CN103116644B - Web topic tendentiousness excavates the method with decision support - Google Patents
Web topic tendentiousness excavates the method with decision support Download PDFInfo
- Publication number
- CN103116644B CN103116644B CN201310059170.2A CN201310059170A CN103116644B CN 103116644 B CN103116644 B CN 103116644B CN 201310059170 A CN201310059170 A CN 201310059170A CN 103116644 B CN103116644 B CN 103116644B
- Authority
- CN
- China
- Prior art keywords
- topic
- sub
- information
- cluster
- classification
- 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
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Machine Translation (AREA)
Abstract
The present invention discloses a kind of web topic tendentiousness and excavates the method with decision support, comprises step: S1. network information extraction and storage, by Web Mining technology, on the internet obtaining information, and by result stored in database and local file system; S2. the viewpoint topic detection of information and tracking, utilize thematic comment data, detects and identify the interested viewpoint theme with integrated semantic, and continue follow the tracks of and pay close attention to this viewpoint theme; S3. viewpoint theme emotion tendency identification, carries out the classification of topic emotion tendency to the much-talked-about topic of enterprise, excavates the emotion tendency of viewpoint theme.The present invention, by obtaining relative commercial information from internet, fast and effeciently excavates the theme tendentiousness tendency that enterprise is relevant, realizes instant business wisdom, better for enterprise provides decision support service from mass network information.
Description
Technical field
The present invention relates to the theme tendentiousness excavation of web data and the method for decision support, especially for theme emotional orientation analysis and the decision support of magnanimity web data.
Background technology
Global financial crisis brings effect of depth to many conventional industries, makes industry personage and investor more recognize the importance of believable bussiness imformation and acquiring technology thereof.For enterprise, these technology can assist them to form business decision quickly and effectively, effectively manage risk and control, and improve their commercial competitiveness and finally make them win in market competition.Based on above-mentioned common recognition, the demand of industrial community to Web information mining and intelligent decision service becomes day by day urgent.Web information mining and intelligent decision service relate to technology for information acquisition, Text Classification, text cluster technology, topic identification and tracking technique and text tendency analysis etc.These technology one are to being the field that domestic and international information worker pays close attention to.Text retrieval conference (TREC), information retrieval special interest group meeting (SIGIR), text detection and tracking meeting (TDT) etc. are all the topmost international conference and the forum that show this type of technology newest research results.
Current research person proposes many network text sentiment classification algorithms, mainly concentrates on the text tendency analysis of Sentence-level and chapter level.Current research work can be divided into two kinds of Research Thinkings: based on the method for emotion knowledge and the method for feature based classification.The former mainly relies on some existing sentiment dictionary or domain lexicon, and the combination evaluation unit with feeling polarities in subjective text calculates, and obtains the polarity of subjective text.The latter mainly uses the method for machine learning, chooses a large amount of significant features to complete classification task.These two kinds of Research Thinkings have a lot of representational research work.In the method for feature based classification, the method for machine learning is applied in the emotional semantic classification task of chapter level by Pang first.They attempt employing n-gram word feature and part of speech feature, and compared for NavieBayes, MaxEntropy and SupportVectorMachine(SVM) three kinds of disaggregated models, find that unigram characteristic effect is best.But Cui proves by experiment, when corpus is less time, the effect of unigram is more excellent, but increasing along with corpus, n-gram (n>3) has played more and more important effect.Kim is except investigating traditional n-gram model, and what also introduce position feature and evaluate that word feature carrys out sentence completion level passes judgement on classification.It is a three-layer classification task that Sentence-level emotional semantic classification task is then refined by Zhao, utilize the interaction of class label between each layer, and consider interacting of emotion between upper and lower sentence, and using ConditionalRandomField(CRF) these features merge by model.Be similar to subjective and objective information classification task, the research emphasis of the method for feature based is the discovery of validity feature, and the research of the problem such as feature selecting and Fusion Features.Except passing judgement on except binary classification to subjective text message, some research work are also had to carry out finer emotional semantic classification task.Pang will pass judgement on grade and be divided three classes, and employ one-vs-all multivariate classification algorithm and return sorting algorithm complete emotional semantic classification.Goldberg then employs a kind of half sorting algorithm instructed based on figure, and what complete comment passes judgement on the classification comprising four grades.
In sum, the method at present for the tendentiousness sentiment analysis of enterprise hot spots topic on internet and excavation is also little, still has distance apart from instant business wisdom.Therefore, the method and system that the excavation of a kind of web topic tendentiousness sentiment analysis and decision support are provided is necessary, to make up the deficiencies in the prior art.Topic detection and tracking can find theme and content association relevant for theme together automatically automatically from web data stream, the web topic relevant to enterprise carries out tendentiousness sentiment analysis and excavation, realize instant business wisdom, can better for enterprise provides decision support service.
Summary of the invention
Based on this, for above-mentioned problems of the prior art, the object of the present invention is to provide a kind of web topic tendentiousness to excavate the method with decision support, be intended to the tendentiousness sentiment analysis for enterprise hot spots topic on internet and excavation, the decision-making for enterprise provides reference and support.
For achieving the above object, technical solution of the present invention is:
Web topic tendentiousness excavates the method with decision support, comprises step:
S1. network information extraction and storage, by Web Mining technology, on the internet obtaining information, and by result stored in database and local file system;
S2. the viewpoint topic detection of information and tracking, utilize thematic comment data, detects and identify the interested viewpoint theme with integrated semantic, and continue follow the tracks of and pay close attention to this viewpoint theme;
S3. viewpoint theme emotion tendency identification, carries out the classification of topic emotion tendency to the much-talked-about topic of enterprise, excavates the emotion tendency of viewpoint theme.
Further, described step S1 also comprises:
S11. natural language processing carries out pre-service to raw network information, comprising: Chinese word segmentation, part-of-speech tagging, stop words process, named entity recognition.
Further, in described step S2, the viewpoint topic detection of network information and the process of tracking specifically comprise:
S21. from the information that network collects, through the information classification based on template, filtered noise information;
S22. by the relevant information after filtration, adopt the increment clustering method based on the function of time, realize the detection of sub-topic, and result is stored in database subsystem topic table;
S23. according to the testing result of sub-topic, extract summary and the keyword of sub-topic, and revise sub-topic table relevant information;
S24. according to the information of sub-topic, again according to the increment clustering method of similarity-rough set between window, carry out topic detection, and extract keyword, obtain topic information stored in database;
S25. according to the quantity of information in time of information in topic and topic, find much-talked-about topic, and present to user.
Further, the process of the detection of described step S22 neutron topic specifically comprises:
S221. every section of document in sequential processes information;
S222. hierarchy clustering method is utilized to carry out cluster to untreated document;
If S223. not history of existence cluster, then according to current cluster result, store sub-topic;
If S224. history of existence cluster, then to the sub-topic that the sub-topic of history and new cluster go out, again carry out hierarchical clustering;
S225. by the sub-topic that newly produces stored in database;
S226. the relation of sub-topic and document is upgraded;
S227. the keyword of new generation and updated sub-topic, multi-document summary information is calculated stored in database.
Further, in described step S24, the process of the detection of topic specifically comprises:
S241. sequential processes every sub-topic;
S242. the vector of first sub-topic becomes the cluster centre of first cluster automatically;
If S243. similarity is greater than certain threshold value, then this sub-topic is assigned to this cluster;
S244., when certain cluster distributed in a sheet topic time, the cluster centre of this cluster is recalculated;
If S245. any cluster do not distributed in certain sub-topic, then this sub-topic becomes a new cluster, is also the cluster centre of this cluster simultaneously;
S246. the topic will newly produced, adds database to;
S247. the information of topic is upgraded.
Further, in described step S3, the process of network themes emotion tendency identification specifically comprises:
S31. train topic sentiment classification model, read the topic language material and sentiment dictionary that have marked, utilize svm classifier algorithm, obtain topic sentiment classification model by training;
S32. sub-topic emotional semantic classification, antithetical phrase topic extracts affective characteristics, utilizes topic sentiment classification model and svm classifier algorithm to obtain sub-topic emotional semantic classification result;
S33. topic emotional semantic classification, utilizes the result of sub-topic emotional semantic classification, builds the graph model based on sub-topic, exports topic emotional semantic classification result according to graph model;
Further, the process of topic sentiment classification model is trained specifically to comprise in described step S31:
S311. the topic language material marked is read in;
S312. by natural language processing, obtain through Chinese word segmentation and the good topic language material of part-of-speech tagging;
S313. according to sentiment dictionary and grammatical pattern storehouse, from topic language material, affective characteristics is extracted, structure topic classification based training data set;
S314. sorter reads training dataset, utilizes svm classifier algorithm, obtains topic sentiment classification model by training.
Further, the process of described step S32 neutron topic emotional semantic classification specifically comprises:
S321. sub-topic to be sorted is read in;
S322. by natural language processing, obtain through Chinese word segmentation and the good sub-topic of part-of-speech tagging;
S323. according to sentiment dictionary and grammatical pattern storehouse, from sub-topic, affective characteristics is extracted, structure test data set;
S324. sorter read test data and the topic sentiment classification model that trains before, utilize svm classifier algorithm, export sub-topic emotional semantic classification result.
Further, in described step S33, the process of topic emotional semantic classification specifically comprises:
S331. topic to be sorted is read in;
S332. topic to be sorted is resolved, obtain sub-topic collection;
S333. call sub-topic emotion classifiers, every sub-topic is classified, obtains sub-topic emotional semantic classification result;
S334. according to the similarity between sub-topic, build LexRank graph model, the graph model constructed by utilization, calculates importance and the redundance of sub-topic, final output topic emotional semantic classification result.
Compared with prior art, the present invention has following beneficial effect: the present invention obtains relative commercial information by Web Mining and information extraction technique from internet, bussiness imformation is analyzed, find new topic, and continue follow the tracks of and pay close attention to this topic, by the emotion tendency obtaining topic of topic and emotion trend.The present invention fast and effeciently can excavate the relevant theme tendentiousness tendency of enterprise from mass network information, realizes instant business wisdom, can better for enterprise provides decision support service.
Accompanying drawing explanation
Fig. 1 is embodiments of the invention one schematic flow sheets.
Fig. 2 is embodiments of the invention two schematic flow sheets.
Embodiment
Below in conjunction with drawings and Examples, the present invention is further detailed explanation.
Embodiment one
The schematic flow sheet of the embodiment of the present invention one has been shown in Fig. 1.
As shown in Figure 1, in this embodiment, a kind of web topic tendentiousness excavates the method with decision support, comprises step:
S101. network information extraction and storage, by Web Mining technology, on the internet obtaining information, and by result stored in database and local file system;
S102. natural language processing carries out pre-service to raw network information, comprising: Chinese word segmentation, part-of-speech tagging, stop words process, named entity recognition;
S103. the viewpoint topic detection of information and tracking, utilize thematic comment data, detects and identify the interested viewpoint theme with integrated semantic.And continue follow the tracks of and pay close attention to this viewpoint theme;
S104. viewpoint theme emotion tendency identification, carries out the classification of topic emotion tendency to the much-talked-about topic of enterprise, excavates the emotion tendency of viewpoint theme.
Embodiment two
The schematic flow sheet of the embodiment of the present invention two has been shown in Fig. 2.
As shown in Figure 2, in this embodiment,
Web topic tendentiousness excavates the method with decision support, comprises step:
S201. network information extraction and storage, by Web Mining technology, on the internet obtaining information, and by result stored in database and local file system;
S202. natural language processing carries out pre-service to raw network information, comprising: Chinese word segmentation, part-of-speech tagging, stop words process, named entity recognition;
S203. the information will collected from network, through the information classification based on template, filtered noise information;
S204. by the relevant information after filtration, adopt the increment clustering method based on the function of time, realize the detection of sub-topic, and result is stored in database subsystem topic table;
S205. according to the testing result of sub-topic, extract summary and the keyword of sub-topic, and revise sub-topic table relevant information;
S206. according to the information of sub-topic, again according to the increment clustering method of similarity-rough set between window, carry out topic detection, and extract keyword, obtain topic information stored in database;
S207. according to the quantity of information in time of information in topic and topic, find much-talked-about topic, and present to user;
S208. train topic sentiment classification model, read the topic language material and sentiment dictionary that have marked, utilize svm classifier algorithm, obtain topic sentiment classification model by training;
S209. sub-topic emotional semantic classification, antithetical phrase topic extracts affective characteristics, utilizes topic sentiment classification model and svm classifier algorithm to obtain sub-topic classification results;
S210. topic emotional semantic classification, utilizes the result of sub-topic emotional semantic classification, builds the graph model based on sub-topic, exports topic emotional semantic classification result according to graph model.
Embodiment three
Web topic tendentiousness excavates the method with decision support, comprises step:
S301. network information extraction and storage, by Web Mining technology, on the internet obtaining information, and by result stored in database and local file system;
S302. natural language processing carries out pre-service to raw network information, comprising: Chinese word segmentation, part-of-speech tagging, stop words process, named entity recognition;
S303. from the information that network collects, through the information classification based on template, filtered noise information;
S304. every section of document in sequential processes information;
S305. hierarchy clustering method is utilized to carry out cluster to untreated document;
If S306. not history of existence cluster, then according to current cluster result, store sub-topic;
If S307. history of existence cluster, then to the sub-topic that the sub-topic of history and new cluster go out, again carry out hierarchical clustering;
S308. by the sub-topic that newly produces stored in database;
S309. the relation of sub-topic and document is upgraded;
S310. the information such as keyword, multi-document summary of new generation and updated sub-topic that calculates is stored in database;
S311. according to the testing result of sub-topic, extract summary and the keyword of sub-topic, and revise sub-topic table relevant information;
S312. sequential processes every sub-topic;
S313. the vector of first sub-topic becomes the cluster centre of first cluster automatically;
If S314. similarity is greater than certain threshold value, then this sub-topic is assigned to this cluster;
S315., when certain cluster distributed in a sheet topic time, the cluster centre of this cluster is recalculated;
If S316. any cluster do not distributed in certain sub-topic, then this sub-topic becomes a new cluster, is also the cluster centre of this cluster simultaneously;
S317. the topic will newly produced, adds database to;
S318. the information of topic is upgraded;
S319. according to the quantity of information in time of information in topic and topic, find much-talked-about topic, and present to user;
S320. the topic language material marked is read in;
S321. by natural language processing, obtain through Chinese word segmentation and the good topic language material of part-of-speech tagging;
S322. according to sentiment dictionary and grammatical pattern storehouse, from topic language material, affective characteristics is extracted, structure topic classification based training data set;
S323. sorter reads training dataset, utilizes svm classifier algorithm, obtains topic sentiment classification model by training;
S324. sub-topic to be sorted is read in;
S325. by natural language processing, obtain through Chinese word segmentation and the good sub-topic of part-of-speech tagging;
S326. according to sentiment dictionary and grammatical pattern storehouse, from sub-topic, affective characteristics is extracted, structure test data set;
S327 sorter read test data and the topic sentiment classification model trained before, utilize svm classifier algorithm, exports sub-topic emotional semantic classification result;
S328. topic to be sorted is read in;
S329. topic to be sorted is resolved, obtain sub-topic collection;
S330. call sub-topic emotion classifiers, every sub-topic is classified, obtains sub-topic emotional semantic classification result;
S331. according to the similarity between sub-topic, build LexRank graph model, the graph model constructed by utilization, calculates importance and the redundance of sub-topic, exports topic emotional semantic classification result.
As adopted reptile to be responsible for from targeted website downloading web pages internet, and resolve and information extraction webpage, result is stored in database and local file system.Adopt focused crawler, filter and irrelevant the linking of theme according to certain web page analysis algorithm, the link remained with also puts it into the URL queue waited for and capturing.Then, the webpage URL that it will select next step to capture according to certain search strategy from queue, and repeat said process, until stop when reaching a certain condition of system.In addition, allly will to be stored by system by the webpage of crawler capturing, carry out certain analysis, filtration, and set up index, so that retrieval and indexing afterwards.
In sub-topic detection and topic detection, concrete clustering method is as follows:
First pre-service is carried out to text, then extract and select speech feature thus reasonable representation speech, finally carry out topic cluster according to speech characteristic sum topic feature calculation similarity.After carrying out topic cluster, then upgrade topic feature.First, each speech is regarded as a topic only containing a speech, and calculate the similarity of each speech team.Secondly, the similarity of each class bunch is calculated.The similarity of class bunch A and class bunch B can be regarded as the arithmetic average of the similarity of the speech team in each class bunch.Finally, suppose that A and B is that class that similarity is the highest is bunch right, if similarity is higher than the threshold value preset, then class bunch A, B is merged into a new class bunch, and continues to perform second step, otherwise stop topic cluster.
These are only the preferred embodiments of the present invention, but design concept of the present invention is not limited thereto, all insubstantial modifications utilizing this design to make the present invention, also all fall within protection scope of the present invention.
Claims (4)
1. web topic tendentiousness excavates the method with decision support, it is characterized in that, comprises step:
S1. network information extraction and storage, by Web Mining technology, on the internet obtaining information, and by result stored in database and local file system;
S2. the viewpoint topic detection of information and tracking, utilize thematic comment data, detects and identify the interested viewpoint theme with integrated semantic, and continue follow the tracks of and pay close attention to this viewpoint theme;
S3. viewpoint theme emotion tendency identification, carries out the classification of topic emotion tendency to the much-talked-about topic of enterprise, excavates the emotion tendency of viewpoint theme;
Described step S1 also comprises:
S11. natural language processing carries out pre-service to raw network information, comprising: Chinese word segmentation, part-of-speech tagging, stop words process, named entity recognition;
In described step S2, the process of viewpoint topic detection and tracking specifically comprises:
S21. from the information that network collects, through the information classification based on template, filtered noise information;
S22. by the relevant information after filtration, adopt the increment clustering method based on the function of time, realize the detection of sub-topic, and result is stored in database subsystem topic table;
S23. according to the testing result of sub-topic, extract summary and the keyword of sub-topic, and revise sub-topic table relevant information;
S24. according to the information of sub-topic, again according to the increment clustering method of similarity-rough set between window, carry out topic detection, and extract keyword, obtain topic information stored in database;
S25. according to the quantity of information in time of information in topic and topic, find much-talked-about topic, and present to user;
The process of the detection of described step S22 neutron topic specifically comprises:
S221. every section of document in sequential processes relevant information;
S222. hierarchy clustering method is utilized to carry out cluster to untreated document;
If S223. not history of existence cluster, then according to current cluster result, store sub-topic;
If S224. history of existence cluster, then to the sub-topic that the sub-topic of history and new cluster go out, again carry out hierarchical clustering;
S225. by the sub-topic that newly produces stored in database;
S226. the relation of sub-topic and document is upgraded;
S227. the keyword of new generation and updated sub-topic, multi-document summary information is calculated stored in database;
In described step S24, the process of the detection of topic specifically comprises:
S241. sequential processes every sub-topic;
S242. the vector of first sub-topic becomes the cluster centre of first cluster automatically;
If S243. similarity is greater than certain threshold value, then this sub-topic is assigned to this cluster;
S244., when certain cluster distributed in a sheet topic time, the cluster centre of this cluster is recalculated;
If S245. any cluster do not distributed in certain sub-topic, then this sub-topic becomes a new cluster, is also the cluster centre of this cluster simultaneously;
S246. the topic will newly produced, adds database to;
S247. the information of topic is upgraded;
In described step S3, the process of network themes emotion tendency identification specifically comprises:
S31. train topic sentiment classification model, read the topic language material and sentiment dictionary that have marked, utilize svm classifier algorithm, obtain topic sentiment classification model by training;
S32. sub-topic emotional semantic classification, antithetical phrase topic extracts affective characteristics, utilizes topic sentiment classification model and svm classifier algorithm to obtain sub-topic emotional semantic classification result;
S33. topic emotional semantic classification, utilizes the result of sub-topic emotional semantic classification, builds the graph model based on sub-topic, exports topic emotional semantic classification result according to graph model.
2. web topic tendentiousness according to claim 1 excavates the method with decision support, it is characterized in that, trains the process of topic sentiment classification model specifically to comprise in described step S31:
S311. the topic language material marked is read in;
S312. by natural language processing, obtain through Chinese word segmentation and the good topic language material of part-of-speech tagging;
S313. according to sentiment dictionary and grammatical pattern storehouse, from topic language material, affective characteristics is extracted, structure topic classification based training data set;
S314. sorter reads training dataset, utilizes svm classifier algorithm, obtains topic sentiment classification model by training.
3. web topic tendentiousness according to claim 1 excavates the method with decision support, and it is characterized in that, the process of described step S32 neutron topic emotional semantic classification specifically comprises:
S321. sub-topic to be sorted is read in;
S322. by natural language processing, obtain through Chinese word segmentation and the good sub-topic of part-of-speech tagging;
S323. according to sentiment dictionary and grammatical pattern storehouse, from sub-topic, affective characteristics is extracted, structure test data set;
S324. sorter read test data and the topic sentiment classification model that trains before, utilize svm classifier algorithm, export sub-topic emotional semantic classification result.
4. web topic tendentiousness according to claim 1 excavates the method with decision support, and it is characterized in that, in described step S33, the process of topic emotional semantic classification specifically comprises:
S331. topic to be sorted is read in;
S332. topic to be sorted is resolved, obtain sub-topic collection;
S333. call sub-topic emotion classifiers, every sub-topic is classified, obtains sub-topic emotional semantic classification result;
S334. according to the similarity between sub-topic, build LexRank graph model, the graph model constructed by utilization, calculates importance and the redundance of sub-topic, final output topic emotional semantic classification result.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201310059170.2A CN103116644B (en) | 2013-02-26 | 2013-02-26 | Web topic tendentiousness excavates the method with decision support |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201310059170.2A CN103116644B (en) | 2013-02-26 | 2013-02-26 | Web topic tendentiousness excavates the method with decision support |
Publications (2)
Publication Number | Publication Date |
---|---|
CN103116644A CN103116644A (en) | 2013-05-22 |
CN103116644B true CN103116644B (en) | 2016-04-13 |
Family
ID=48415017
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201310059170.2A Active CN103116644B (en) | 2013-02-26 | 2013-02-26 | Web topic tendentiousness excavates the method with decision support |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN103116644B (en) |
Families Citing this family (16)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103473280B (en) * | 2013-08-28 | 2017-02-08 | 中国科学院合肥物质科学研究院 | Method for mining comparable network language materials |
CN103841121B (en) * | 2014-03-28 | 2017-03-29 | 中国科学技术大学 | A kind of comment and interaction systems and method based on local file |
CN104123395A (en) * | 2014-08-13 | 2014-10-29 | 北京赛科世纪数码科技有限公司 | Decision making method and system based on big data |
CN104331394A (en) * | 2014-08-29 | 2015-02-04 | 南通大学 | Text classification method based on viewpoint |
CN104504087A (en) * | 2014-12-25 | 2015-04-08 | 中国科学院电子学研究所 | Low-rank decomposition based delicate topic mining method |
CN104794212B (en) * | 2015-04-27 | 2018-04-10 | 清华大学 | Context sensibility classification method and categorizing system based on user comment text |
CN104965823A (en) * | 2015-07-30 | 2015-10-07 | 成都鼎智汇科技有限公司 | Big data based opinion extraction method |
CN105159972A (en) * | 2015-08-26 | 2015-12-16 | 苏州大学张家港工业技术研究院 | Classification method and system for evaluation types |
CN105787026B (en) * | 2016-02-24 | 2019-07-09 | 人民网股份有限公司 | The display methods and device of information flow |
CN106126502B (en) * | 2016-07-07 | 2018-10-30 | 四川长虹电器股份有限公司 | A kind of emotional semantic classification system and method based on support vector machines |
CN106528538A (en) * | 2016-12-07 | 2017-03-22 | 竹间智能科技(上海)有限公司 | Method and device for intelligent emotion recognition |
CN106855879A (en) * | 2016-12-14 | 2017-06-16 | 竹间智能科技(上海)有限公司 | The robot that artificial intelligence psychology is seeked advice from music |
CN109902230A (en) * | 2019-02-13 | 2019-06-18 | 北京航空航天大学 | A kind of processing method and processing device of news data |
CN112231470A (en) * | 2019-06-28 | 2021-01-15 | 上海智臻智能网络科技股份有限公司 | Topic mining method and device, storage medium and terminal |
CN111428510B (en) * | 2020-03-10 | 2023-04-07 | 蚌埠学院 | Public praise-based P2P platform risk analysis method |
CN114218381B (en) * | 2021-12-08 | 2022-08-30 | 北京中科闻歌科技股份有限公司 | Method, device, equipment and medium for identifying position |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102073647A (en) * | 2009-11-23 | 2011-05-25 | 北京科技大学 | E-Science environment-oriented multi-domain Web text feature extracting system and method |
CN102646114A (en) * | 2012-02-17 | 2012-08-22 | 清华大学 | News topic timeline abstract generating method based on breakthrough point |
-
2013
- 2013-02-26 CN CN201310059170.2A patent/CN103116644B/en active Active
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102073647A (en) * | 2009-11-23 | 2011-05-25 | 北京科技大学 | E-Science environment-oriented multi-domain Web text feature extracting system and method |
CN102646114A (en) * | 2012-02-17 | 2012-08-22 | 清华大学 | News topic timeline abstract generating method based on breakthrough point |
Non-Patent Citations (1)
Title |
---|
基于话题检测和情感分析的互联网热点分析与监控技术研究;吴泽衡;《中国优秀硕士学位论文全文数据库》;20111215(第12期);第15-66页 * |
Also Published As
Publication number | Publication date |
---|---|
CN103116644A (en) | 2013-05-22 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN103116644B (en) | Web topic tendentiousness excavates the method with decision support | |
CN104820629B (en) | A kind of intelligent public sentiment accident emergent treatment system and method | |
Arulanandam et al. | Extracting crime information from online newspaper articles | |
CN102708096B (en) | Network intelligence public sentiment monitoring system based on semantics and work method thereof | |
CN102207948B (en) | Method for generating incident statement sentence material base | |
Venugopalan et al. | Exploring sentiment analysis on twitter data | |
CN101609450A (en) | Web page classification method based on training set | |
CN104268200A (en) | Unsupervised named entity semantic disambiguation method based on deep learning | |
CN112650848A (en) | Urban railway public opinion information analysis method based on text semantic related passenger evaluation | |
CN104268197A (en) | Industry comment data fine grain sentiment analysis method | |
CN104408093A (en) | News event element extracting method and device | |
Chawla et al. | Product opinion mining using sentiment analysis on smartphone reviews | |
CN105608200A (en) | Network public opinion tendency prediction analysis method | |
CN103544255A (en) | Text semantic relativity based network public opinion information analysis method | |
CN103077164A (en) | Text analysis method and text analyzer | |
CN102937960A (en) | Device and method for identifying and evaluating emergency hot topic | |
CN101114298A (en) | Method for gaining oral vocabulary entry, device and input method system thereof | |
CN101609459A (en) | A kind of extraction system of affective characteristic words | |
CN107943909A (en) | User demand trend method for digging and device, storage medium based on comment data | |
CN101393555A (en) | Rubbish blog detecting method | |
CN103324700A (en) | Noumenon concept attribute learning method based on Web information | |
CN111914087A (en) | Public opinion analysis method | |
CN101968801A (en) | Method for extracting key words of single text | |
CN102073641A (en) | Method, device and program for processing consumer-generated media information | |
CN102508830A (en) | Method and system for extracting social network from news document |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant |