CN111144929A - Comment object and word combined extraction method for automobile industry user generated content - Google Patents
Comment object and word combined extraction method for automobile industry user generated content Download PDFInfo
- Publication number
- CN111144929A CN111144929A CN201911229735.0A CN201911229735A CN111144929A CN 111144929 A CN111144929 A CN 111144929A CN 201911229735 A CN201911229735 A CN 201911229735A CN 111144929 A CN111144929 A CN 111144929A
- Authority
- CN
- China
- Prior art keywords
- dictionary
- words
- evaluation
- user
- word
- 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
- 238000000605 extraction Methods 0.000 title claims abstract description 44
- 238000011156 evaluation Methods 0.000 claims abstract description 55
- 238000012549 training Methods 0.000 claims abstract description 34
- 238000000034 method Methods 0.000 claims abstract description 19
- 238000012360 testing method Methods 0.000 claims abstract description 10
- 230000004927 fusion Effects 0.000 claims abstract description 5
- 238000007781 pre-processing Methods 0.000 claims abstract description 5
- 238000011157 data evaluation Methods 0.000 claims abstract description 4
- 239000013598 vector Substances 0.000 claims description 5
- 230000006870 function Effects 0.000 description 7
- 230000014509 gene expression Effects 0.000 description 7
- 230000011218 segmentation Effects 0.000 description 6
- 238000011160 research Methods 0.000 description 4
- 239000000284 extract Substances 0.000 description 3
- 238000005065 mining Methods 0.000 description 3
- 238000010276 construction Methods 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- JEIPFZHSYJVQDO-UHFFFAOYSA-N iron(III) oxide Inorganic materials O=[Fe]O[Fe]=O JEIPFZHSYJVQDO-UHFFFAOYSA-N 0.000 description 2
- 238000003058 natural language processing Methods 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 238000005336 cracking Methods 0.000 description 1
- 230000002950 deficient Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 230000008451 emotion Effects 0.000 description 1
- 238000005187 foaming Methods 0.000 description 1
- 239000000446 fuel Substances 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000012552 review Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0201—Market modelling; Market analysis; Collecting market data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/374—Thesaurus
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9535—Search customisation based on user profiles and personalisation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- Strategic Management (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Accounting & Taxation (AREA)
- Development Economics (AREA)
- Finance (AREA)
- General Engineering & Computer Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Computational Linguistics (AREA)
- Game Theory and Decision Science (AREA)
- Economics (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Machine Translation (AREA)
Abstract
The invention discloses a comment object and word combined extraction method for user generated content in the automobile industry, which comprises the following steps: acquiring user data, integrating and preprocessing the user data to enable the format and naming mode of the user data to be uniform, and selecting partial data as a training sample set for manual marking; constructing a professional word dictionary in the automobile field; performing part-of-speech tagging on the preprocessed user data, designing an extraction rule of user data evaluation feature words, and extracting the feature words evaluated by the user to construct an evaluation feature word dictionary; designing the characteristics of a CRFs template, and introducing the constructed dictionary into the characteristic template; converting the training and testing data into a format of a feature template required by CRF + +, writing the features into the feature template in sequence, training the CRF + + template to obtain a training model, and testing by using the training model to realize the joint extraction of the evaluation object and the evaluation word of multi-feature fusion. The method reduces the interference of noise content on the extraction result, and can be suitable for the extraction of user generated content in the automobile field.
Description
Technical Field
The invention relates to the field of text mining and natural language processing, in particular to a comment object and word joint extraction method for user generated content in the automobile industry.
Background
The structure of the automobile product itself is complex, and it is difficult for a user to obtain desired information from a strict product specification or official configuration information. In recent years, vertical automobile websites are rapidly developed, users are willing to share own purchasing decision experience and use experience in public praise, forum and other channels including complaints, a large amount of user generated contents are generated in an internet service mode dominated by the users, and rich user demand information is stored in large-scale unstructured data, so that a basis is provided for other users to make purchasing decisions, reference is provided for improvement of automobile products, knowledge in user comment contents on the internet is fully identified, the demands of the users can be more accurately known, and update iteration of the automobile products is promoted.
The information extraction aims to convert the language text described in an unstructured or semi-structured mode into structured data, the extraction of the comment content of the user mainly identifies comment objects and comment words from the text to extract structured data, and the structured data has important significance for product competition and updating of enterprises.
The identification of the comment objects and comment words is a key for information extraction, and the common identification methods are mainly divided into three types, namely a rule-based method, a statistical-based method and a mixed method combining rules and statistics. The entity identification method based on the rules needs to manually establish the rules, and the establishment of the rules depends on extremely strong professional knowledge and text formats. Common algorithms based on statistical methods are hidden markov models, maximum entropy models, conditional random fields, etc.
Based on the method, the research is based on organization and mining of product information in the automobile field in the Internet environment, a website platform facing to the automobile vertical field is combined with text mining and natural language processing technologies, sequence labels of CRFs (conditional random fields) are utilized to analyze the characteristics of the user comment content, and a method capable of accurately identifying and jointly extracting comment objects and comment words is constructed.
In the existing research, many extraction methods are used for individually identifying extracted comment objects and comment words, and the existing technology has the following problems when analyzing entity extraction in the automobile field:
1) the information extraction technology needs to be adapted to new fields, corresponding research is carried out in a plurality of professional fields at present, and the research of the information extraction technology facing the content generated by automobile users is still very deficient.
2) The word bank in the automobile professional field is insufficient in breadth and depth, and a part of the user expressing content is lacking. The expression of the user usually uses an abbreviation of a function attribute or a similar meaning of a wrongly written or mispronounced character and function in the expression of the character, and the situations can cause the accuracy of the entity extraction result to be reduced.
3) The user generated content is varied, different relations exist between the comment objects and the comment words, such as satisfaction of the user on product attribute representation, complaints and the like, for the extraction of the comment objects and the comment words, the characteristics of the comment words and the relation between the comment objects and the comment words are fully considered, the extraction difficulty is reduced, and the accuracy is improved.
4) The independent extraction of the comment objects or the corresponding comment words cannot well reflect the context environment of the text, and due to the non-normative language in the network comment and the use of emerging network words, the independent extraction ignores the relationship between the comment objects and the comment words, so that part of the comment objects or the comment words can be filtered out, and the extraction result is deviated.
Disclosure of Invention
The invention provides a comment object and word combined extraction method for automobile industry user generated content, which is characterized in that the automobile industry user generated content is combined with a professional word dictionary and an evaluation feature word dictionary in the automobile field by using sequence labeling of CRFs (conditional random fields) to extract knowledge by fusing various features, and the details are described as follows:
a comment object and word joint extraction method for automobile industry user generated content comprises the following steps:
acquiring user data, integrating and preprocessing the user data to enable the format and naming mode of the user data to be uniform, and selecting partial data as a training sample set for manual marking;
constructing a professional word dictionary in the automobile field; performing part-of-speech tagging on the preprocessed user data, designing an extraction rule of user data evaluation feature words, and extracting the feature words evaluated by the user to construct an evaluation feature word dictionary;
designing the characteristics of a CRFs template, and introducing the constructed dictionary into the characteristic template;
converting the training and testing data into a format of a feature template required by CRF + +, writing the features into the feature template in sequence, training the CRF + + template to obtain a training model, and testing by using the training model to realize the joint extraction of the evaluation object and the evaluation word of multi-feature fusion.
Wherein, the automobile field professional dictionary comprises:
the method comprises the steps of using an automobile body knowledge base, using harmonic words and mispronounced words in the content generated by a user in the automobile field, and using word2vec training word vectors to extract similar words of product attributes.
Further, the feature dictionary includes: and adding a relation dictionary, a satisfaction dictionary and a complaint dictionary.
Wherein, the introducing the constructed dictionary into the feature template specifically comprises:
the characteristics comprise lexical characteristics, automobile field professional word dictionary characteristics, evaluation characteristic word dictionary characteristics, dependency relationship characteristics and relative position characteristics between evaluation objects and evaluation words.
The technical scheme provided by the invention has the beneficial effects that:
1. on the premise of utilizing the comment content characteristics and the relationship type defined by the functional application angle of the product, the relationship between the evaluation object and the evaluation word is fully considered, wherein the relationship comprises the dependence relationship and the relative position relationship of the comment object and the comment word;
2. the invention constructs a professional word dictionary and an evaluation characteristic word dictionary in the automobile field, introduces the professional word dictionary and the evaluation characteristic word dictionary into a CRF + + (CRFs realization system) template, and performs fusion of a plurality of characteristics;
3. the method reduces the interference of noise content on the extraction result, can adapt to the extraction of user generated content in the automobile field, and ensures the accuracy of the extraction result.
Drawings
FIG. 1 is a framework of a review object and word joint extraction method for automobile industry user-generated content;
FIG. 2 is a schematic diagram of construction of a professional word dictionary in the automotive field;
FIG. 3 is a drawing of the extraction rules and template design for CRFs (conditional random fields).
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, embodiments of the present invention are described in further detail below.
Example 1
A comment object and word combined extraction method for automobile industry user generated content is disclosed, referring to FIG. 1, and the method comprises the following steps:
101: the method comprises the steps of obtaining user data of the Internet in the automobile industry, storing the user data into a local database, integrating and preprocessing the user data (removing stop words and word segmentation), enabling the format and naming mode of the user data to be uniform, randomly selecting partial data as a training sample set for manual marking, and marking in a form of a comment object and a comment word >;
102: constructing a professional word dictionary in the automobile field;
the automobile field professional word dictionary mainly comprises an automobile body knowledge base, harmonic words and mispronounced words in automobile field user generated contents, and similar words for extracting product attributes by using word2vec training word vectors.
103: performing part-of-speech tagging on the user data preprocessed in the step 101, designing an extraction rule of user data evaluation feature words, extracting the feature words evaluated by the user to construct an evaluation feature word dictionary, and mainly comprising the following steps of: adding a relation dictionary, a satisfaction dictionary and a complaint dictionary;
in a specific implementation, a user usually uses words such as "poor", "reliable", and the like to describe the opinion on the attribute of the automobile product, and such words are defined as evaluation feature words.
Adding a relation dictionary: when an automobile product lacks a certain functional attribute or a user has a demand for a certain functional attribute, the user usually uses words such as 'none', 'no', 'lack' and the like, and defines the words as characteristic words of an increasing relationship to construct an increasing relationship dictionary.
Satisfaction dictionary: when a user is satisfied with the attributes of the automobile product, the user usually uses related expressions such as 'good' and 'reliable' and extracts the evaluation words of the types to construct a satisfaction dictionary by using the 'most satisfactory points' in the public praise data.
A complaint dictionary: when the user is not satisfied with the automobile product attribute, the user usually uses related complaint words such as 'bad', 'serious' and the like, and extracts the complaint words of the types mentioned above by using the 'most unsatisfactory point' in the word-of-mouth data and the user complaint data to construct a complaint dictionary.
104: designing the characteristics of a CRFs (conditional random field) template, and introducing dictionaries constructed in the steps 102 and 103 into the characteristic template, so that the designed characteristics mainly comprise lexical characteristics (results of word segmentation and part of speech tagging), professional word dictionary characteristics in the automobile field, evaluation characteristic word dictionary (increased relation dictionary, satisfaction dictionary and complaint dictionary) characteristics, dependency relationship characteristics between evaluation objects and evaluation words (direct dependency relationship exists between the evaluation objects and the evaluation words, such as a cardinal meaning relationship, a centering relationship and the like) and relative position characteristics;
105: converting the training and testing data into a format of a feature template required by a CRF + + (CRFs realization system), writing the features in the step 104 into the feature template in sequence, training the CRF + + template to obtain a training model, and testing by using the training model to realize the joint extraction of the evaluation object and the evaluation word of the multi-feature fusion.
The CRF + + is an implementation system of CRFs, the tool package used is CRF + +0.58, and CRF templates are trained in CRF + + 0.58.
In summary, in the embodiment of the present invention, through the above steps 101 to 105, joint extraction of the content comment object and the comment word generated by the user in the automobile industry is realized.
Example 2
The scheme of example 1 is further described below with reference to specific methods and examples, which are described in detail below:
201: acquiring public praise content of the automobile vertical website;
1) the specific data mainly comprises:
① vehicle type data, including the brand of the vehicle type, the type of the vehicle type, the time of coming to market, and the corresponding configuration of 'engine', 'gearbox', 'safety air bag', 'tire pressure monitoring', etc.
② user comment content includes vehicle type purchased, time purchased, "most satisfactory point", "most unsatisfactory point", "space", "power", "operation", "appearance", "interior", "fuel consumption", "comfort", "cost performance", and other product attributes.
2) Cleaning, integrating and preprocessing the acquired data
① the text content is preprocessed to delete duplicate data and incomplete content data.
②, integrating the acquired text content to ensure the uniform designation of the vehicle type name and each function attribute name.
3) Randomly selecting partial data for manual marking
Randomly selecting partial data as a training sample set, removing stop words, segmenting words and manually marking the stop words, the segmenting words, and designing a relatively simple marking form so as to avoid excessively complex marking: the text content evaluation object is marked as ' CO ', the evaluation word is marked as ' CW ', other irrelevant words are marked as ' WU ', the last column extracts an evaluation object-evaluation word ', and the training sample set is used for the subsequent training of a CRF + + (CRFs realization system) model.
202: constructing a professional word dictionary in the field of automobiles
The automobile field professional word dictionary is mainly used as a self-defined dictionary of word segmentation to increase the accuracy of word segmentation and introduce a characteristic template for guidance.
The automobile field professional word dictionary mainly comprises an automobile body knowledge base and corresponding statement words for automobile product attribute users. The automobile ontology knowledge base comprises: the functions and performances of the large and various parts of the automobile system, such as a power system, a transmission system, a chassis, an electric system, automobile body accessories, electric appliances and the like, specific parts of each system and various performance evaluation indexes for evaluating the running, steering, braking and the like of the automobile; the corresponding expression words of the automobile product attribute users comprise word2vec training word vectors, word vector models of user generated contents are obtained, similarity among words is calculated, near-meaning words of product attributes are extracted, and harmonic words and wrongly-recognized words of some users during expression are included as follows: the trunk, the seat and the like.
203: designing an extraction rule of the evaluation feature words for extraction, and constructing an evaluation feature word dictionary
Wherein, the evaluation feature word dictionary mainly comprises: and adding a relation dictionary, a satisfaction dictionary and a complaint dictionary. The specific construction method is as follows:
adding a relation dictionary: when an automobile product does not have a certain function or a user has a need for a certain function, the user usually uses words such as "none", "lack", and the like, and the words are characteristic words for increasing the relationship. Similar characteristic expressions are found mainly by using word2vec, added into a dictionary, and manually screened and supplemented.
Satisfaction dictionary: deleting repeated abnormal contents by using the most satisfied point in the public praise data, performing word segmentation and part-of-speech tagging on the text contents, wherein most of evaluation words are adjectives and verbs when a user expresses the evaluation words, and the designed extraction rule is as follows: pos ═ a, pos ═ v.
A complaint dictionary: the most unsatisfied point and complaint data in the public praise data are used to design a rule and a satisfaction dictionary. The evaluation words or emotion words in the comment feature dictionary are not further divided and are classified into comment words, and the following expression is constructed:
Token | Pos | Result |
vehicle body | n | |
Rust formation | v | Entering a complaint dictionary |
Painted surface | n | |
Foaming | v | Entering a complaint dictionary |
Cracking of | v | Entering a complaint dictionary |
Severe severity of disease | a | Entering a complaint dictionary |
204: designing extraction characteristics of a CRFs (conditional random field) characteristic template, introducing the constructed dictionary into the characteristic template, wherein the designed characteristics comprise lexical characteristics, automobile field professional word dictionary characteristics, evaluation characteristic word dictionary (adding relationship dictionary, satisfaction dictionary and complaint dictionary) characteristics, syntactic characteristics and relative position characteristics, and the method is specifically designed as follows:
the lexical characteristics are as follows: the method comprises the steps of word feature and part-of-speech tagging. The word characteristics are word segmentation results such as 'car door', 'rust' and 'severe', the part-of-speech characteristics are part-of-speech tagging results, and the part-of-speech characteristics indicate the role of a word in a sentence. The comment subjects are often nouns or noun phrases, and the comment words are mostly adjectives, verbs, such as "rusty v", "severe a".
The automobile field professional dictionary features: according to the automobile field professional word dictionary constructed in the step 202, if the tested text exists in the automobile field professional word dictionary, the text is marked as "zh", otherwise, the text is marked as "nozh";
evaluating the characteristic dictionary characteristics: according to the added relation dictionary, the satisfaction dictionary and the complaint dictionary constructed in step 203, if the tested texts exist in the three dictionaries, the tested texts are respectively marked as "add", "sa" and "unsa", and otherwise, the tested texts are "no".
Syntactic characteristics: and extracting appropriate dependency relationship characteristic information from the sentence analysis result. Direct dependency relationship often exists between the evaluation object and the evaluation word, and the core idea of analyzing by using the dependency relationship is specifically as follows: firstly, an evaluation object is positioned according to a professional dictionary in the automobile field, a dependency relationship is searched by taking the evaluation object as a center, and a corresponding evaluation word is obtained after screening processing. The evaluation objects are often nouns or noun phrases, and the majority of evaluation words are adjectives or verbs and also have a small part of nouns, so the dependency relationship mainly utilized by the invention is SBV (main and predicate relationship) and ATT (centering relationship).
Relative position characteristics: the distance between the evaluation object and the evaluation word is often relatively close, and the evaluation object often appears in front of or behind the evaluation word. For short and irregular texts, the relative position features are particularly important when the dependency relationship does not exist.
The feature extraction and related description of the CRFs template are as follows:
205: and (3) defining a CRF + + (CRFs realization system) template, converting the training and testing data into a format of a characteristic template required by the CRF + +, training and realizing the joint extraction of evaluation objects and evaluation words by using a training model.
In concrete implementation, an open source tool package CRF + +0.58 with the highest comprehensive performance is selected for training, a training template is defined as the graph shown in FIG. 3, data are converted into a data format required by CRF + +, the features extracted in the step 204 are written into the template in sequence, a training model is obtained by using a data set marked by template training, and finally the trained model is used for testing to complete extraction of comment objects and comment words.
In the embodiment of the present invention, except for the specific description of the model of each device, the model of other devices is not limited, as long as the device can perform the above functions.
Those skilled in the art will appreciate that the drawings are only schematic illustrations of preferred embodiments, and the above-described embodiments of the present invention are merely provided for description and do not represent the merits of the embodiments.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.
Claims (4)
1. A comment object and word joint extraction method for automobile industry user generated content is characterized by comprising the following steps:
acquiring user data, integrating and preprocessing the user data to enable the format and naming mode of the user data to be uniform, and selecting partial data as a training sample set for manual marking;
constructing a professional word dictionary in the automobile field; performing part-of-speech tagging on the preprocessed user data, designing an extraction rule of user data evaluation feature words, and extracting the feature words evaluated by the user to construct an evaluation feature word dictionary;
designing the characteristics of a CRFs template, and introducing the constructed dictionary into the characteristic template;
converting the training and testing data into a format of a feature template required by CRF + +, writing the features into the feature template in sequence, training the CRF + + template to obtain a training model, and testing by using the training model to realize the joint extraction of the evaluation object and the evaluation word of multi-feature fusion.
2. The method for extracting comment objects and words jointly oriented to automobile industry user generated content according to claim 1, wherein the automobile field professional word dictionary comprises:
the method comprises the steps of using an automobile body knowledge base, using harmonic words and mispronounced words in the content generated by a user in the automobile field, and using word2vec training word vectors to extract similar words of product attributes.
3. The method for extracting comment objects and words jointly from user-generated content in the automobile industry according to claim 1, wherein the feature dictionary comprises: and adding a relation dictionary, a satisfaction dictionary and a complaint dictionary.
4. The method for jointly extracting comment objects and words of user-generated content for the automobile industry according to claim 1, wherein the step of introducing the constructed dictionary into the feature template specifically comprises the steps of:
the characteristics comprise lexical characteristics, automobile field professional word dictionary characteristics, evaluation characteristic word dictionary characteristics, dependency relationship characteristics and relative position characteristics between evaluation objects and evaluation words.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911229735.0A CN111144929A (en) | 2019-12-04 | 2019-12-04 | Comment object and word combined extraction method for automobile industry user generated content |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911229735.0A CN111144929A (en) | 2019-12-04 | 2019-12-04 | Comment object and word combined extraction method for automobile industry user generated content |
Publications (1)
Publication Number | Publication Date |
---|---|
CN111144929A true CN111144929A (en) | 2020-05-12 |
Family
ID=70517590
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201911229735.0A Pending CN111144929A (en) | 2019-12-04 | 2019-12-04 | Comment object and word combined extraction method for automobile industry user generated content |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111144929A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111724196A (en) * | 2020-05-14 | 2020-09-29 | 天津大学 | Method for improving quality of automobile product based on user experience |
CN111723162A (en) * | 2020-06-19 | 2020-09-29 | 广州小鹏车联网科技有限公司 | Dictionary processing method, processing device, server and voice interaction system |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104268160A (en) * | 2014-09-05 | 2015-01-07 | 北京理工大学 | Evaluation object extraction method based on domain dictionary and semantic roles |
CN107704556A (en) * | 2017-09-28 | 2018-02-16 | 北京车慧互动广告有限公司 | A kind of sentiment analysis method and system in automobile industry subdivision field |
CN110442728A (en) * | 2019-06-28 | 2019-11-12 | 天津大学 | Sentiment dictionary construction method based on word2vec automobile product field |
-
2019
- 2019-12-04 CN CN201911229735.0A patent/CN111144929A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104268160A (en) * | 2014-09-05 | 2015-01-07 | 北京理工大学 | Evaluation object extraction method based on domain dictionary and semantic roles |
CN107704556A (en) * | 2017-09-28 | 2018-02-16 | 北京车慧互动广告有限公司 | A kind of sentiment analysis method and system in automobile industry subdivision field |
CN110442728A (en) * | 2019-06-28 | 2019-11-12 | 天津大学 | Sentiment dictionary construction method based on word2vec automobile product field |
Non-Patent Citations (1)
Title |
---|
冯仓龙: ""面向商品评价的情感要素抽取"" * |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111724196A (en) * | 2020-05-14 | 2020-09-29 | 天津大学 | Method for improving quality of automobile product based on user experience |
CN111723162A (en) * | 2020-06-19 | 2020-09-29 | 广州小鹏车联网科技有限公司 | Dictionary processing method, processing device, server and voice interaction system |
CN111723162B (en) * | 2020-06-19 | 2023-08-25 | 北京小鹏汽车有限公司 | Dictionary processing method, processing device, server and voice interaction system |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108804521B (en) | Knowledge graph-based question-answering method and agricultural encyclopedia question-answering system | |
CN111145052A (en) | Structured analysis method and system of judicial documents | |
CN111291156B (en) | Knowledge graph-based question and answer intention recognition method | |
CN106649260B (en) | Product characteristic structure tree construction method based on comment text mining | |
Seerat et al. | Opinion Mining: Issues and Challenges(A survey) | |
CN111931506B (en) | Entity relationship extraction method based on graph information enhancement | |
CN109376251A (en) | A kind of microblogging Chinese sentiment dictionary construction method based on term vector learning model | |
CN105512687A (en) | Emotion classification model training and textual emotion polarity analysis method and system | |
CN107562919B (en) | Multi-index integrated software component retrieval method and system based on information retrieval | |
CN113312922B (en) | Improved chapter-level triple information extraction method | |
CN112069312A (en) | Text classification method based on entity recognition and electronic device | |
CN111144929A (en) | Comment object and word combined extraction method for automobile industry user generated content | |
CN107818173B (en) | Vector space model-based Chinese false comment filtering method | |
CN111259645A (en) | Referee document structuring method and device | |
CN112949713A (en) | Text emotion classification method based on ensemble learning of complex network | |
CN112860896A (en) | Corpus generalization method and man-machine conversation emotion analysis method for industrial field | |
CN112800184A (en) | Short text comment emotion analysis method based on Target-Aspect-Opinion joint extraction | |
CN107451116B (en) | Statistical analysis method for mobile application endogenous big data | |
CN111241290A (en) | Comment tag generation method and device and computing equipment | |
CN111737424A (en) | Question matching method, device, equipment and storage medium | |
CN111159342A (en) | Park text comment emotion scoring method based on machine learning | |
CN112685440B (en) | Structural query information expression method for marking search semantic role | |
CN113792542A (en) | Intention understanding method fusing syntactic analysis and semantic role pruning | |
CN113392183A (en) | Characterization and calculation method of children domain map knowledge | |
Velmurugan et al. | Mining implicit and explicit rules for customer data using natural language processing and apriori algorithm |
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 | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20200512 |
|
WD01 | Invention patent application deemed withdrawn after publication |