CN107133282B - Improved evaluation object identification method based on bidirectional propagation - Google Patents
Improved evaluation object identification method based on bidirectional propagation Download PDFInfo
- Publication number
- CN107133282B CN107133282B CN201710248437.0A CN201710248437A CN107133282B CN 107133282 B CN107133282 B CN 107133282B CN 201710248437 A CN201710248437 A CN 201710248437A CN 107133282 B CN107133282 B CN 107133282B
- Authority
- CN
- China
- Prior art keywords
- word
- evaluation
- evaluation object
- candidate
- candidate evaluation
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
- G06F16/2465—Query processing support for facilitating data mining operations in structured databases
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/253—Grammatical analysis; Style critique
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Databases & Information Systems (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- General Health & Medical Sciences (AREA)
- Fuzzy Systems (AREA)
- Mathematical Physics (AREA)
- Probability & Statistics with Applications (AREA)
- Software Systems (AREA)
- Data Mining & Analysis (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Machine Translation (AREA)
Abstract
The invention discloses an improved evaluation object identification method based on two-way propagation, which comprises the steps of carrying out dependency syntax analysis on all comments of a class of products one by one and extracting a word pair list with a specific relationship from the comments; then, a small amount of emotional words are used as seeds, iterative identification of evaluation elements is carried out in the word pair list, and a rough candidate evaluation object set and an evaluation word set are obtained; extracting accurate evaluation objects from the candidate evaluation object set by using a rule with high accuracy; and extracting the rest evaluation objects by utilizing similarity calculation, PMI and association rules based on word vectors according to the accurate evaluation objects, and finally obtaining a complete evaluation object set. The method and the device can accurately acquire the attributes or aspects of the products described in the product reviews, and the attributes or aspects are used as a knowledge base for review analysis, so that the accuracy and the integrity of the review analysis are improved.
Description
Technical Field
The invention relates to the field of opinion mining, in particular to an improved evaluation object identification method based on bidirectional propagation.
Background
The internet has gradually penetrated into the aspects of social life, and along with the rise of the mobile internet, the penetration mode of the internet becomes more diversified and the content is richer. For example, social platforms, e-commerce, online payment, internet finance, blogs, BBS, etc., which are currently receiving much attention, have already provided products and services to a large number of users in different forms using the internet as a carrier. Meanwhile, the common users no longer receive the products or services only unilaterally, and the participation degree of the internet of the common users is continuously improved. After a user purchases or obtains a service, the behavior of online commenting on a product or the service is particularly remarkable. The comment information of the user on the product or service reflects the opinion and attitude of the user on the product quality or service level. The comment information has important significance for consumers and merchants: for the consumers, the consumers can objectively obtain the information of each dimension of the product through the comment information of other consumers, and the merchants can improve the product or make a sales strategy according to the comment information fed back by the customers.
The user comment information has the characteristics of huge amount and different standards, so that a great deal of manpower and time are consumed only by a traditional manual review method, and at the moment, a machine is needed to help a human to process the huge user comment information and quickly arrange adult-understandable structured information, which is also the purpose of opinion mining technology.
Opinion mining techniques mainly use natural language processing, information identification and extraction, data mining and the like as means to identify and extract valuable point of view information from a large amount of text information. One important task is the extraction of evaluation objects. The existing evaluation object extraction method is lack of a method with high accuracy and high recall rate.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provides an improved evaluation object identification method based on bidirectional propagation, which can effectively extract an evaluation object from an evaluation corpus.
The purpose of the invention can be realized by the following technical scheme:
an improved evaluation object identification method based on bidirectional propagation, comprising the following steps:
s1, obtaining a large amount of comment corpora of a class of products;
s2, performing dependency syntax analysis on each comment corpus by using a dependency syntax tool, and extracting all specific dependency relationship pairs < word _ object, word _ sentiment >;
s3, utilizing the seed emotion dictionary, and using an evaluation element iterative recognition algorithm to perform iterative recognition on the word _ object and word _ sentiment > according to the dependency relationship extracted in the step S2 until convergence to obtain a candidate evaluation object set CO and a candidate evaluation word set CS;
s4, extracting word frequency larger than threshold lambda from candidate evaluation object set CO1The words are used as accurate evaluation objects, and word frequency which is greater than a threshold value lambda is extracted from a candidate evaluation word set CS1The word frequency is a candidate evaluation object or a candidate evaluation word in a dependency relationship pair<word_object,word_sentiment>The number of occurrences in (a);
and S5, extracting the residual candidate evaluation objects in the step S4 by using word2vec, the association rule and PMI-IR to obtain a final accurate evaluation object set O.
Further, in step S1, the products of the category are products on websites of e-commerce and virtual products, in step S2, the dependency syntax tool is a chinese language processing tool LTP, and in step S3, the seed emotion dictionary is a positive and negative evaluation dictionary.
Further, in step S2, the specific dependency relationship pair is a word pair < word _ object, word _ sense > that satisfies the four syntax relationships of SBV, VOB, ATT, and CMP.
Further, the specific process of step S3 is as follows:
step S31, initializing the candidate evaluation object set CO to be empty, and initializing the candidate evaluation word set CS to be empty;
step S32, adding words word _ sense belonging to the seed emotion dictionary in the dependency relationship pair < word _ object, word _ sense > into the candidate evaluation word set CS;
step S33, adding word _ objects which correspond to the words in the candidate evaluation word set CS and belong to nouns into the candidate evaluation object set CO;
step S34, adding word _ sentient which corresponds to the words in the candidate evaluation object set CO and belongs to the adjectives into the candidate evaluation word set CS;
step S35, repeating step S33 and step S34 until the candidate evaluation object set CO and the candidate evaluation word set CS are no longer changed.
Further, in step S4, the threshold λ1The value range is as follows: lambda [ alpha ]1E is N, where is λ1＝10。
Further, the specific process of step S5 is as follows:
step S51, performing word2vec training on the large amount of comment corpus obtained in the step S1 by using an open source word2vec tool to obtain a word vector of each word, wherein the form of the word vector is [ w [ ]i1,wi2，…wik…，wim]Wherein w isikIs the value of the k-th dimension of the word direction of the ith word, and m is the wordThe dimension of the vector;
step S52, traversing the residual candidate evaluation objects in the step S4, and based on the word vectors obtained by training in the step S51, enabling the similarity of the word vectors with the accurate evaluation objects to be larger than a threshold lambda2Adding the evaluation object into an accurate evaluation object set O;
step S53, based on the dependency relationship obtained in step S2, performing association rule discovery on the word _ object and the word _ sentiment > on the candidate evaluation objects remaining in step S4 and the accurate evaluation object set O obtained in step S52 to obtain an association rule < object, sentiment > set, and adding the object into the accurate evaluation object set O;
step S54, calculating PMI-IR value of the dependency relationship pair containing the low-frequency candidate object and the accurate evaluation word by utilizing the search engine, and enabling the PMI-IR value to be larger than the threshold value lambda3Is dependent on the relationship pair<word_object,word_sentiment>The evaluation object in (1) is added to the accurate evaluation object set O.
Further, in step S52, the word vector similarity calculation formula is:
wherein v isiWord vector, v, representing the ith wordjWord vectors, w, representing the jth wordikValue, w, of the k-th dimension of the word vector representing the ith wordjkThe value of the k-th dimension of the word vector representing the jth word, and m is the dimension of the word vector.
Further, in step S52, the threshold λ2The value range is as follows: lambda [ alpha ]2∈(0,1]Here, take λ2＝0.7。
Further, in step S54, the PMI-IR has the formula:
where hit (x) is the number of hits of the search term x in the search engine, which is a constant term.
Further, in step S54, the threshold λ3The value range is as follows: lambda [ alpha ]3∈(-∞,0]Here, take λ3＝-6。
Compared with the prior art, the invention has the following advantages and beneficial effects:
1. the method adopts a technical scheme of evaluation object identification based on bidirectional propagation, particularly processes the Chinese comment text through dependency syntax, fully excavates the dependency relationship between words in the text, and performs iterative identification on the evaluation object and the evaluation word from the emotional words, thereby achieving the effect of improving the identification accuracy and recall rate of the evaluation object.
2. The invention adopts a technical scheme of recommending the evaluation object based on the word vector, measures the similarity between the evaluation object to be evaluated and the accurate evaluation object by using the word vector obtained by large-scale corpus training, and recommends the evaluation object with large similarity, thereby achieving the effect of improving the identification accuracy and recall rate of the evaluation object from the aspect of semantic similarity.
3. The invention adopts the technical scheme of recommending the evaluation object based on the association rule, and the scheme achieves the effect of improving the identification recall rate of the evaluation object by means of recommending the evaluation object to be evaluated which has strong association with the accurate evaluation word.
4. The technical scheme of recommending the evaluation object based on the PMI-IR information of the search engine is adopted, and the effect of improving the identification recall rate of the evaluation object is achieved by means of recommending the evaluation object to be evaluated with a high PMI-IR value of an accurate evaluation word.
Drawings
FIG. 1 is an overall flow chart of an embodiment of the present invention.
Fig. 2 is a flowchart of an evaluation element iterative identification algorithm according to an embodiment of the present invention.
Fig. 3 is a flowchart of recommending an evaluation object by using word2vec, an association rule, and PMI-IR according to an embodiment of the present invention.
Detailed Description
The present invention will be described in further detail with reference to examples and drawings, but the present invention is not limited thereto.
Example (b):
the embodiment provides an improved evaluation object identification method based on bidirectional propagation, and a flow chart of the method is shown in fig. 1, and the method comprises the following steps:
s1, obtaining a large amount of comment corpora of a class of products;
s2, performing dependency syntax analysis on each comment corpus by using a dependency syntax tool, and extracting all specific dependency relationship pairs < word _ object, word _ sentiment >;
s3, utilizing the seed emotion dictionary, and using an evaluation element iterative recognition algorithm to perform iterative recognition on the word _ object and word _ sentiment > according to the dependency relationship extracted in the step S2 until convergence to obtain a candidate evaluation object set CO and a candidate evaluation word set CS;
s4, extracting word frequency larger than threshold lambda from candidate evaluation object set CO1The words are used as accurate evaluation objects, and word frequency which is greater than a threshold value lambda is extracted from a candidate evaluation word set CS1The term of (a) is taken as an accurate evaluation term, and the threshold value lambda1E is N, where is λ110, wherein the word frequency is a candidate evaluation object or a candidate evaluation word in dependency relationship pair<word_object,word_sentiment>The number of occurrences in (a);
and S5, extracting the residual candidate evaluation objects in the step S4 by using word2vec, the association rule and PMI-IR to obtain a final accurate evaluation object set O.
The method comprises the steps of carrying out dependency syntax analysis on all comments of a class of products item by item, and extracting a word pair list with a specific relationship from the comments; then, a small amount of emotion dictionaries are used as seeds, iterative identification of evaluation elements is carried out in the word pair list, and a rough candidate evaluation object set and an evaluation word set are obtained; extracting accurate evaluation objects from the candidate evaluation object set by using a rule with high accuracy; and extracting the rest evaluation objects by utilizing similarity calculation, PMI and association rules based on word vectors according to the accurate evaluation objects, and finally obtaining a complete evaluation object set.
The flowchart of step S3 is shown in fig. 2, and the specific process is as follows:
step S31, initializing the candidate evaluation object set CO to be empty, and initializing the candidate evaluation word set CS to be empty;
step S32, adding words word _ sense belonging to the seed emotion dictionary in the dependency relationship pair < word _ object, word _ sense > into the candidate evaluation word set CS;
step S33, adding word _ objects which correspond to the words in the candidate evaluation word set CS and belong to nouns into the candidate evaluation object set CO;
step S34, adding word _ sentient which corresponds to the words in the candidate evaluation object set CO and belongs to the adjectives into the candidate evaluation word set CS;
step S35, repeating step S33 and step S34 until the candidate evaluation object set CO and the candidate evaluation word set CS are no longer changed.
The flowchart of step S5 is shown in fig. 3, and the specific process is as follows:
step S51, performing word2vec training on the large amount of comment corpus obtained in the step S1 by using an open source word2vec tool to obtain a word vector of each word, wherein the form of the word vector is [ w [ ]i1,wi2，…wik…，wim]Wherein w isikIs the value of the k-th dimension of the word vector of the ith word, and m is the dimension of the word vector;
step S52, traversing the residual candidate evaluation objects in the step S4, and based on the word vectors obtained by training in the step S51, enabling the similarity of the word vectors with the accurate evaluation objects to be larger than a threshold lambda2Adding the evaluation object to the accurate evaluation object set O, and determining the threshold lambda2The value range is as follows: lambda [ alpha ]2∈(0,1]Here, take λ20.7, wherein the word vector similarity calculation formula is:
wherein v isiWord vector, v, representing the ith wordjWord vectors, w, representing the jth wordikValue, w, of the k-th dimension of the word vector representing the ith wordjkThe value of the k-th dimension of the word vector representing the jth word, and m is the dimension of the word vector. (ii) a
Step S53, based on the dependency relationship obtained in step S2, performing association rule discovery on the word _ object and the word _ sentiment > on the candidate evaluation objects remaining in step S4 and the accurate evaluation object set O obtained in step S52 to obtain an association rule < object, sentiment > set, and adding the object into the accurate evaluation object set O;
step S54, calculating PMI-IR value of the dependency relationship pair containing the low-frequency candidate object and the accurate evaluation word by utilizing the search engine, and enabling the PMI-IR value to be larger than the threshold value lambda3Is dependent on the relationship pair<word_object,word_sentiment>The evaluation object in (1) is added into an accurate evaluation object set O, and the threshold value lambda is3The value range is as follows: lambda [ alpha ]3∈(-∞,0]Here, take λ3-6, the formula of PMI-IR is:
where hit (x) is the number of hits of the search term x in the search engine, which is a constant term.
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 (8)
1. An improved evaluation object identification method based on bidirectional propagation is characterized by comprising the following steps:
s1, obtaining a large amount of comment corpora of a class of products;
s2, performing dependency syntax analysis on each comment corpus by using a dependency syntax tool, and extracting all specific dependency relationship pairs < word _ object, word _ sentiment >;
s3, utilizing the seed emotion dictionary, and using an evaluation element iterative recognition algorithm to perform iterative recognition on the word _ object and word _ sentiment > according to the dependency relationship extracted in the step S2 until convergence to obtain a candidate evaluation object set CO and a candidate evaluation word set CS; the specific process is as follows:
step S31, initializing the candidate evaluation object set CO to be empty, and initializing the candidate evaluation word set CS to be empty;
step S32, adding words word _ sense belonging to the seed emotion dictionary in the dependency relationship pair < word _ object, word _ sense > into the candidate evaluation word set CS;
step S33, adding word _ objects which correspond to the words in the candidate evaluation word set CS and belong to nouns into the candidate evaluation object set CO;
step S34, adding word _ sentient which corresponds to the words in the candidate evaluation object set CO and belongs to the adjectives into the candidate evaluation word set CS;
step S35, repeating the step S33 and the step S34 until the candidate evaluation object set CO and the candidate evaluation word set CS are not changed;
s4, extracting word frequency larger than threshold lambda from candidate evaluation object set CO1The words are used as accurate evaluation objects, and word frequency which is greater than a threshold value lambda is extracted from a candidate evaluation word set CS1The word frequency is a candidate evaluation object or a candidate evaluation word in a dependency relationship pair<word_object,word_sentiment>The number of occurrences in (a);
s5, extracting the remaining candidate evaluation objects in the step S4 by using word2vec, association rules and PMI-IR to obtain a final accurate evaluation object set O; the specific process is as follows:
step S51, performing word2vec training on the large amount of comment corpus obtained in the step S1 by using an open source word2vec tool to obtain a word vector of each word, wherein the form of the word vector is [ w [ ]i1,wi2，…wik…，wim]Wherein w isikIs the value of the k-th dimension of the word vector of the ith word, and m is the dimension of the word vector;
step S52, traversing the residual candidate evaluation objects in the step S4, and based on the word vectors obtained by training in the step S51, enabling the similarity of the word vectors with the accurate evaluation objects to be larger than a threshold lambda2Adding the evaluation object into an accurate evaluation object set O;
step S53, based on the dependency relationship obtained in step S2, performing association rule discovery on the word _ object and the word _ sentiment > on the candidate evaluation objects remaining in step S4 and the accurate evaluation object set O obtained in step S52 to obtain an association rule < object, sentiment > set, and adding the object into the accurate evaluation object set O;
step S54, calculating PMI-IR value of the dependency relationship pair containing the low-frequency candidate object and the accurate evaluation word by utilizing the search engine, and enabling the PMI-IR value to be larger than the threshold value lambda3Is dependent on the relationship pair<word_object,word_sentiment>The evaluation object in (1) is added to the accurate evaluation object set O.
2. The improved evaluation object identification method based on bidirectional propagation as claimed in claim 1, characterized in that: in step S1, the products of the category refer to products on websites of e-commerce and virtual products, in step S2, the dependency syntax tool is a chinese language processing tool LTP, and in step S3, the seed emotion dictionary is a positive and negative evaluation dictionary.
3. The improved evaluation object identification method based on bidirectional propagation as claimed in claim 1, characterized in that: in step S2, the specific dependency relationship pair is a word pair < word _ object, word _ sense > that satisfies four syntax relationships of SBV, VOB, ATT, and CMP.
4. The improved evaluation object identification method based on bidirectional propagation as claimed in claim 1, characterized in that: in step S4, the threshold λ1The value range is as follows: lambda [ alpha ]1E is N, where is λ1＝10。
5. The improved evaluation object identification method based on bidirectional propagation as claimed in claim 1, characterized in that: in step S52, the word vector similarity calculation formula is:
wherein v isiWord vector, v, representing the ith wordjWord vectors, w, representing the jth wordikValue, w, of the k-th dimension of the word vector representing the ith wordjkThe value of the k-th dimension of the word vector representing the jth word, and m is the dimension of the word vector.
6. The improved evaluation object identification method based on bidirectional propagation as claimed in claim 1, characterized in that: in step S52, the threshold λ2The value range is as follows: lambda [ alpha ]2∈(0,1]Here, take λ2＝0.7。
7. The improved evaluation object identification method based on bidirectional propagation as claimed in claim 1, characterized in that: in step S54, the PMI-IR has the formula:
where hit (x) is the number of hits of the search term x in the search engine, and is a constant item, and object and sense are respectively the object and sense in the association rule < object, sense > set obtained in step S53, and refer to the evaluation object and the evaluation term that meet the association rule.
8. The improved evaluation object identification method based on bidirectional propagation as claimed in claim 1, characterized in that: in step S54, the threshold λ3The value range is as follows: lambda [ alpha ]3∈(-∞,0]Herein, thisIs taken from3＝-6。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710248437.0A CN107133282B (en) | 2017-04-17 | 2017-04-17 | Improved evaluation object identification method based on bidirectional propagation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710248437.0A CN107133282B (en) | 2017-04-17 | 2017-04-17 | Improved evaluation object identification method based on bidirectional propagation |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107133282A CN107133282A (en) | 2017-09-05 |
| CN107133282B true CN107133282B (en) | 2020-12-22 |
Family
ID=59714963
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710248437.0A Active CN107133282B (en) | 2017-04-17 | 2017-04-17 | Improved evaluation object identification method based on bidirectional propagation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107133282B (en) |
Families Citing this family (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108255808B (en) * | 2017-12-29 | 2021-10-22 | 东软集团股份有限公司 | Text division method and device, storage medium and electronic equipment |
| CN108614814B (en) * | 2018-04-20 | 2022-02-15 | 广东工业大学 | Evaluation information extraction method, device and equipment |
| CN110175325B (en) * | 2019-04-26 | 2023-07-11 | 南京邮电大学 | Comment analysis method based on word vector and syntactic characteristics and visual interaction interface |
| CN111144507B (en) | 2019-12-30 | 2021-06-08 | 北京百度网讯科技有限公司 | Emotion analysis model pre-training method and device and electronic equipment |
| CN111339752B (en) * | 2020-02-18 | 2023-04-25 | 哈尔滨工业大学 | Evaluation object-oriented emotion analysis method for multi-task joint learning |
| CN111563384B (en) * | 2020-04-22 | 2022-10-25 | 华南理工大学 | Evaluation object identification method and device for E-commerce products and storage medium |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102866989B (en) * | 2012-08-30 | 2016-09-07 | 北京航空航天大学 | Viewpoint abstracting method based on word dependence relationship |
| CN104268160B (en) * | 2014-09-05 | 2017-06-06 | 北京理工大学 | A kind of OpinionTargetsExtraction Identification method based on domain lexicon and semantic role |
| US20170068906A1 (en) * | 2015-09-09 | 2017-03-09 | Microsoft Technology Licensing, Llc | Determining the Destination of a Communication |
| CN106257455B (en) * | 2016-07-08 | 2019-09-17 | 闽江学院 | A kind of Bootstrapping method extracting viewpoint evaluation object based on dependence template |
-
2017
- 2017-04-17 CN CN201710248437.0A patent/CN107133282B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN107133282A (en) | 2017-09-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107133282B (en) | Improved evaluation object identification method based on bidirectional propagation | |
| CN110825876B (en) | Movie comment viewpoint emotion tendency analysis method | |
| CN107133224B (en) | Language generation method based on subject word | |
| CN108628833B (en) | Method and device for determining summary of original content and method and device for recommending original content | |
| CN103646088B (en) | Product comment fine-grained emotional element extraction method based on CRFs and SVM | |
| CN111563384B (en) | Evaluation object identification method and device for E-commerce products and storage medium | |
| CN111767741A (en) | Text emotion analysis method based on deep learning and TFIDF algorithm | |
| WO2015192734A1 (en) | Information processing method and apparatus | |
| CN108694647B (en) | Method and device for mining merchant recommendation reason and electronic equipment | |
| CN111125349A (en) | Graph model text abstract generation method based on word frequency and semantics | |
| CN108255813B (en) | Text matching method based on word frequency-inverse document and CRF | |
| CN105809473B (en) | Training method for matching model parameters, service recommendation method and corresponding device | |
| CN106708966A (en) | Similarity calculation-based junk comment detection method | |
| CN107944911B (en) | Recommendation method of recommendation system based on text analysis | |
| CN105630768A (en) | Cascaded conditional random field-based product name recognition method and device | |
| CN106649272A (en) | Named entity recognizing method based on mixed model | |
| CN108256968B (en) | E-commerce platform commodity expert comment generation method | |
| CN105183833A (en) | User model based microblogging text recommendation method and recommendation apparatus thereof | |
| CN106599054A (en) | Method and system for title classification and push | |
| CN103049470A (en) | Opinion retrieval method based on emotional relevancy | |
| WO2018068648A1 (en) | Information matching method and related device | |
| CN105740382A (en) | Aspect classification method for short comment texts | |
| CN109598586A (en) | A kind of recommended method based on attention model | |
| CN109063147A (en) | Online course forum content recommendation method and system based on text similarity | |
| CN110955750A (en) | Combined identification method and device for comment area and emotion polarity, and electronic equipment |
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 |