts; the original data crawled by the crawler has a plurality of irrelevant texts, such as a large number of abbreviations, expression symbols, repeated punctuations, sentences with unknown meanings and the like, the texts do not contain emotional tendencies and useful information, and only can seriously interfere the text expression, so that the emotional analysis is difficult to be accurate, and therefore, the interfering texts need to be removed; in addition, some comments in the data set are repeated, so that the repeated comments need to be removed;
data annotation includes, but is not limited to, annotating the reviews with their scoring information, such as a score of 0-2 being negative, a score of 3 being neutral, and a score of 4-5 being positive.
Step 3, performing word vectorization on the comment text in the preprocessed data set by adopting a BERT model to obtain a comment text word vector; wherein the content of the first and second substances,
the BERT model is called a transform-based bidirectional encoder, and is a very powerful pre-training model used in the field of natural language processing; the input embedding of the BERT model consists of three embeddings: position embedding, segment embedding, and word embedding, since BERT can take sentence pairs as input, segment embedding represents whether a word is in a first sentence or a second sentence, and position embedding represents the position of a word in a sentence. The structure of the BERT model is shown in FIG. 2, and has the characteristic of bidirectional, which means that information can be extracted from the context of the selected text in the training process; the pretraining process of BERT mainly includes two tasks: through the two tasks, the word vector output of the pre-training Model can accurately represent the deep semantic information of the text combined with the context, and finally the pre-training BERT Model can be used as a word vectorizer to convert the input words into word vectors containing the deep semantic information.
Step 4, inputting the comment text word vector into a BilSTM + Attention model for coarse-grained emotion analysis, and training to obtain a neural network model for emotion analysis;
the method specifically comprises the following steps:
step 4-1: the LSTM is a neural network with the capability of memorizing long-term and short-term information, and can well solve the problem of long-term dependence; the BilSTM is a bidirectional LSTM, is formed by combining a forward LSTM and a backward LSTM, and can learn forward information and backward information at the same time; the block diagram of the LSTM single neuron and the block diagram of the BiLSTM are shown in fig. 3 and 4.
The algorithm formula and analysis of the LSTM single neuron is as follows:
calculating a forgetting gate, selecting information to be forgotten: the hidden state ht-1 at the previous moment and the input word xt at the current moment are input, the value ft of the forgetting gate is output, and the calculation formula is as follows: f. of t ＝σ(W f ×[h t-1 ，x t ]+b f ) Wherein Wf and bf are weight matrix and offset vector.
Calculating a memory gate, selecting information to be memorized: the hidden layer state ht-1 at the previous moment and the input word xt at the current moment are input, and the values it and the temporary cell state of the memory gate are output
The calculation formula is as follows:
where Wi and bi are the weight matrix and offset vector.
Calculating the cell state at the current moment: the input is the value it of the memory gate, the value ft of the forgetting gate, the temporary cell state
And the cell state Ct-1 at the previous moment, the output is the cell state Ct at the current moment, and the calculation formula is as follows: />
Calculating hidden layer states of an output gate and the current moment: the input is a hidden layer state ht-1 at the previous moment, the output is an output gate ot, the output is a value ot of the output gate and a hidden layer state ht at the current moment, and the calculation formula is as follows: o t ＝σ(W o [h t-1 ,x t ]+b o )，h t ＝o t *tanh(C t )。
The BilSTM respectively adopts forward LSTM and backward LSTM for each word sequence, then output at the same moment is spliced, the output after splicing simultaneously contains forward and backward information, and the calculation method of the output at the moment t comprises the following steps:
where T denotes the time series length and ht (T =1,2, \8230;, T) is the final output.
Step 4-2: and (2) inputting the output of the step (4-1) into an Attention layer, wherein the Attention layer can fuse the output of the BilSt at each moment, a weighted summation mode is adopted during fusion, the weight value is determined by the importance of the word represented by the hidden vector at the moment, and the word with higher importance in the sentence has a higher weight value, so that the information of the more important word can be more concerned when the hidden information of each word in the sentence is fused, the key word expressing the emotional tendency can be concerned, and the accuracy of emotion analysis is improved.
The specific algorithm formula and analysis are as follows:
u t ＝tanh(w s h t +b s )
wherein h is t Is the output of BilSTM at time t, α t Representing a weight value. W is a group of s And b s Respectively weight matrix and offset vector, u s The training method is a parameter vector which comprises a large number of parameters for training, and the parameters can enable the neural network to gradually learn which words are more important, so that higher weight values can be calculated. V is a vector output by the Attention layer and contains deep hidden information of the whole sentence.
Step 4-3: and (3) inputting the output of the step (4-2) into a full connection layer, mapping the output into a vector with the size of 3, and then performing softmax activation on the vector to obtain the final output, wherein the output is predicted probability distribution, namely the predicted probability that the emotional tendency of the input sentence is positive, neutral and negative. And training the whole neural network model for subsequent emotion analysis.
Step 5, adopting an LDA topic model to perform topic extraction on the preprocessed data set to obtain a topic-attribute word; short sentences containing attribute words in the data set are screened, and corresponding topics are marked; the clustered attribute words are all words describing the same theme, and the short sentence with the attribute words is the theme of the comment;
the method specifically comprises the following steps:
step 5-1: performing word segmentation on the data set preprocessed in the step 2 by using a jieba word segmentation tool, inputting a processed text into an LDA topic model, setting the number of topics generated after clustering to be 5, finally obtaining 5 topics, further obtaining attribute words under each topic, wherein the clustered attribute words are all used for describing the same topic, selecting 10 attribute words with the highest frequency under each topic, deducing a proper topic word according to the 10 attribute words, and finally obtaining 5 topic words, wherein 10 attribute words related to the topic words are obtained under each topic word; wherein, the first and the second end of the pipe are connected with each other,
the LDA topic model is mainly used for deducing topic distribution of a document and can be used for identifying potential topic information in a text. The LDA model recognizes that topics can be represented by a distribution of words and articles can be represented by a distribution of topics. According to the LDA model, when an article is required to be generated, the distribution of topics and vocabularies needs to be determined, the distribution of documents and topics is determined, then a topic is randomly generated according to the distribution of the documents and the topics, a word is randomly generated according to the topic through the distribution of the topics and the vocabularies, and the process of the generated word is repeated until a complete document is generated. The distribution of the document and the subject is the subject distribution of the document which is obtained from the Dirichlet distribution alpha, and the distribution of the subject and the vocabulary is the vocabulary distribution which is obtained from the Dirichlet distribution beta and corresponds to the subject.
Step 5-2: and (5) screening short sentences with the attribute words according to the attribute words obtained in the step (5-1), if a plurality of attribute words appear in one comment, dividing the comment to obtain a plurality of short sentences with the attribute words, and labeling the short sentences with corresponding topics according to the attribute words.
Step 6, inputting the short sentence sets marked with the themes into a trained neural network model for fine-grained emotion analysis to obtain the emotion tendency of each theme; wherein the content of the first and second substances,
for each comment, the sentiment tendency of a plurality of subjects described by the comment can be obtained through the short sentence obtained after segmentation; and the emotional tendency of each theme in the whole short sentence set is counted, so that the evaluation condition of each theme and aspect of the book can be obtained, and the comprehensive and systematic evaluation of the book can be obtained.
The invention also provides a text fine-grained emotion analysis system, which comprises the following components:
an obtaining module, configured to implement step 1;
a preprocessing module for implementing the step 2;
a word vectorization module for implementing the step 3;
a training module for implementing the step 4;
a theme extraction module for implementing the step 5;
and the analysis module is used for realizing the step 6.
The present invention also provides a computer readable storage medium having stored thereon executable instructions that, when executed by a processor, cause the processor to perform the above-described text fine-grained emotion analysis method.
The present invention also provides a computing device comprising: one or more memories storing executable instructions; and one or more processors executing executable instructions to implement the text fine-grained sentiment analysis method.
The invention has the advantages that:
according to the invention, word vectors containing deep semantic information are obtained through BERT word vectorization, then the word vectorized text is input into a BilSTM + attention model, deep characteristic information can be further extracted, so that the prediction effect of emotion tendency is effectively improved, topics described by the text can be obtained through topic clustering by an LDA topic model, so that emotion analysis can be performed on the granularity of the topic level, and finally the accuracy of text fine-grained emotion analysis is improved to a certain extent by combining with a Bert + BilSTM front-edge deep learning model.
The above is only a preferred embodiment of the present invention, and is not intended to limit the present invention, and various modifications and changes will occur to those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (9)
1. A text fine-grained emotion analysis method is characterized by comprising the following steps:
obtaining a comment text data set;
preprocessing the comment text data set, wherein the preprocessing comprises data cleaning and data labeling;
performing word vectorization on the comment text in the preprocessed data set by adopting a BERT model to obtain a comment text word vector;
inputting the comment text word vector into a BilSTM + Attention model for coarse-grained emotion analysis, and training to obtain a neural network model for emotion analysis;
adopting an LDA topic model to perform topic extraction on the preprocessed data set to obtain a topic-attribute word; short sentences containing attribute words in the data set are screened, and corresponding topics are marked;
and inputting the short sentence sets marked with the themes into the trained neural network model for fine-grained emotional analysis to obtain the emotional tendency of each theme.
2. The text fine-grained emotion analysis method of claim 1, wherein a web crawler technology is used to obtain the comment text data set.
3. A method for fine-grained sentiment analysis of text according to claim 1, wherein the data cleansing includes removing irrelevant text and repeated comment text, the irrelevant text including but not limited to acronyms, emoticons, repeated punctuation and nonsense sentences;
the data labeling method comprises but is not limited to labeling the comments by using the rating information of the comments, wherein the scores of 0-2 are labeled as negative, the scores of 3 are labeled as neutral, and the scores of 4-5 are labeled as positive.
4. The method for analyzing fine-grained emotion of text according to claim 1, wherein said comment text word vector is inputted into a BilSTM + Attention model for coarse-grained emotion analysis, and a neural network model for emotion analysis is obtained by training; the method comprises the following steps:
extracting implicit information of each word in the comment text sentence by using the BilSTM;
carrying out weighted fusion on the hidden information of each word by adopting an attention mechanism to obtain the hidden information of the whole sentence;
implicit information of the whole sentence is output through a full connection layer and a softmax activation function, emotional tendency of the whole sentence is predicted, and finally a neural network model for emotion analysis is obtained through training.
5. The text fine-grained emotion analysis method of claim 1, wherein the subject extraction is performed on the preprocessed data set by using an LDA subject model to obtain a subject-attribute word; short sentences containing attribute words in the data set are screened, and corresponding topics are marked; the method comprises the following steps:
performing word segmentation on the preprocessed data set;
inputting the text after word segmentation into an LDA topic model, and extracting topics according to the number of preset topics to finally obtain N topics and M attribute words related to the N topics;
and screening short sentences with the attribute words according to the obtained attribute words, and labeling the short sentences with corresponding topics according to the attribute words.
6. The method for analyzing text fine-grained emotion according to claim 5, wherein when a plurality of attribute words are screened out from a sentence of comments, the comments need to be segmented to obtain a plurality of short sentences with the attribute words.
7. A text fine-grained emotion analysis system for implementing the text fine-grained emotion analysis method defined in any one of claims 1 to 6, comprising:
the obtaining module is used for obtaining a comment text data set;
the preprocessing module is used for preprocessing the comment text data set, and the preprocessing comprises data cleaning and data labeling;
the word vectorization module is used for carrying out word vectorization on the comment text in the preprocessed data set by adopting a BERT model to obtain a comment text word vector;
the training module is used for inputting the comment text word vectors into a BilSTM + Attention model for coarse-grained emotion analysis, and training to obtain a neural network model for emotion analysis;
the theme extraction module is used for extracting themes of the preprocessed data set by adopting an LDA theme model to obtain theme-attribute words; short sentences containing attribute words in the data set are screened, and corresponding topics are marked;
and the analysis module is used for inputting the short sentence sets marked with the themes into the trained neural network model for fine-grained sentiment analysis to obtain the sentiment tendency of each theme.
8. A computer-readable storage medium having stored thereon executable instructions that, when executed by a processor, cause the processor to perform a method of fine-grained sentiment analysis of text according to any one of claims 1-6.
9. A computing device, comprising: one or more memories storing executable instructions; one or more processors executing the executable instructions to implement the method of fine grained sentiment analysis of text according to any one of claims 1-6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211441031.1A CN115906824A (en) | 2022-11-17 | 2022-11-17 | Text fine-grained emotion analysis method, system, medium and computing equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211441031.1A CN115906824A (en) | 2022-11-17 | 2022-11-17 | Text fine-grained emotion analysis method, system, medium and computing equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115906824A true CN115906824A (en) | 2023-04-04 |
Family
ID=86487421
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211441031.1A Pending CN115906824A (en) | 2022-11-17 | 2022-11-17 | Text fine-grained emotion analysis method, system, medium and computing equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115906824A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116432107A (en) * | 2023-05-30 | 2023-07-14 | 江苏大学 | Fine granularity emotion classification method |
-
2022
- 2022-11-17 CN CN202211441031.1A patent/CN115906824A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116432107A (en) * | 2023-05-30 | 2023-07-14 | 江苏大学 | Fine granularity emotion classification method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112001185B (en) | Emotion classification method combining Chinese syntax and graph convolution neural network | |
| CN106776581B (en) | Subjective text emotion analysis method based on deep learning | |
| CN112001187B (en) | Emotion classification system based on Chinese syntax and graph convolution neural network | |
| CN112001186A (en) | Emotion classification method using graph convolution neural network and Chinese syntax | |
| Chang et al. | Research on detection methods based on Doc2vec abnormal comments | |
| CN112256866B (en) | Text fine-grained emotion analysis algorithm based on deep learning | |
| Bsir et al. | Enhancing deep learning gender identification with gated recurrent units architecture in social text | |
| CN111339260A (en) | BERT and QA thought-based fine-grained emotion analysis method | |
| Azim et al. | Text to emotion extraction using supervised machine learning techniques | |
| CN112395417A (en) | Network public opinion evolution simulation method and system based on deep learning | |
| CN112434164B (en) | Network public opinion analysis method and system taking topic discovery and emotion analysis into consideration | |
| Kandhro et al. | Performance analysis of hyperparameters on a sentiment analysis model | |
| Hasan et al. | Sentiment analysis using out of core learning | |
| Jayasudha et al. | A survey on sentimental analysis of student reviews using natural language processing (NLP) and text mining | |
| Fouadi et al. | Applications of deep learning in Arabic sentiment analysis: research perspective | |
| CN115906824A (en) | Text fine-grained emotion analysis method, system, medium and computing equipment | |
| Kadagadkai et al. | Summarization tool for multimedia data | |
| CN116108840A (en) | Text fine granularity emotion analysis method, system, medium and computing device | |
| CN111414755A (en) | Network emotion analysis method based on fine-grained emotion dictionary | |
| Basri et al. | A deep learning based sentiment analysis on bang-lish disclosure | |
| Zim et al. | Exploring Word2Vec embedding for sentiment analysis of Bangla raw and romanized text | |
| CN114443846A (en) | Classification method and device based on multi-level text abnormal composition and electronic equipment | |
| CN114817533A (en) | Bullet screen emotion analysis method based on time characteristics | |
| CN115129807A (en) | Fine-grained classification method and system for social media topic comments based on self-attention | |
| Purba et al. | Document level emotion detection from bangla text using machine learning techniques |
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 |