CN112070139A - Text classification method based on BERT and improved LSTM - Google Patents
Text classification method based on BERT and improved LSTM Download PDFInfo
- Publication number
- CN112070139A CN112070139A CN202010898906.5A CN202010898906A CN112070139A CN 112070139 A CN112070139 A CN 112070139A CN 202010898906 A CN202010898906 A CN 202010898906A CN 112070139 A CN112070139 A CN 112070139A
- Authority
- CN
- China
- Prior art keywords
- text
- bert
- vector
- lstm
- gate
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 17
- 239000013598 vector Substances 0.000 claims abstract description 61
- 238000007781 pre-processing Methods 0.000 claims abstract description 6
- 230000009467 reduction Effects 0.000 claims abstract description 4
- 238000012545 processing Methods 0.000 claims abstract description 3
- 238000001914 filtration Methods 0.000 claims description 6
- 230000002457 bidirectional effect Effects 0.000 claims description 3
- 238000012217 deletion Methods 0.000 claims description 3
- 230000037430 deletion Effects 0.000 claims description 3
- 230000004913 activation Effects 0.000 claims description 2
- 230000001413 cellular effect Effects 0.000 claims description 2
- 239000011159 matrix material Substances 0.000 claims description 2
- 230000011218 segmentation Effects 0.000 claims description 2
- 238000013145 classification model Methods 0.000 abstract description 11
- 230000000694 effects Effects 0.000 abstract description 6
- 238000000605 extraction Methods 0.000 abstract description 4
- 210000002569 neuron Anatomy 0.000 abstract description 4
- 230000009286 beneficial effect Effects 0.000 abstract description 2
- 238000012549 training Methods 0.000 description 11
- 230000008859 change Effects 0.000 description 4
- 210000004027 cell Anatomy 0.000 description 3
- 238000013527 convolutional neural network Methods 0.000 description 3
- 238000013135 deep learning Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 238000012795 verification Methods 0.000 description 3
- 238000003062 neural network model Methods 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- 239000002699 waste material Substances 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 230000008451 emotion Effects 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000007477 logistic regression Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 230000000306 recurrent effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000012827 research and development Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 238000000844 transformation Methods 0.000 description 1
- 238000010200 validation analysis Methods 0.000 description 1
Images
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
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/205—Parsing
- G06F40/211—Syntactic parsing, e.g. based on context-free grammar [CFG] or unification grammars
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/289—Phrasal analysis, e.g. finite state techniques or chunking
-
- 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
- G06N3/045—Combinations of networks
-
- 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
- G06N3/049—Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
-
- 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
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Probability & Statistics with Applications (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Machine Translation (AREA)
Abstract
The invention belongs to the field of text recognition, and discloses a text classification method based on BERT and improved LSTM, which comprises the following steps: preprocessing input text data; inputting the preprocessed text data into a BERT model for processing to obtain a word vector sequence; carrying out depth coding on the vector sequence by utilizing an improved LSTM network to obtain a characteristic vector; reducing the dimension of the feature vector by using a full connection layer; and classifying the feature vectors subjected to dimension reduction by using a classifier. The invention distinguishes the importance of words in the text through the improved LSTM, thereby improving the learning quality and efficiency of the neuron, and the text classification model has high fitting speed and good classification effect; the BERT of the text classification model captures context information, thereby being convenient for identifying ambiguous words, laying a good foundation for feature extraction and being beneficial to improving the text classification precision.
Description
Technical Field
The invention belongs to the field of text recognition, and particularly relates to a text classification method based on BERT and improved LSTM.
Background
The main application fields of text classification are microblog emotion analysis, user comment mining, information retrieval, classification of news groups, word semantic resolution and the like. Before the 90 s of the 20 th century, automatic text classification mainly adopts a knowledge-based engineering mode, namely, manual classification by professionals, and has the defects of high cost, time waste and labor waste. Since the 90 s, researchers began applying various statistical and machine learning methods to automatic text classification, such as Support Vector Machine (SVM), AdaBoost, naive bayes, KNN, and Logistic regression. In recent years, with the rapid development of deep learning and various neural network models, the text classification method based on deep learning has attracted close attention and research in academia and industry, and the recurrent neural networks LSTM, GRU and the convolutional neural network CNN are widely applied to text classification.
In the current text classification method, input is often a non-dynamic word vector or word vector, the word vector or word vector cannot be changed according to the context of the word vector or word vector, and the information coverage is relatively single; most of the adopted feature extraction models are CNN and RNN models in deep learning, and fine-grained adjustment of different importance levels of input information streams in input dimensions is lacked.
Disclosure of Invention
The invention aims to solve the problems, and provides a text classification method based on BERT and improved LSTM, which adds a contribution gate in the existing LSTM unit to pay attention to the importance of different elements of a vector sequence of a text, improves the learning efficiency of neurons, accelerates the fitting speed of a text classification model, and improves the classification effect of the text classification model.
The technical scheme of the invention is a text classification method based on BERT and improved LSTM, a text classification model comprises the BERT, the improved LSTM and a classifier which are connected in sequence, and the text classification method comprises the following steps:
step 1: preprocessing input text data;
step 2: inputting the preprocessed text data into a BERT model for processing to obtain a word vector sequence;
and step 3: carrying out depth coding on the vector sequence by utilizing an improved LSTM network to obtain a characteristic vector;
and 4, step 4: reducing the dimension of the feature vector by using a full connection layer;
and 5: and classifying the feature vectors subjected to dimension reduction by using a classifier.
Further, in step 1, the preprocessing of the text data includes punctuation filtering, abbreviation filling, space deletion and illegal character filtering.
Further, the step 2 specifically includes:
1) utilizing the trained BERT model to perform word segmentation on the text of the preprocessed text data set T ', and obtaining a word vector set T' ═ { T1″，t2″，...，tn", the text of the text data set is converted into a word vector t" of fixed length1＝{w1，w2，...，wL}; 2) inputting the word vector set T' into a Token embedding layer, a Segment embedding layer and a Position embedding layer in BERT to respectively obtain vector codes V1Sentence coding V2And a position code V3；
3) Will V1、V2、V3Adding the words and the vectors, inputting the words and the vectors into a bidirectional Transformer in BERT to obtain a word vector sequence S ═ S1，s2，...，sn}。
Further, the LSTM unit of the improved LSTM network comprises a contribution gate, a forgetting gate, an input gate and an output gate, wherein the contribution gate is based on the cell state c at the last momentt-1Hidden state ht-1And the input information of the current time to generate and input vector xtAttention vector a with the same dimensiontAttention vector atAs inputs to the forgetting gate, the input gate, and the output gate.
Further, step 5 performs probability classification on the feature vector with the dimension reduced in step 4 by using a softmax classifier, and outputs a probability prediction vector P ═ { P ═ P1，p2，...，pC}，piI 1,2, C denotes the probability that the text belongs to a particular class, C being the total number of classes; p with the maximum probability valueiThe corresponding classification is determined as the category of the text.
Compared with the prior art, the invention has the beneficial effects that:
1) the text classification method of the invention distinguishes the importance of the words in the text through the improved LSTM, thereby improving the learning quality and efficiency of the neurons, and the text classification model has high fitting speed and good classification effect;
2) the BERT of the text classification model captures context information, so that ambiguous words can be conveniently recognized, a good foundation is laid for feature extraction, and the text classification precision is improved;
3) the text classification model combining BERT and improved LSTM has good universality and is suitable for text classification in different technical fields.
Drawings
The invention is further illustrated by the following figures and examples.
Fig. l is a flowchart illustrating a text classification method according to an embodiment of the present invention.
Fig. 2 is a schematic structural diagram of a text classification model according to an embodiment of the present invention.
Fig. 3 is a schematic diagram of an LSTM unit in accordance with an embodiment of the present invention.
FIG. 4 is a schematic diagram showing the comparison of the improved LSTM with the verification accuracies of sRNN, LSTM and GRU.
FIG. 5 is a graphical representation of the improved LSTM vs. the loss of validation for sRNN, LSTM, GRU.
Detailed Description
Embodiments select the chinese dataset THUCNews for classification testing. As shown in FIG. 2, the text classification model of an embodiment includes BERT, modified LSTM, classifiers, which include a fully connected layer and a Softmax layer.
As shown in fig. 1, the text classification method based on BERT and modified LSTM includes the following steps:
step 1: inputting a text data set T, preprocessing sentences in the text, including punctuation mark filtration, abbreviation filling, space deletion and illegal character filtration, determining the length threshold of the sentences by combining the length distribution and mean square deviation of the sentences in the text data set, and forming uniform sentence length to finally obtain a text data set T';
step 2: vectorizing the text data set T' by using the trained BERT model;
segmenting the text in the T' by using the trained BERT model to obtain data T ″ { T ″1″，t2″，...，tn", where each text is converted to a word vector t" of fixed length L1＝{w1，w2，...，wL}；
Putting the text in the T' into a Token Embedding layer, a Segment Embedding layer and a Position Embedding layer in the BERT to respectively obtain vector codes V1Sentence coding V2And a position code V3；
Will V1、V2、V3Adding the words, inputting the words into a bidirectional Transformer in BERT, and outputting a word vector sequence S ═ S corresponding to T ″1，s2，...，snH, where each word vector subsequence siBy the word vector v (w) of the ith textj) Composition, i represents each word in the text;
the BERT model of the examples was referred to the BERT disclosed in the paper "Pre-training of Deep biological transformations for Language Understanding", published by the Google research and development team 2018.
And step 3: the improved LSTM network is used for carrying out feature learning on the word vector sequence S, potential features F are extracted, and the structure of an LSTM unit of the improved LSTM network is shown in figure 3;
and 4, step 4: reducing the dimension of the feature vector F by using a full connection layer;
and 5: performing probability classification on the feature vector of the dimensionality reduction step 4 by using a softmax classifier, and outputting a probability prediction vector P ═ P1，p2，...，pC}，piI 1,2, C denotes the probability that the text belongs to a particular class, C being the total number of classes; and determining the corresponding classification with the maximum probability value as the text category.
The LSTM unit of the improved LSTM network comprises a contribution gate, a forgetting gate, an input gate and an output gate, wherein the contribution gate is based on the cell state c at the last momentt-1Hidden state ht-1And the input information of the current time to generate and input vector xtAttention having the same dimensionForce vector atAttention vector atAnd xtCombining to obtain optimized input vector xt', as inputs to a forgetting gate, an input gate, and an output gate;
at＝σa(Waxt+Uaht-1+Mact-1+ba)
xt′＝(xt+ht-1)οat
forget the door:
ft＝σg(Wfxt′+bf)
an input gate:
it＝σg(Wixt′+bi)
an output gate:
ot＝σg(Woxt′+bo)
cell state:
ct＝ftοct-1+itοσc(Wcxt′+bc)
hidden state:
ht＝otοσh(ct)
wherein h istHidden state at the current time t, ctIs the cellular state at the present time t, Wa、Ua、Ma、Wf、Wi、Wo、WcAre respectively a weight matrix, ba、bf、bi、bo、bcAre respectively deviation terms; sigmaa、σg、σc、σhRespectively, an activation function; o denotes element-by-element dot multiplication operation.
Compared with models such as word2vec, word vectors generated by the word2vec model are fixed, and word polysemous differences are given to the BERT according to context related information to generate more accurate feature representation, so that the model performance is improved.
The improved LSTM gives attention to important elements in input information to neurons in the input dimension aspect because of the capability of paying attention to the important elements, so that the fitting speed of the model is increased, and the training effect of the model is improved.
In order to verify the effectiveness of the improved LSTM, the Chinese data set THUCNews is compared with other neural network models sRNN, LSTM and GRU, 10 times, 30 times and 50 times (Epochs) are respectively trained, the model precision (Accuracy) and the model fitting speed (Convergent) are compared, and the experimental results are shown in Table 1.
TABLE 1 precision comparison table for improved LSTM and sRNN, LSTM, GRU models
As can be seen from Table 1, the accuracy of the improved LSTM is optimal when the feature extraction layer selects and uses the text classification model of the improved LSTM for training different times, the accuracy of the improved LSTM is improved by 1.86% relative to the LSTM after 10 times of training, the accuracy of the improved LSTM is improved by 1.14% relative to the LSTM after 30 times of training, and the accuracy of the improved LSTM is improved by 0.94% relative to the LSTM after 50 times of training.
As can be seen from fig. 4 and 5, the verification accuracy curve of the improved LSTM does not change drastically any more when the LSTM is iterated 6 times, and the corresponding verification loss is also a minimum value, the model is close to fitting, and the model achieves the best effect.
Compared with the improved LSTM, the accuracy change range of the sRNN, the GRU and the LSTM is larger when different training times are used. Under different training times, the model training effect of the improved LSTM is similar, the model precision is 92.02% after 10 times of training, the model precision is 92.86% after 50 times of training, and the difference is only 0.8%.
The above description is only for the preferred embodiments of the present invention, but the protection scope of the present invention is not limited thereto, and any person skilled in the art can substitute or change the technical solution of the present invention and the inventive concept within the scope of the present invention, which is disclosed by the present invention, and the equivalent or change thereof belongs to the protection scope of the present invention.
Claims (5)
1. The text classification method based on BERT and improved LSTM is characterized by comprising the following steps:
step 1: preprocessing input text data;
step 2: inputting the preprocessed text data into a BERT model for processing to obtain a word vector sequence;
and step 3: carrying out depth coding on the vector sequence by utilizing an improved LSTM network to obtain a characteristic vector;
and 4, step 4: reducing the dimension of the feature vector by using a full connection layer;
and 5: and classifying the feature vectors subjected to dimension reduction by using a classifier.
2. The BERT and LSTM-based text classification method of claim 1, wherein the preprocessing of the text data in step 1 comprises punctuation filtering, abbreviation padding, space deletion and illegal character filtering.
3. The BERT and LSTM-based text classification method according to claim 1, wherein the step 2 specifically comprises:
1) utilizing a trained BERT model to perform word segmentation on the text of the preprocessed text data set T', and obtaining a word vector set T ═ T1”,t2”,...,tn"}, the text of the text data set is converted into a word vector t of fixed length1”＝{w1,w2,...,wL}；
2) Inputting the word vector set T' into a Token embedding layer, a Segment embedding layer and a Position embedding layer in BERT to respectively obtain vector codes V1Sentence coding V2And a position code V3；
3) Will V1、V2、V3Adding the words and the vectors, inputting the words and the vectors into a bidirectional Transformer in BERT to obtain a word vector sequence S ═ S1,s2,...,sn}。
4. According to claim 1The text classification method based on BERT and improved LSTM is characterized in that LSTM units of the improved LSTM network comprise a contribution gate, a forgetting gate, an input gate and an output gate, and the contribution gate is based on the cell state c at the last momentt-1Hidden state ht-1And the input information of the current time to generate and input vector xtAttention vector a with the same dimensiontAttention vector atAnd xtCombining to obtain optimized input vector xt', as inputs to a forgetting gate, an input gate, and an output gate;
at＝σa(Waxt+Uaht-1+Mact-1+ba)
forget the door:
ft＝σg(Wfxt'+bf)
an input gate:
it＝σg(Wixt'+bi)
an output gate:
ot＝σg(Woxt'+bo)
cell state:
hidden state:
wherein h istHidden state at the current time t, ctIs the cellular state at the present time t, Wa、Ua、Ma、Wf、Wi、Wo、WcAre respectively a weight matrix, ba、bf、bi、bo、bcAre respectively deviation terms; sigmaa、σg、σc、σhRespectively, an activation function;
representing an element-by-element dot product operation.
5. The BERT and LSTM based text classification method according to any of claims 1-4, wherein step 5 uses softmax layer to perform probability classification on the step 4 reduced-dimension feature vector, and outputs a probability prediction vector P ═ { P ═ P { (m }1,p2,...,pC}，piI 1,2, C denotes the probability that the text belongs to a particular class, C being the total number of classes; p with the maximum probability valueiThe corresponding classification is determined as the category of the text.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010898906.5A CN112070139B (en) | 2020-08-31 | 2020-08-31 | Text classification method based on BERT and improved LSTM |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010898906.5A CN112070139B (en) | 2020-08-31 | 2020-08-31 | Text classification method based on BERT and improved LSTM |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112070139A true CN112070139A (en) | 2020-12-11 |
| CN112070139B CN112070139B (en) | 2023-12-26 |
Family
ID=73665222
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010898906.5A Active CN112070139B (en) | 2020-08-31 | 2020-08-31 | Text classification method based on BERT and improved LSTM |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112070139B (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112818123A (en) * | 2021-02-08 | 2021-05-18 | 河北工程大学 | Emotion classification method for text |
| CN113177120A (en) * | 2021-05-11 | 2021-07-27 | 中国人民解放军国防科技大学 | Method for quickly editing information based on Chinese text classification |
| CN113821637A (en) * | 2021-09-07 | 2021-12-21 | 北京微播易科技股份有限公司 | Long text classification method and device, computer equipment and readable storage medium |
| CN114240495A (en) * | 2021-12-16 | 2022-03-25 | 成都新潮传媒集团有限公司 | Method and device for predicting business opportunity conversion probability and computer readable storage medium |
| CN115048447A (en) * | 2022-06-27 | 2022-09-13 | 华中科技大学 | Database natural language interface system based on intelligent semantic completion |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106599933A (en) * | 2016-12-26 | 2017-04-26 | 哈尔滨工业大学 | Text emotion classification method based on the joint deep learning model |
| CN109101584A (en) * | 2018-07-23 | 2018-12-28 | 湖南大学 | A kind of sentence classification improved method combining deep learning with mathematical analysis |
| CN109344244A (en) * | 2018-10-29 | 2019-02-15 | 山东大学 | A kind of the neural network relationship classification method and its realization system of fusion discrimination information |
| CN109918491A (en) * | 2019-03-12 | 2019-06-21 | 焦点科技股份有限公司 | A kind of intelligent customer service question matching method of knowledge based library self study |
| CN109992783A (en) * | 2019-04-03 | 2019-07-09 | 同济大学 | Chinese term vector modeling method |
| CN109992648A (en) * | 2019-04-10 | 2019-07-09 | 北京神州泰岳软件股份有限公司 | The word-based depth text matching technique and device for migrating study |
| CN110362817A (en) * | 2019-06-04 | 2019-10-22 | 中国科学院信息工程研究所 | A kind of viewpoint proneness analysis method and system towards product attribute |
| CN110717334A (en) * | 2019-09-10 | 2020-01-21 | 上海理工大学 | Text emotion analysis method based on BERT model and double-channel attention |
| CN110781306A (en) * | 2019-10-31 | 2020-02-11 | 山东师范大学 | English text aspect layer emotion classification method and system |
-
2020
- 2020-08-31 CN CN202010898906.5A patent/CN112070139B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106599933A (en) * | 2016-12-26 | 2017-04-26 | 哈尔滨工业大学 | Text emotion classification method based on the joint deep learning model |
| CN109101584A (en) * | 2018-07-23 | 2018-12-28 | 湖南大学 | A kind of sentence classification improved method combining deep learning with mathematical analysis |
| CN109344244A (en) * | 2018-10-29 | 2019-02-15 | 山东大学 | A kind of the neural network relationship classification method and its realization system of fusion discrimination information |
| CN109918491A (en) * | 2019-03-12 | 2019-06-21 | 焦点科技股份有限公司 | A kind of intelligent customer service question matching method of knowledge based library self study |
| CN109992783A (en) * | 2019-04-03 | 2019-07-09 | 同济大学 | Chinese term vector modeling method |
| CN109992648A (en) * | 2019-04-10 | 2019-07-09 | 北京神州泰岳软件股份有限公司 | The word-based depth text matching technique and device for migrating study |
| CN110362817A (en) * | 2019-06-04 | 2019-10-22 | 中国科学院信息工程研究所 | A kind of viewpoint proneness analysis method and system towards product attribute |
| CN110717334A (en) * | 2019-09-10 | 2020-01-21 | 上海理工大学 | Text emotion analysis method based on BERT model and double-channel attention |
| CN110781306A (en) * | 2019-10-31 | 2020-02-11 | 山东师范大学 | English text aspect layer emotion classification method and system |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112818123A (en) * | 2021-02-08 | 2021-05-18 | 河北工程大学 | Emotion classification method for text |
| CN113177120A (en) * | 2021-05-11 | 2021-07-27 | 中国人民解放军国防科技大学 | Method for quickly editing information based on Chinese text classification |
| CN113177120B (en) * | 2021-05-11 | 2024-03-08 | 中国人民解放军国防科技大学 | Quick information reorganizing method based on Chinese text classification |
| CN113821637A (en) * | 2021-09-07 | 2021-12-21 | 北京微播易科技股份有限公司 | Long text classification method and device, computer equipment and readable storage medium |
| CN114240495A (en) * | 2021-12-16 | 2022-03-25 | 成都新潮传媒集团有限公司 | Method and device for predicting business opportunity conversion probability and computer readable storage medium |
| CN115048447A (en) * | 2022-06-27 | 2022-09-13 | 华中科技大学 | Database natural language interface system based on intelligent semantic completion |
| CN115048447B (en) * | 2022-06-27 | 2023-06-16 | 华中科技大学 | Database natural language interface system based on intelligent semantic completion |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112070139B (en) | 2023-12-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110502749B (en) | Text relation extraction method based on double-layer attention mechanism and bidirectional GRU | |
| US11631007B2 (en) | Method and device for text-enhanced knowledge graph joint representation learning | |
| CN110119765B (en) | Keyword extraction method based on Seq2Seq framework | |
| CN109376242B (en) | Text classification method based on cyclic neural network variant and convolutional neural network | |
| CN110059188B (en) | Chinese emotion analysis method based on bidirectional time convolution network | |
| CN111160037B (en) | Fine-grained emotion analysis method supporting cross-language migration | |
| CN112070139B (en) | Text classification method based on BERT and improved LSTM | |
| CN112732916B (en) | BERT-based multi-feature fusion fuzzy text classification system | |
| CN111401061A (en) | Method for identifying news opinion involved in case based on BERT and Bi L STM-Attention | |
| CN111368086A (en) | CNN-BilSTM + attribute model-based sentiment classification method for case-involved news viewpoint sentences | |
| CN112069831B (en) | Method for detecting unreal information based on BERT model and enhanced hybrid neural network | |
| CN110287323B (en) | Target-oriented emotion classification method | |
| CN108319666A (en) | A kind of electric service appraisal procedure based on multi-modal the analysis of public opinion | |
| CN111177376A (en) | Chinese text classification method based on BERT and CNN hierarchical connection | |
| CN107451118A (en) | Sentence-level sensibility classification method based on Weakly supervised deep learning | |
| CN110175221B (en) | Junk short message identification method by combining word vector with machine learning | |
| CN114462420A (en) | False news detection method based on feature fusion model | |
| CN112199503B (en) | Feature-enhanced unbalanced Bi-LSTM-based Chinese text classification method | |
| CN112561718A (en) | Case microblog evaluation object emotion tendency analysis method based on BilSTM weight sharing | |
| CN111353032B (en) | Community question and answer oriented question classification method and system | |
| CN114722835A (en) | Text emotion recognition method based on LDA and BERT fusion improved model | |
| CN114417851A (en) | Emotion analysis method based on keyword weighted information | |
| CN115630156A (en) | Mongolian emotion analysis method and system fusing Prompt and SRU | |
| CN114417872A (en) | Contract text named entity recognition method and system | |
| CN113987120A (en) | Public sentiment emotion classification method based on deep learning |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |