CN109213861B - Traveling evaluation emotion classification method combining At _ GRU neural network and emotion dictionary - Google Patents
Traveling evaluation emotion classification method combining At _ GRU neural network and emotion dictionary Download PDFInfo
- Publication number
- CN109213861B CN109213861B CN201810862476.4A CN201810862476A CN109213861B CN 109213861 B CN109213861 B CN 109213861B CN 201810862476 A CN201810862476 A CN 201810862476A CN 109213861 B CN109213861 B CN 109213861B
- Authority
- CN
- China
- Prior art keywords
- emotion
- word
- dictionary
- evaluation
- tourism
- 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.)
- Active
Links
Images
Classifications
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Machine Translation (AREA)
Abstract
The invention relates to a tourism evaluation emotion classification method combining an At _ GRU neural network and an emotion dictionary, which is used for realizing the evaluation semantic classification of tourism users on the whole tourism according to the evaluation text of tourists on the whole journey of the tourism and comprises the following steps: 1) and an emotional characteristic processing stage: vectorizing the emotional characteristics in the travel comments by constructing a composite type special travel emotional dictionary; 2) a data preprocessing stage: carrying out context vector splicing on the original comment text training word vector, and fusing the spliced vector and the vectorized emotional characteristics to be used as the input of a bidirectional GRU neural network; 3) a bidirectional GRU text semantic classification model stage: and training a bidirectional GRU neural network and classifying the tourism evaluation emotion. Compared with the prior art, the method has the advantages of high precision, consideration of the accuracy of the emotion dictionary and the robustness of machine learning and the like.
Description
Technical Field
The invention relates to the field of natural language processing and deep learning, in particular to a tourism evaluation emotion classification method combining an At _ GRU neural network and an emotion dictionary.
Background
The travel route evaluation is to record the feedback of the tourist on a specific travel route formulated by a certain scenic spot on a travel website, is to most directly express the satisfaction degree or suggestion of the tourist on the travel route, and is a link vertically connecting the tourist and the travel website. Through the evaluation of the travel route, the passenger can elaborate the route for reference of the travel arrangement, accommodation condition, traffic arrangement and the like, and the travel company can also directly listen to the opinions, quickly respond to the details of improving and adjusting the travel route and the like, improve the service and strengthen the satisfaction degree of the passenger. Therefore, the travel route evaluation is rapidly and accurately analyzed and processed in detail, and accurate evaluation grade and classification are obtained, so that the optimization speed of the travel route can be greatly improved, the feedback gap is shortened, the manual analysis cost is reduced, and the service quality of a travel company is effectively improved.
The tourist route evaluation information is important information for recording the feedback of passengers, and is mainly expressed as a natural language paragraph in a short text form. Meanwhile, in recent years, natural language processing technology is in rapid development, and particularly, text semantic analysis is taken as an important research object, namely, text is subjected to structured extraction, analysis and understanding, and is associated from a semantic level, so that text meaning is accurately understood. The semantic analysis method comprises a traditional method and a deep learning method, wherein the deep learning method can extract more effective text features and has higher accuracy compared with the traditional method. A large number of scholars do relevant research at home and abroad, how to bring forward multi-feature combination semantic mining based on decision trees in the world, the Xiamanyu et al adopt an ICTCCLAS word segmentation technology and word frequency statistics to carry out commodity evaluation feature mining, but deep learning training cannot be introduced, so that more accurate features cannot be well extracted, and the Lijie et al adopt a CNN model to carry out short text analysis, but cannot fully utilize context semantic information, so that the problem of inaccurate classification is caused.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provides a tourism evaluation emotion classification method combining an At _ GRU neural network and an emotion dictionary.
The purpose of the invention can be realized by the following technical scheme:
a travel evaluation emotion classification method combining an At _ GRU neural network and an emotion dictionary is used for realizing the evaluation semantic classification of a travel user on the whole travel according to an evaluation text of a tourist on the whole travel, and comprises the following steps:
1) and an emotional characteristic processing stage: vectorizing the emotional characteristics in the travel comments by constructing a composite type special travel emotional dictionary;
2) a data preprocessing stage: carrying out context vector splicing on the original comment text training word vector, and fusing the spliced vector and the vectorized emotional characteristics to be used as the input of a bidirectional GRU neural network;
3) a bidirectional GRU text semantic classification model stage: and training a bidirectional GRU neural network and classifying the tourism evaluation emotion.
The step 1) specifically comprises the following steps:
11) constructing a composite emotion dictionary special for tourism: counting various existing emotion dictionaries, combining words with the same polarity to form a composite emotion dictionary, vectorizing the composite emotion dictionary and the segmented tourism evaluation words respectively, acquiring Euclidean distances between each word in the dictionary and all words of tourism evaluation, selecting the words with similar distances as similar emotion words, and fusing to form the composite special emotion dictionary for tourism;
12) semantic logic rule processing: reading travel comments, dividing the comments into sentences by taking Chinese and English punctuations as sentence marks, and acquiring the sentiment value M (M) of each sentence according to the part of speech of the sentence1,m2,m3,m4,m5)，m1-m5Whether the clauses contain negative words, degree adverbs and emotional words or not is respectively determined, and finally the emotion value of each clause is combined to serve as the emotion polarity of the comment;
13) vectorizing emotional characteristics: and (5) performing dimension increase and vectorization on the emotion polarity of the processed clause.
In the step 11), the existing emotion dictionaries comprise a Qinghua university dictionary, a Hownet emotion dictionary of the Hownet and a simplified Chinese emotion dictionary of Taiwan university.
The step 2) specifically comprises the following steps:
21) word segmentation, removal of stop words and training word vectors: the original comment text information is defined as S and contains a constructed word set W (W)1,w2,...,wn) N is the number of words in the sentence S, the original comment text is participled by adopting an ICTCCLAS tool, stop words are removed, and a Word set W (W) is divided by adopting a Word2vec tool1,w2,...,wn) Performing word vector training, each word wiA word vector form represented as 50 dimensions;
22) combining the stitching context vectors: defining the ith word w in a sentenceiAll the remaining sentence information contained on the left is Cl(wi) And the left sentence information on the right is Cr(wi) Converting the word vector into a word vector and then generating a combined spliced 50-dimensional word vector;
23) and (3) feature vector fusion: and performing quantity product processing and fusion on the spliced 50-dimensional word vector and the vectorized emotional characteristics to form a final 55-dimensional word vector, and taking the final 55-dimensional word vector as the input of the GRU network.
In the step 22), the specific definition expression is as follows:
cl(wi)＝f(W(l)cl(wi-1)+W(sl)e(wi-1))
cr(wi)＝f(W(r)cr(wi+1)+W(sr)e(wi+1))
wherein, W(l)W(r)For converting a hidden layer, i.e. a context, into a matrix for the next hidden layer, W(sl)For the matrix used to combine the semantics of the text relationship left of the current word with the next word, W(sr)For a matrix combining the semantics of the text relationship of the current word to the right of the last word, f is a non-linear activation function, cl(wi-1) Is the i-1 th word wi-1All remaining sentence information contained on the left, e (w)i-1) Is the i-1 th word wi-1The word vector form of e (w)i+1) Is the i +1 th word wi+1Word vector form of cr(wi+1) Is the i +1 th word wi+1All remaining sentence information contained on the right.
The step 3) specifically comprises the following steps:
31) training a bidirectional GRU network: constructing a bidirectional GRU network, loading a 55-dimensional word vector training set into an At _ GRU model from the forward direction and the reverse direction of a sentence respectively, and performing parameter tuning to finish training;
32) for the trained bidirectional GRU network model, new tourist user evaluation is subjected to data preprocessing to form word vectors, the word vectors are loaded into the model for emotion classification, natural language emotion analysis of each user evaluation is realized, emotion polarity classification in three degrees of satisfaction, generality and dissatisfaction is finally shown in the form of 5 dimensions of tour guide service, forced consumption, traffic routes, route arrangement and accommodation and catering, the emotion polarity classification is respectively shown by 1, 0 and-1, and experience feedback of passengers in each tour route to the 5 dimensions is shown.
Compared with the prior art, the invention has the following advantages:
the tourism exclusive dictionary is constructed, a great neural network training corpus is provided, and compared with the original general emotion dictionary, the classification precision can be increased.
And secondly, enhancing context semantic relation through a word vector splicing form, combining emotion polarity characteristics generated based on semantic logic rules, and generating a brand new characteristic vector through a specific fusion formula. The method has the accuracy of the emotion dictionary and the robustness of machine learning.
And thirdly, through bidirectional GRU neural network training, compared with the traditional unidirectional neural network, the semantic information of the sentence can be effectively analyzed in the forward direction and the reverse direction.
Drawings
FIG. 1 is a diagram of a semantic logic structure.
FIG. 2 is a diagram of word vector context stitching.
FIG. 3 is a flow chart of a model framework architecture.
Detailed Description
The invention provides a high-precision tourism evaluation emotion classification method with learning performance and combining an At _ GRU neural network and an emotion dictionary, which comprises the following three steps as shown in figure 3:
first, emotional characteristic processing stage
(1) Constructing a tourism emotion dictionary: counting various emotion dictionaries such as a Qinghua university dictionary, a Hownet emotion dictionary, a simplified Chinese emotion dictionary of Taiwan university, and the like, merging words with the same polarity, enriching words with different polarities, perfecting and creating a composite emotion dictionary, performing word vectorization on the composite emotion dictionary and travel evaluation after word segmentation respectively, calculating Euclidean distances between each word in the dictionary and all words in the travel evaluation, selecting the words with similar distances as similar emotion words, and finally fusing and creating the composite emotion dictionary special for travel.
(2) Semantic logic rule processing: reading each piece of travel comment, and separating the comment by taking Chinese and English punctuation marks (.,!) as a sentence mark; an emotion value is acquired for each clause of the comment, the calculation and analysis process is as shown in fig. 1, in the figure, Y represents that the emotion value is 1, N represents that the emotion value is 0, the emotion value of each clause is analyzed in the dimension of whether negative words, adverb words, emotion words, exclamation sentences and question sentences are contained or not, the emotion value is positively represented by '1', the emotion value is negatively represented by '-1', the emotion value is neutrally represented by '0', and a 5-dimensional vector is formed to serve as an emotion polarity vector of the comment.
(3) Vectorizing emotional characteristics: and performing dimension increasing and word vectorization on the emotion polarity of the processed clauses.
Second, data preprocessing stage
(1) Word segmentation, stop word removal and word vector training: the original text information is defined as S, containing a constructed set of words W (W)1,w2,...,wn) And n represents the number of words of sentence S. After segmenting the original text and removing stop words using ICTCCLAS tool, Word set W (W) is segmented using Word2vec tool1,w2,...,wn) Performing word vector training, each word wiA word vector form with dimension 50 is generated.
(2) Combining the stitching context vectors: defining the ith word w in a sentenceiAll the remaining sentence information contained on the left is Cl(wi) And the left sentence information on the right is Cr(wi) Converting the word vector into a word vector and then generating a combined spliced 50-dimensional word vector, wherein the specific definition expression is as follows:
cl(wi)＝f(W(l)cl(wi-1)+W(sl)e(wi-1))
cr(wi)＝f(W(r)cr(wi+1)+W(sr)e(wi+1))
wherein, W(l)W(r)For converting a hidden layer, i.e. a context, into a matrix for the next hidden layer, W(sl)For the matrix used to combine the semantics of the text relationship left of the current word with the next word, W(sr)For a matrix combining the semantics of the text relationship of the current word to the right of the last word, f is a non-linear activation function, cl(wi-1) Is the i-1 th word wi-1All remaining sentence information contained on the left, e (w)i-1) Is the i-1 th word wi-1Word vector form of，e(wi+1) Is the i +1 th word wi+1Word vector form of cr(wi+1) Is the i +1 th word wi+1All the remaining sentence information contained on the right;
three, two-way GRU text semantic classification model stage
(1) Bidirectional GRU network training: constructing a bidirectional GRU network, and training a 55-dimensional word vector training set Y { Y }1,y2,...,ynAnd (5) loading the sentence into the At _ GRU model in the forward direction and the reverse direction respectively to form two-way model training.
(2) The system loads the trained network model, then loads the trained network model into a new tourism user evaluation model, and the model carries out emotion classification on the new tourism user evaluation model, so that emotion polarity classification of three degrees of satisfaction, generality and dissatisfaction of each user evaluation in dimensions such as route planning, forced consumption, traffic convenience, time arrangement and the like is realized, clear and intuitive understanding is provided for tourists on a plurality of concerned aspects of the tourism line, and a better line decision is provided.
Examples
For the following travel comment text "is indeed one of the four gardens in China! No matter the scale or the scenery is, none of the trails which are peculiar to south of the river, pavilion and slope of the curved path through the pylons! A water pond. The following steps are carried out:
1. data pre-processing
1) The data is processed by word segmentation, noise removal and the like, and the following results are finally generated:
"four-large gardens in China! No matter the scene is a large-scale scene, the special small bridge flowing water in south of the Yangtze river, the pavilion, the bent path and the pylorus are inclined! One time as much lotus leaves in the north of the lotus pond in water, the lotus flowers shake and shine, and the water pond is used for more than ten days.
2) Performing word vector training on each word to generate a 50-dimensional word vector as follows:
"gardens 0.151640.30177-0.167630.176840.317190.33973-0.43478-0.31086-0.44999-0.294860.166080.11963-0.41328"
3) Concatenating the word vectors of each word and its context-dependent word to generate a completely new 50-dimensional word vector
2. Emotional characteristic processing stage
1) The method is characterized by comprising the steps of sorting and integrating a plurality of known emotion dictionaries, such as a Homeenet emotion dictionary, a simplified Chinese emotion dictionary of Taiwan university, an emotion dictionary of Qinghua university and the like.
2) And (3) vectorizing each word in the emotion dictionary by using the words, calculating Euclidean distance between the word vector and the evaluation word vector generated in the first step, selecting the word with the similarity close to 10% in each emotion dictionary, and forming and constructing a special tourism emotion dictionary.
3) And analyzing and processing emotion polarity of each travel evaluation based on the exclusive travel emotion dictionary, analyzing whether degree adverbs, question reversers and the like exist or not, if yes, marking 1 and not marking 0, and finally generating a 6-dimensional vector, such as 0, 1, 1.1, 0, 0 and 1.
4) And performing fusion processing on the word vectors generated in the first step to finally generate 55-dimensional word vectors V1' 0.317190.33973-0.43478-0.31086-0.44999-0.294860.16608-0.41328.
3. Bidirectional GRU text semantic classification model stage
1) Building GRU neural network, and loading training set word vector V1
2) Training the network, manually adjusting parameters, and finally generating a model M
3) The test set is loaded and ultimately produces the classification effect R (1, 0, 1, -1, 0).
The invention strengthens context semantic relation through a word vector splicing form and combines the emotion polarity characteristics generated based on semantic logic rules to generate a brand new characteristic vector through fusion. The method has the accuracy of the emotion dictionary and the robustness of machine learning. The experimental result proves that the provided semantic classification method has higher accuracy.
Claims (2)
1. A travel evaluation emotion classification method combining an At _ GRU neural network and an emotion dictionary is used for realizing the evaluation semantic classification of a travel user on the whole travel according to an evaluation text of a tourist on the whole travel, and is characterized by comprising the following steps of:
1) and an emotional characteristic processing stage: the method specifically comprises the following steps of performing vectorization processing on emotional features in the travel comments by constructing a composite type special travel emotion dictionary:
11) constructing a composite emotion dictionary special for tourism: counting various existing emotion dictionaries, combining words with the same polarity to form a composite emotion dictionary, vectorizing the composite emotion dictionary and the segmented tourism evaluation words respectively, acquiring Euclidean distances between each word in the dictionary and all words of tourism evaluation, selecting the words with similar distances as similar emotion words, and fusing to form the composite special emotion dictionary for tourism;
12) semantic logic rule processing: reading travel comments, dividing the comments into sentences by taking Chinese and English punctuations as sentence marks, acquiring the emotion value of each sentence according to the part of speech of the sentence, and taking the emotion value of each sentence as the emotion polarity of the comment;
13) vectorizing emotional characteristics: performing dimension increase and vectorization on the emotion polarity of the processed clauses;
2) a data preprocessing stage: the method comprises the following steps of carrying out context vector splicing on original comment text training word vectors, fusing the spliced vectors and vectorized emotional features to serve as input of a bidirectional GRU neural network, and specifically comprising the following steps:
21) word segmentation, removal of stop words and training word vectors: the original comment text information is defined as S and contains a constructed word set W (W)1,w2,...,wn) N is the number of words in the sentence S, the original comment text is participled by adopting an ICTCCLAS tool, stop words are removed, and a Word set W (W) is divided by adopting a Word2vec tool1,w2,...,wn) Performing word vector training, each word wiA word vector form represented as 50 dimensions;
22) combining the stitching context vectors: defining the ith word w in a sentenceiAll the remaining sentence information contained on the left is Cl(wi) And the left sentence information on the right is Cr(wi) Converting the word vector into a word vector and then generating a combined spliced 50-dimensional word vector, wherein the specific definition expression is as follows:
cl(wi)＝f(W(l)cl(wi-1)+W(sl)e(wi-1))
cr(wi)＝f(W(r)cr(wi+1)+W(sr)e(wi+1))
wherein, W(l)W(r)For converting a hidden layer, i.e. a context, into a matrix for the next hidden layer, W(sl)For the matrix used to combine the semantics of the text relationship left of the current word with the next word, W(sr)For a matrix combining the semantics of the text relationship of the current word to the right of the last word, f is a non-linear activation function, cl(wi-1) Is the i-1 th word wi-1All remaining sentence information contained on the left, e (w)i-1) Is the i-1 th word wi-1The word vector form of e (w)i+1) Is the i +1 th word wi+1Word vector form of cr(wi+1) Is the i +1 th word wi+1All the remaining sentence information contained on the right;
23) and (3) feature vector fusion: performing quantity product processing on the spliced 50-dimensional word vector and the vectorized emotional characteristics, fusing to form a final 55-dimensional word vector, and taking the final 55-dimensional word vector as the input of the GRU network;
3) a bidirectional GRU text semantic classification model stage: the method specifically comprises the following steps of training a bidirectional GRU neural network and classifying tourism evaluation emotions:
31) training a bidirectional GRU network: constructing a bidirectional GRU network, loading a 55-dimensional word vector training set into an At _ GRU model from the forward direction and the reverse direction of a sentence respectively, and performing parameter tuning to finish training;
32) for the trained bidirectional GRU network model, new tourist user evaluation is subjected to data preprocessing to form word vectors, the word vectors are loaded into the model for emotion classification, natural language emotion analysis of each user evaluation is realized, emotion polarity classification in three degrees of satisfaction, generality and dissatisfaction is finally shown in the form of 5 dimensions of tour guide service, forced consumption, traffic routes, route arrangement and accommodation and catering, the emotion polarity classification is respectively shown by 1, 0 and-1, and experience feedback of passengers in each tour route to the 5 dimensions is shown.
2. The method for classifying traveling assessment emotion according to claim 1, wherein the At _ GRU neural network and emotion dictionary are combined, in the step 11), the existing emotion dictionaries comprise a Qinghua university dictionary, a Hownet emotion dictionary, and a Taiwan university simplified Chinese emotion dictionary.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810862476.4A CN109213861B (en) | 2018-08-01 | 2018-08-01 | Traveling evaluation emotion classification method combining At _ GRU neural network and emotion dictionary |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810862476.4A CN109213861B (en) | 2018-08-01 | 2018-08-01 | Traveling evaluation emotion classification method combining At _ GRU neural network and emotion dictionary |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109213861A CN109213861A (en) | 2019-01-15 |
| CN109213861B true CN109213861B (en) | 2022-03-29 |
Family
ID=64987780
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810862476.4A Active CN109213861B (en) | 2018-08-01 | 2018-08-01 | Traveling evaluation emotion classification method combining At _ GRU neural network and emotion dictionary |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109213861B (en) |
Families Citing this family (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109858973A (en) * | 2019-02-18 | 2019-06-07 | 成都中科大旗软件有限公司 | A kind of analysis method of regional tourism industry development |
| CN109948665B (en) * | 2019-02-28 | 2020-11-27 | 中国地质大学（武汉） | Human activity type classification method and system based on long-time and short-time memory neural network |
| CN110059922A (en) * | 2019-03-11 | 2019-07-26 | 北京比速信息科技有限公司 | Satisfaction evaluation method on the line of data is commented on based on internet tourist |
| CN109933664B (en) * | 2019-03-12 | 2021-09-07 | 中南大学 | Fine-grained emotion analysis improvement method based on emotion word embedding |
| CN110019882B (en) * | 2019-03-18 | 2022-01-28 | 新浪网技术(中国)有限公司 | Advertisement creative classification method and system |
| CN110083825A (en) * | 2019-03-21 | 2019-08-02 | 昆明理工大学 | A kind of Laotian sentiment analysis method based on GRU model |
| CN110083829A (en) * | 2019-04-03 | 2019-08-02 | 平安科技（深圳）有限公司 | Feeling polarities analysis method and relevant apparatus |
| CN110196976B (en) * | 2019-05-10 | 2020-10-16 | 新华三大数据技术有限公司 | Text emotional tendency classification method and device and server |
| CN110245299B (en) * | 2019-06-19 | 2022-02-08 | 中国人民解放军国防科技大学 | Sequence recommendation method and system based on dynamic interaction attention mechanism |
| CN110717090A (en) * | 2019-08-30 | 2020-01-21 | 昆山市量子昆慈量子科技有限责任公司 | Network public praise evaluation method and system for scenic spots and electronic equipment |
| CN111259651A (en) * | 2020-01-21 | 2020-06-09 | 北京工业大学 | User emotion analysis method based on multi-model fusion |
| CN111340385A (en) * | 2020-03-10 | 2020-06-26 | 深圳华侨城创新研究院有限公司 | Scientific measuring method for measuring joy index of tourist attraction |
| CN112711664B (en) * | 2020-12-31 | 2022-09-20 | 山西三友和智慧信息技术股份有限公司 | Text emotion classification method based on TCN + LSTM |
| CN113033186B (en) * | 2021-05-31 | 2021-08-17 | 江苏联著实业股份有限公司 | Error correction early warning method and system based on event analysis |
| CN116533529B (en) * | 2023-05-12 | 2023-09-29 | 湖州东尼新能源有限公司 | Intelligent control method and system for ultrasonic welding PC (polycarbonate) sheet |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104572892A (en) * | 2014-12-24 | 2015-04-29 | 中国科学院自动化研究所 | Text classification method based on cyclic convolution network |
| CN107168945A (en) * | 2017-04-13 | 2017-09-15 | 广东工业大学 | A kind of bidirectional circulating neutral net fine granularity opinion mining method for merging multiple features |
| CN107609132A (en) * | 2017-09-18 | 2018-01-19 | 杭州电子科技大学 | One kind is based on Ontology storehouse Chinese text sentiment analysis method |
| CN108108433A (en) * | 2017-12-19 | 2018-06-01 | 杭州电子科技大学 | A kind of rule-based and the data network integration sentiment analysis method |
-
2018
- 2018-08-01 CN CN201810862476.4A patent/CN109213861B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104572892A (en) * | 2014-12-24 | 2015-04-29 | 中国科学院自动化研究所 | Text classification method based on cyclic convolution network |
| CN107168945A (en) * | 2017-04-13 | 2017-09-15 | 广东工业大学 | A kind of bidirectional circulating neutral net fine granularity opinion mining method for merging multiple features |
| CN107609132A (en) * | 2017-09-18 | 2018-01-19 | 杭州电子科技大学 | One kind is based on Ontology storehouse Chinese text sentiment analysis method |
| CN108108433A (en) * | 2017-12-19 | 2018-06-01 | 杭州电子科技大学 | A kind of rule-based and the data network integration sentiment analysis method |
Non-Patent Citations (1)
| Title |
|---|
| 基于多元特征融合和 LSTM神经网络的中文评论情感分析;李科;《中国优秀硕士学位论文全文数据库信息科技辑》;20180115;论文第3-4章 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109213861A (en) | 2019-01-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109213861B (en) | Traveling evaluation emotion classification method combining At _ GRU neural network and emotion dictionary | |
| CN109933664B (en) | Fine-grained emotion analysis improvement method based on emotion word embedding | |
| CN106777275B (en) | Entity attribute and property value extracting method based on more granularity semantic chunks | |
| CN109829159B (en) | Integrated automatic lexical analysis method and system for ancient Chinese text | |
| Sánchez et al. | ICFHR2014 competition on handwritten text recognition on transcriptorium datasets (HTRtS) | |
| CN109766544B (en) | Document keyword extraction method and device based on LDA and word vector | |
| CN108563638B (en) | Microblog emotion analysis method based on topic identification and integrated learning | |
| CN111931506B (en) | Entity relationship extraction method based on graph information enhancement | |
| CN109977413A (en) | A kind of sentiment analysis method based on improvement CNN-LDA | |
| CN110276054B (en) | Insurance text structuring realization method | |
| CN110362819B (en) | Text emotion analysis method based on convolutional neural network | |
| CN112231472B (en) | Judicial public opinion sensitive information identification method integrated with domain term dictionary | |
| CN110929034A (en) | Commodity comment fine-grained emotion classification method based on improved LSTM | |
| CN111897917B (en) | Rail transit industry term extraction method based on multi-modal natural language features | |
| CN112818698B (en) | Fine-grained user comment sentiment analysis method based on dual-channel model | |
| CN112434164B (en) | Network public opinion analysis method and system taking topic discovery and emotion analysis into consideration | |
| CN106446147A (en) | Emotion analysis method based on structuring features | |
| CN110119443A (en) | A kind of sentiment analysis method towards recommendation service | |
| CN111159342A (en) | Park text comment emotion scoring method based on machine learning | |
| CN112561718A (en) | Case microblog evaluation object emotion tendency analysis method based on BilSTM weight sharing | |
| CN110321434A (en) | A kind of file classification method based on word sense disambiguation convolutional neural networks | |
| CN110717341A (en) | Method and device for constructing old-Chinese bilingual corpus with Thai as pivot | |
| CN107818173B (en) | Vector space model-based Chinese false comment filtering method | |
| CN113312922A (en) | Improved chapter-level triple information extraction method | |
| CN107451116B (en) | Statistical analysis method for mobile application endogenous big data |
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 |