CN116432107A - Fine granularity emotion classification method - Google Patents
Fine granularity emotion classification method Download PDFInfo
- Publication number
- CN116432107A CN116432107A CN202310627129.4A CN202310627129A CN116432107A CN 116432107 A CN116432107 A CN 116432107A CN 202310627129 A CN202310627129 A CN 202310627129A CN 116432107 A CN116432107 A CN 116432107A
- Authority
- CN
- China
- Prior art keywords
- emotion
- neural network
- information
- emotion classification
- inputting
- 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
- 230000008451 emotion Effects 0.000 title claims abstract description 52
- 238000000034 method Methods 0.000 title claims abstract description 18
- 239000013598 vector Substances 0.000 claims abstract description 25
- 238000013528 artificial neural network Methods 0.000 claims abstract description 15
- 238000004458 analytical method Methods 0.000 claims abstract description 14
- 238000012549 training Methods 0.000 claims abstract description 13
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 9
- 238000012545 processing Methods 0.000 claims abstract description 7
- 238000007781 pre-processing Methods 0.000 claims abstract description 6
- 238000000926 separation method Methods 0.000 claims description 6
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 claims description 3
- 238000004140 cleaning Methods 0.000 claims description 3
- 230000002708 enhancing effect Effects 0.000 claims description 3
- 238000005457 optimization Methods 0.000 claims description 3
- 235000019580 granularity Nutrition 0.000 description 6
- 238000010801 machine learning Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 230000011218 segmentation Effects 0.000 description 2
- 238000010276 construction Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000002996 emotional effect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003058 natural language processing Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000006467 substitution reaction 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
-
- 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
- G06F40/30—Semantic analysis
-
- 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
- G06N3/044—Recurrent networks, e.g. Hopfield networks
- G06N3/0442—Recurrent networks, e.g. Hopfield networks characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU]
-
- 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
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- 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
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (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)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Machine Translation (AREA)
Abstract
The invention provides a fine granularity emotion classification method, which is characterized by comprising the following steps of: s1, acquiring sentences to be analyzed and corresponding aspect words thereof, and preprocessing; s2, splicing and inputting the sentences obtained in the S1 and the corresponding aspect words thereof into a SKEP pre-training model for processing to obtain semantic information after each word vector and the context information are integrated; s3, extracting features of the semantic information obtained in the step S2 by adopting a multi-layer neural network to obtain deeper information; and S4, inputting the deeper information extracted in the S3 into an output layer to perform fine-grained emotion prediction, and obtaining an analysis result of an emotion analysis method. The invention adopts the SKEP-BiLSTM-attention-fused threshold convolutional neural network model, enhances the association of aspect words and comment sentences, introduces BiLSTM and threshold convolutional neural network, and can deeply extract emotion characteristics.
Description
Technical Field
The invention belongs to the field of text emotion analysis of natural language processing, and particularly relates to a fine granularity emotion classification method.
Background
With the advent of the information age, the internet was filled with a wide variety of text content, which contained rich information, and especially in the emerging service industry, from thousands of comment information, it was possible to show whether merchants had good praise and thereby judge their quality of service. How to extract emotion tendencies of various granularities from a plurality of comment texts is a practical problem worthy of research.
Fine-grained emotion classification is an important subtask of emotion classification, which is a deeper analysis of subjective text. It is not only context-dependent, but also related to given aspect information, and its task is to analyze deeply a plurality of comment objects appearing in the text, separating the emotional polarities of the plurality of objects. The fine granularity emotion analysis can identify emotion tendencies of each given aspect word in one comment, and thus a more accurate and comprehensive emotion analysis conclusion is obtained, and information loss is avoided.
The existing emotion analysis methods can be divided into emotion dictionary-based emotion analysis methods, traditional machine learning-based emotion analysis methods and neural network deep learning-based emotion analysis methods. The emotion analysis method based on the emotion dictionary in the early stage mainly relies on manual construction of the emotion dictionary, has poor effect on network new words, and needs to expand the dictionary for use. Based on a machine learning method: the statistical machine learning algorithm is used to extract features and output emotion results, but the context information of the context text cannot be fully utilized.
Therefore, a new fine-granularity emotion classification method capable of improving the accuracy of the pushout emotion classification is needed.
Disclosure of Invention
In order to solve the problems in the prior art, the fine-grained emotion classification method provided by the invention comprises the following steps:
s1, acquiring sentences to be analyzed and corresponding aspect words thereof, and preprocessing;
s2, splicing and inputting the preprocessed sentences and the corresponding aspect words thereof into a SKEP pre-training model for processing to obtain semantic information after integrating each word vector with the context information;
s3, extracting features of the semantic information obtained in the step S2 by adopting a multi-layer neural network to obtain deeper information;
and S4, inputting the deeper information extracted in the S3 into an output layer to perform fine-grained emotion prediction, and obtaining an emotion classification result.
Further, the pretreatment specifically comprises:
carrying out escape processing on the sentence to be analyzed and the emoji expression contained in the corresponding aspect word, converting the sentence to be analyzed into text description, and enhancing the emotion characteristics of the original sentence; carrying out data cleaning on useless information equivalent to stop words; and word segmentation is performed on the sentences.
Further, the S2 specifically is:
splicing the preprocessed sentences and the corresponding aspect words into an input text sequence in the form of a text start symbol, a sentence to be analyzed, a text separation and end symbol, an aspect word, a text separation and end symbol; and inputting the text sequence into the SKEP pre-training model to obtain semantic information after word vector representation and context information are integrated.
Further, the step S3 specifically includes:
inputting the semantic information extracted by the SKEP pre-training model into a multi-layer neural network, outputting a forward vector and a backward vector at the position of each word after the multi-layer neural network receives the semantic information, and then splicing the two vectors corresponding to each time step to obtain a final output vector; and applying an Attention mechanism to the context vectors to obtain semantic vectors of the whole text sequence, and obtaining deeper information through a threshold convolutional neural network.
Further, the multi-layer neural network is composed of a threshold convolutional neural network and BiLSTM.
Further, the S4 specifically is:
inputting the deeper information obtained in the step S3 into an output layer, wherein the output layer comprises a full-connection layer and softmax, and p outputs are obtained, wherein p represents different emotion polarity numbers contained in emotion analysis tasks; and finally, fine granularity emotion classification is completed.
Furthermore, an Adam optimizer is adopted to optimize the SKEP pre-training model, and cross entropy is used as a loss function in the optimization process.
The invention has the technical effects that:
the invention adopts the SKEP-BiLSTM-attention-fused threshold convolutional neural network model, enhances the association of aspect words and comment sentences, introduces BiLSTM and threshold convolutional neural network, and can deeply extract emotion characteristics.
Drawings
The accompanying drawings illustrate various embodiments by way of example in general and not by way of limitation, and together with the description and claims serve to explain the inventive embodiments. Wherever possible, the same reference numbers will be used throughout the drawings to refer to the same or like parts. Such embodiments are illustrative and not intended to be exhaustive or exclusive of the present apparatus or method.
FIG. 1 shows a schematic diagram of the steps of the present invention;
FIG. 2 shows a schematic flow chart of the present invention;
fig. 3 shows a schematic architecture of the present invention.
Detailed Description
It should be noted that, in the case of no conflict, the embodiments and features in the embodiments may be combined with each other. The present application will be described in detail below with reference to the accompanying drawings in conjunction with embodiments.
As shown in figure 1 of the drawings,
preprocessing data;
s1, acquiring sentences to be analyzed and corresponding aspect words thereof;
s2, splicing and inputting the sentence acquired in the step S1 after preprocessing and the corresponding aspect words thereof into a SKEP pre-training model for processing to acquire semantic information after integrating each word vector with the context information;
s3, extracting features of the semantic information obtained in the step S2 by adopting a multi-layer neural network to obtain deeper information;
and S4, inputting the deeper information extracted in the step S3 into a full-connection layer for fine-grained emotion prediction, and obtaining an emotion classification result.
The pretreatment is specifically as follows:
carrying out escape processing on the sentence to be analyzed and the emoji expression contained in the corresponding aspect word, converting the sentence to be analyzed into text description, and enhancing the emotion characteristics of the original sentence; carrying out data cleaning on useless information equivalent to stop words; and word segmentation is performed on the sentences.
The step S2 is specifically as follows:
splicing the preprocessed sentences and the corresponding aspect words into an input text sequence in the form of a text start symbol, sentences to be analyzed, text separation and end symbol, aspect words, text separation and end symbol; and inputting the text sequence into the SKEP pre-training model to obtain semantic information after word vector representation and context information are integrated.
The step S3 is specifically as follows:
inputting the semantic information extracted by the SKEP pre-training model into a multi-layer neural network, outputting a forward vector and a backward vector at the position of each word after the multi-layer neural network receives the semantic information, and then splicing the two vectors corresponding to each time step to obtain a final output vector; and applying an Attention mechanism to the context vectors to obtain semantic vectors of the whole text sequence, and obtaining deeper information through a threshold convolutional neural network.
The multi-layer neural network is composed of a threshold convolution neural network and BiLSTM.
The step S4 specifically comprises the following steps:
inputting the deeper information obtained in the step S3 into an output layer, wherein the output layer comprises a full-connection layer and softmax, and p outputs are obtained, wherein p represents different emotion polarity numbers contained in emotion analysis tasks; and finally, fine granularity emotion classification is completed.
And optimizing the SKEP pre-training model by adopting an Adam optimizer, wherein cross entropy is used as a loss function in the optimization process.
The foregoing is only a preferred embodiment of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art, who is within the scope of the present invention, should make equivalent substitutions or modifications according to the technical solution of the present invention and the inventive concept thereof, and should be covered by the scope of the present invention.
Claims (7)
1. A fine granularity emotion classification method, the method comprising:
s1, acquiring sentences to be analyzed and corresponding aspect words thereof, and preprocessing;
s2, splicing and inputting the preprocessed sentences and the corresponding aspect words thereof into a SKEP pre-training model for processing to obtain semantic information after integrating each word vector with the context information;
s3, extracting features of the semantic information obtained in the step S2 by adopting a multi-layer neural network to obtain deeper information;
and S4, inputting the deeper information extracted in the S3 into an output layer to perform fine-grained emotion prediction, and obtaining an emotion classification result.
2. The fine-grained emotion classification method according to claim 1, characterized in that the preprocessing is specifically:
carrying out escape processing on the sentence to be analyzed and the emoji expression contained in the corresponding aspect word, converting the sentence to be analyzed into text description, and enhancing the emotion characteristics of the original sentence; and cleaning data of useless information equivalent to stop words and separating words from sentences.
3. The fine granularity emotion classification method according to claim 1, wherein S2 is specifically:
splicing the processed sentences and the corresponding aspect words into an input text sequence in the form of a text start symbol, a sentence to be analyzed, a text separation and end symbol, an aspect word, a text separation and end symbol; and inputting the text sequence into the SKEP pre-training model to obtain semantic information after word vector representation and context information are integrated.
4. The fine granularity emotion classification method according to claim 1, wherein S3 is specifically:
inputting the semantic information extracted by the SKEP pre-training model into a multi-layer neural network, outputting a forward vector and a backward vector at the position of each word after the multi-layer neural network receives the semantic information, and then splicing the two vectors corresponding to each time step to obtain a final output vector; and applying an Attention mechanism to the context vectors to obtain semantic vectors of the whole text sequence, and obtaining deeper information through a threshold convolutional neural network.
5. The fine granularity emotion classification method of claim 4, wherein the multi-layer neural network is composed of a threshold convolutional neural network and a BiLSTM.
6. The fine granularity emotion classification method according to claim 1, wherein S4 is specifically:
and (3) inputting the deeper information obtained in the step (S3) into an output layer to obtain p outputs, wherein p represents different emotion polarity numbers contained in emotion analysis tasks, and finally finishing fine-granularity emotion classification.
7. The fine granularity emotion classification method of claim 1, wherein the SKEP pre-training model is optimized using an Adam optimizer, and cross entropy is used as a loss function in the optimization process.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310627129.4A CN116432107A (en) | 2023-05-30 | 2023-05-30 | Fine granularity emotion classification method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310627129.4A CN116432107A (en) | 2023-05-30 | 2023-05-30 | Fine granularity emotion classification method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116432107A true CN116432107A (en) | 2023-07-14 |
Family
ID=87081706
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310627129.4A Pending CN116432107A (en) | 2023-05-30 | 2023-05-30 | Fine granularity emotion classification method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116432107A (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109460553A (en) * | 2018-11-05 | 2019-03-12 | 中山大学 | A kind of machine reading understanding method based on thresholding convolutional neural networks |
| CN111881249A (en) * | 2020-06-08 | 2020-11-03 | 江苏大学 | Method for judging text emotion tendentiousness based on recurrent neural network |
| CN115292485A (en) * | 2022-06-14 | 2022-11-04 | 中国民用航空飞行学院 | Emotion classification method based on SKEP-ERNIE-BiGRU network model structure |
| CN115730606A (en) * | 2022-12-07 | 2023-03-03 | 辽宁大学 | BERT model and attention concentration network-based fine-grained emotion analysis method |
| CN115906824A (en) * | 2022-11-17 | 2023-04-04 | 北京工业大学 | Text fine-grained emotion analysis method, system, medium and computing equipment |
-
2023
- 2023-05-30 CN CN202310627129.4A patent/CN116432107A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109460553A (en) * | 2018-11-05 | 2019-03-12 | 中山大学 | A kind of machine reading understanding method based on thresholding convolutional neural networks |
| CN111881249A (en) * | 2020-06-08 | 2020-11-03 | 江苏大学 | Method for judging text emotion tendentiousness based on recurrent neural network |
| CN115292485A (en) * | 2022-06-14 | 2022-11-04 | 中国民用航空飞行学院 | Emotion classification method based on SKEP-ERNIE-BiGRU network model structure |
| CN115906824A (en) * | 2022-11-17 | 2023-04-04 | 北京工业大学 | Text fine-grained emotion analysis method, system, medium and computing equipment |
| CN115730606A (en) * | 2022-12-07 | 2023-03-03 | 辽宁大学 | BERT model and attention concentration network-based fine-grained emotion analysis method |
Non-Patent Citations (2)
| Title |
|---|
| 马费成，赵一鸣: "《大数据环境下的知识组织与服务创新》", 30 November 2021, 武汉：武汉大学出版社, pages: 321 - 328 * |
| 黄晨: "基于深度学习的网络评论情感分析技术研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》, no. 06, pages 138 - 1266 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110532554B (en) | Chinese abstract generation method, system and storage medium | |
| CN106886580B (en) | Image emotion polarity analysis method based on deep learning | |
| CN110853649A (en) | Label extraction method, system, device and medium based on intelligent voice technology | |
| CN112100388A (en) | Method for analyzing emotional polarity of long text news public sentiment | |
| CN113657115B (en) | Multi-mode Mongolian emotion analysis method based on ironic recognition and fine granularity feature fusion | |
| CN111339305A (en) | Text classification method and device, electronic equipment and storage medium | |
| CN111339772B (en) | Russian text emotion analysis method, electronic device and storage medium | |
| CN112861524A (en) | Deep learning-based multilevel Chinese fine-grained emotion analysis method | |
| CN112836053A (en) | Man-machine conversation emotion analysis method and system for industrial field | |
| KR20190123093A (en) | System and Method for Automated Emoticon Recommendation based on Emotion Ontology | |
| CN112579762A (en) | Dialogue emotion analysis method based on semantics, emotion inertia and emotion commonality | |
| Munandar et al. | A multi domains short message sentiment classification using hybrid neural network architecture | |
| CN111460100A (en) | Criminal legal document and criminal name recommendation method and system | |
| CN111666385A (en) | Customer service question-answering system based on deep learning and implementation method | |
| CN116432107A (en) | Fine granularity emotion classification method | |
| CN112464664B (en) | Multi-model fusion Chinese vocabulary repeated description extraction method | |
| CN115827831A (en) | Intention recognition model training method and device | |
| Saraswat et al. | Sentiment Analysis of Audio Files Using Machine Learning and Textual Classification of Audio Data | |
| CN113988047A (en) | Corpus screening method and apparatus | |
| He | Recent works for sentiment analysis using machine learning and lexicon based approaches | |
| CN118095218B (en) | Insurance industry dialogue data enhancement method based on large model | |
| CN115563972B (en) | Training method of structured six-linked emotion analysis model | |
| CN114860937B (en) | Sentence classification method and system based on Chinese bionic literature abstract | |
| CN113658587B (en) | Intelligent voice recognition method and system with high recognition rate based on deep learning | |
| CN115658908B (en) | Five-personality perception method and system based on conversation interaction process |
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 |