TASC:Topic-Adaptive Sentiment Classification on Dynamic Tweets
Sentiment classification is a topic-sensitive task, i.e., a classifier trained from one topic will perform worse on another. This is especially a problem for the tweets sentiment analysis. Since the topics in Twitter are very diverse, it is impossible to train a universal classifier for all topics....
Ausführliche Beschreibung
Gespeichert in:
| Autor*in: |
Shenghua Liu [verfasserIn] |
|---|
| Format: |
Artikel |
|---|---|
| Sprache: |
Englisch |
| Erschienen: |
2015 |
|---|
| Schlagwörter: |
learning (artificial intelligence) semisupervised learning methods |
|---|
| Systematik: |
|
|---|
| Übergeordnetes Werk: |
Enthalten in: IEEE transactions on knowledge and data engineering - New York, NY : IEEE, 1989, 27(2015), 6, Seite 1696-1709 |
|---|---|
| Übergeordnetes Werk: |
volume:27 ; year:2015 ; number:6 ; pages:1696-1709 |
| Links: |
|---|
| DOI / URN: |
10.1109/TKDE.2014.2382600 |
|---|
| Katalog-ID: |
OLC1958213837 |
|---|
| LEADER | 01000caa a2200265 4500 | ||
|---|---|---|---|
| 001 | OLC1958213837 | ||
| 003 | DE-627 | ||
| 005 | 20230714144306.0 | ||
| 007 | tu | ||
| 008 | 160206s2015 xx ||||| 00| ||eng c | ||
| 024 | 7 | |a 10.1109/TKDE.2014.2382600 |2 doi | |
| 028 | 5 | 2 | |a PQ20160617 |
| 035 | |a (DE-627)OLC1958213837 | ||
| 035 | |a (DE-599)GBVOLC1958213837 | ||
| 035 | |a (PRQ)c2119-bb3b284f9fc675eb1420da6ebed2aa58c7d2a180319a1ffa14762a17bd5720070 | ||
| 035 | |a (KEY)0175400920150000027000601696tasctopicadaptivesentimentclassificationondynamict | ||
| 040 | |a DE-627 |b ger |c DE-627 |e rakwb | ||
| 041 | |a eng | ||
| 082 | 0 | 4 | |a 620 |a 530 |a 000 |q DNB |
| 084 | |a SA 5571 |q AVZ |2 rvk | ||
| 084 | |a 54.00 |2 bkl | ||
| 100 | 0 | |a Shenghua Liu |e verfasserin |4 aut | |
| 245 | 1 | 0 | |a TASC:Topic-Adaptive Sentiment Classification on Dynamic Tweets |
| 264 | 1 | |c 2015 | |
| 336 | |a Text |b txt |2 rdacontent | ||
| 337 | |a ohne Hilfsmittel zu benutzen |b n |2 rdamedia | ||
| 338 | |a Band |b nc |2 rdacarrier | ||
| 520 | |a Sentiment classification is a topic-sensitive task, i.e., a classifier trained from one topic will perform worse on another. This is especially a problem for the tweets sentiment analysis. Since the topics in Twitter are very diverse, it is impossible to train a universal classifier for all topics. Moreover, compared to product review, Twitter lacks data labeling and a rating mechanism to acquire sentiment labels. The extremely sparse text of tweets also brings down the performance of a sentiment classifier. In this paper, we propose a semi-supervised topic-adaptive sentiment classification (TASC) model, which starts with a classifier built on common features and mixed labeled data from various topics. It minimizes the hinge loss to adapt to unlabeled data and features including topic-related sentiment words, authors' sentiments and sentiment connections derived from"" mentions of tweets, named as topic-adaptive features. Text and non-text features are extracted and naturally split into two views for co-training. The TASC learning algorithm updates topic-adaptive features based on the collaborative selection of unlabeled data, which in turn helps to select more reliable tweets to boost the performance. We also design the adapting model along a timeline (TASC-t) for dynamic tweets. An experiment on 6 topics from published tweet corpuses demonstrates that TASC outperforms other well-known supervised and ensemble classifiers. It also beats those semi-supervised learning methods without feature adaption. Meanwhile, TASC-t can also achieve impressive accuracy and F-score. Finally, with timeline visualization of "river" graph, people can intuitively grasp the ups and downs of sentiments' evolvement, and the intensity by color gradation. | ||
| 650 | 4 | |a learning (artificial intelligence) | |
| 650 | 4 | |a Adaptation models | |
| 650 | 4 | |a collaborative selection | |
| 650 | 4 | |a semisupervised learning methods | |
| 650 | 4 | |a F-score | |
| 650 | 4 | |a sparse text | |
| 650 | 4 | |a tweets sentiment analysis | |
| 650 | 4 | |a Support vector machines | |
| 650 | 4 | |a adaptive feature | |
| 650 | 4 | |a sentiment labels | |
| 650 | 4 | |a topic-adaptive features | |
| 650 | 4 | |a timeline visualization | |
| 650 | 4 | |a TASC learning algorithm | |
| 650 | 4 | |a Data visualization | |
| 650 | 4 | |a topic-related sentiment words | |
| 650 | 4 | |a TASC model | |
| 650 | 4 | |a semisupervised topic-adaptive sentiment classification | |
| 650 | 4 | |a sentiment connections | |
| 650 | 4 | |a multiclass SVM | |
| 650 | 4 | |a feature extraction | |
| 650 | 4 | |a Sentiment analysis | |
| 650 | 4 | |a pattern classification | |
| 650 | 4 | |a social networking (online) | |
| 650 | 4 | |a TASC-t | |
| 650 | 4 | |a ensemble classifiers | |
| 650 | 4 | |a topic-sensitive task | |
| 650 | 4 | |a color gradation intensity | |
| 650 | 4 | |a supervised classifiers | |
| 650 | 4 | |a universal classifier | |
| 650 | 4 | |a dynamic tweets | |
| 650 | 4 | |a text analysis | |
| 650 | 4 | |a mixed labeled data | |
| 650 | 4 | |a nontext features extraction | |
| 650 | 4 | |a cross-domain | |
| 650 | 4 | |a Google | |
| 650 | 4 | |a river graph | |
| 650 | 4 | |a topic-adaptive | |
| 650 | 4 | |a unlabeled data | |
| 650 | 4 | |a sentiment classification | |
| 650 | 4 | |a Twitter | |
| 650 | 4 | |a Product reviews | |
| 700 | 0 | |a Xueqi Cheng |4 oth | |
| 700 | 0 | |a Fuxin Li |4 oth | |
| 700 | 0 | |a Fangtao Li |4 oth | |
| 773 | 0 | 8 | |i Enthalten in |t IEEE transactions on knowledge and data engineering |d New York, NY : IEEE, 1989 |g 27(2015), 6, Seite 1696-1709 |w (DE-627)130765732 |w (DE-600)1001468-8 |w (DE-576)023036060 |x 1041-4347 |7 nnns |
| 773 | 1 | 8 | |g volume:27 |g year:2015 |g number:6 |g pages:1696-1709 |
| 856 | 4 | 1 | |u http://dx.doi.org/10.1109/TKDE.2014.2382600 |3 Volltext |
| 856 | 4 | 2 | |u http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=6990628 |
| 856 | 4 | 2 | |u http://search.proquest.com/docview/1677626558 |
| 912 | |a GBV_USEFLAG_A | ||
| 912 | |a SYSFLAG_A | ||
| 912 | |a GBV_OLC | ||
| 912 | |a SSG-OLC-TEC | ||
| 912 | |a SSG-OLC-PHY | ||
| 912 | |a SSG-OLC-MAT | ||
| 912 | |a SSG-OPC-BBI | ||
| 936 | r | v | |a SA 5571 |
| 936 | b | k | |a 54.00 |q AVZ |
| 951 | |a AR | ||
| 952 | |d 27 |j 2015 |e 6 |h 1696-1709 | ||
| author_variant |
s l sl |
|---|---|
| matchkey_str |
article:10414347:2015----::acoiaatvsnietlsiiai |
| hierarchy_sort_str |
2015 |
| bklnumber |
54.00 |
| publishDate |
2015 |
| allfields |
10.1109/TKDE.2014.2382600 doi PQ20160617 (DE-627)OLC1958213837 (DE-599)GBVOLC1958213837 (PRQ)c2119-bb3b284f9fc675eb1420da6ebed2aa58c7d2a180319a1ffa14762a17bd5720070 (KEY)0175400920150000027000601696tasctopicadaptivesentimentclassificationondynamict DE-627 ger DE-627 rakwb eng 620 530 000 DNB SA 5571 AVZ rvk 54.00 bkl Shenghua Liu verfasserin aut TASC:Topic-Adaptive Sentiment Classification on Dynamic Tweets 2015 Text txt rdacontent ohne Hilfsmittel zu benutzen n rdamedia Band nc rdacarrier Sentiment classification is a topic-sensitive task, i.e., a classifier trained from one topic will perform worse on another. This is especially a problem for the tweets sentiment analysis. Since the topics in Twitter are very diverse, it is impossible to train a universal classifier for all topics. Moreover, compared to product review, Twitter lacks data labeling and a rating mechanism to acquire sentiment labels. The extremely sparse text of tweets also brings down the performance of a sentiment classifier. In this paper, we propose a semi-supervised topic-adaptive sentiment classification (TASC) model, which starts with a classifier built on common features and mixed labeled data from various topics. It minimizes the hinge loss to adapt to unlabeled data and features including topic-related sentiment words, authors' sentiments and sentiment connections derived from"" mentions of tweets, named as topic-adaptive features. Text and non-text features are extracted and naturally split into two views for co-training. The TASC learning algorithm updates topic-adaptive features based on the collaborative selection of unlabeled data, which in turn helps to select more reliable tweets to boost the performance. We also design the adapting model along a timeline (TASC-t) for dynamic tweets. An experiment on 6 topics from published tweet corpuses demonstrates that TASC outperforms other well-known supervised and ensemble classifiers. It also beats those semi-supervised learning methods without feature adaption. Meanwhile, TASC-t can also achieve impressive accuracy and F-score. Finally, with timeline visualization of "river" graph, people can intuitively grasp the ups and downs of sentiments' evolvement, and the intensity by color gradation. learning (artificial intelligence) Adaptation models collaborative selection semisupervised learning methods F-score sparse text tweets sentiment analysis Support vector machines adaptive feature sentiment labels topic-adaptive features timeline visualization TASC learning algorithm Data visualization topic-related sentiment words TASC model semisupervised topic-adaptive sentiment classification sentiment connections multiclass SVM feature extraction Sentiment analysis pattern classification social networking (online) TASC-t ensemble classifiers topic-sensitive task color gradation intensity supervised classifiers universal classifier dynamic tweets text analysis mixed labeled data nontext features extraction cross-domain Google river graph topic-adaptive unlabeled data sentiment classification Twitter Product reviews Xueqi Cheng oth Fuxin Li oth Fangtao Li oth Enthalten in IEEE transactions on knowledge and data engineering New York, NY : IEEE, 1989 27(2015), 6, Seite 1696-1709 (DE-627)130765732 (DE-600)1001468-8 (DE-576)023036060 1041-4347 nnns volume:27 year:2015 number:6 pages:1696-1709 http://dx.doi.org/10.1109/TKDE.2014.2382600 Volltext http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=6990628 http://search.proquest.com/docview/1677626558 GBV_USEFLAG_A SYSFLAG_A GBV_OLC SSG-OLC-TEC SSG-OLC-PHY SSG-OLC-MAT SSG-OPC-BBI SA 5571 54.00 AVZ AR 27 2015 6 1696-1709 |
| spelling |
10.1109/TKDE.2014.2382600 doi PQ20160617 (DE-627)OLC1958213837 (DE-599)GBVOLC1958213837 (PRQ)c2119-bb3b284f9fc675eb1420da6ebed2aa58c7d2a180319a1ffa14762a17bd5720070 (KEY)0175400920150000027000601696tasctopicadaptivesentimentclassificationondynamict DE-627 ger DE-627 rakwb eng 620 530 000 DNB SA 5571 AVZ rvk 54.00 bkl Shenghua Liu verfasserin aut TASC:Topic-Adaptive Sentiment Classification on Dynamic Tweets 2015 Text txt rdacontent ohne Hilfsmittel zu benutzen n rdamedia Band nc rdacarrier Sentiment classification is a topic-sensitive task, i.e., a classifier trained from one topic will perform worse on another. This is especially a problem for the tweets sentiment analysis. Since the topics in Twitter are very diverse, it is impossible to train a universal classifier for all topics. Moreover, compared to product review, Twitter lacks data labeling and a rating mechanism to acquire sentiment labels. The extremely sparse text of tweets also brings down the performance of a sentiment classifier. In this paper, we propose a semi-supervised topic-adaptive sentiment classification (TASC) model, which starts with a classifier built on common features and mixed labeled data from various topics. It minimizes the hinge loss to adapt to unlabeled data and features including topic-related sentiment words, authors' sentiments and sentiment connections derived from"" mentions of tweets, named as topic-adaptive features. Text and non-text features are extracted and naturally split into two views for co-training. The TASC learning algorithm updates topic-adaptive features based on the collaborative selection of unlabeled data, which in turn helps to select more reliable tweets to boost the performance. We also design the adapting model along a timeline (TASC-t) for dynamic tweets. An experiment on 6 topics from published tweet corpuses demonstrates that TASC outperforms other well-known supervised and ensemble classifiers. It also beats those semi-supervised learning methods without feature adaption. Meanwhile, TASC-t can also achieve impressive accuracy and F-score. Finally, with timeline visualization of "river" graph, people can intuitively grasp the ups and downs of sentiments' evolvement, and the intensity by color gradation. learning (artificial intelligence) Adaptation models collaborative selection semisupervised learning methods F-score sparse text tweets sentiment analysis Support vector machines adaptive feature sentiment labels topic-adaptive features timeline visualization TASC learning algorithm Data visualization topic-related sentiment words TASC model semisupervised topic-adaptive sentiment classification sentiment connections multiclass SVM feature extraction Sentiment analysis pattern classification social networking (online) TASC-t ensemble classifiers topic-sensitive task color gradation intensity supervised classifiers universal classifier dynamic tweets text analysis mixed labeled data nontext features extraction cross-domain Google river graph topic-adaptive unlabeled data sentiment classification Twitter Product reviews Xueqi Cheng oth Fuxin Li oth Fangtao Li oth Enthalten in IEEE transactions on knowledge and data engineering New York, NY : IEEE, 1989 27(2015), 6, Seite 1696-1709 (DE-627)130765732 (DE-600)1001468-8 (DE-576)023036060 1041-4347 nnns volume:27 year:2015 number:6 pages:1696-1709 http://dx.doi.org/10.1109/TKDE.2014.2382600 Volltext http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=6990628 http://search.proquest.com/docview/1677626558 GBV_USEFLAG_A SYSFLAG_A GBV_OLC SSG-OLC-TEC SSG-OLC-PHY SSG-OLC-MAT SSG-OPC-BBI SA 5571 54.00 AVZ AR 27 2015 6 1696-1709 |
| allfields_unstemmed |
10.1109/TKDE.2014.2382600 doi PQ20160617 (DE-627)OLC1958213837 (DE-599)GBVOLC1958213837 (PRQ)c2119-bb3b284f9fc675eb1420da6ebed2aa58c7d2a180319a1ffa14762a17bd5720070 (KEY)0175400920150000027000601696tasctopicadaptivesentimentclassificationondynamict DE-627 ger DE-627 rakwb eng 620 530 000 DNB SA 5571 AVZ rvk 54.00 bkl Shenghua Liu verfasserin aut TASC:Topic-Adaptive Sentiment Classification on Dynamic Tweets 2015 Text txt rdacontent ohne Hilfsmittel zu benutzen n rdamedia Band nc rdacarrier Sentiment classification is a topic-sensitive task, i.e., a classifier trained from one topic will perform worse on another. This is especially a problem for the tweets sentiment analysis. Since the topics in Twitter are very diverse, it is impossible to train a universal classifier for all topics. Moreover, compared to product review, Twitter lacks data labeling and a rating mechanism to acquire sentiment labels. The extremely sparse text of tweets also brings down the performance of a sentiment classifier. In this paper, we propose a semi-supervised topic-adaptive sentiment classification (TASC) model, which starts with a classifier built on common features and mixed labeled data from various topics. It minimizes the hinge loss to adapt to unlabeled data and features including topic-related sentiment words, authors' sentiments and sentiment connections derived from"" mentions of tweets, named as topic-adaptive features. Text and non-text features are extracted and naturally split into two views for co-training. The TASC learning algorithm updates topic-adaptive features based on the collaborative selection of unlabeled data, which in turn helps to select more reliable tweets to boost the performance. We also design the adapting model along a timeline (TASC-t) for dynamic tweets. An experiment on 6 topics from published tweet corpuses demonstrates that TASC outperforms other well-known supervised and ensemble classifiers. It also beats those semi-supervised learning methods without feature adaption. Meanwhile, TASC-t can also achieve impressive accuracy and F-score. Finally, with timeline visualization of "river" graph, people can intuitively grasp the ups and downs of sentiments' evolvement, and the intensity by color gradation. learning (artificial intelligence) Adaptation models collaborative selection semisupervised learning methods F-score sparse text tweets sentiment analysis Support vector machines adaptive feature sentiment labels topic-adaptive features timeline visualization TASC learning algorithm Data visualization topic-related sentiment words TASC model semisupervised topic-adaptive sentiment classification sentiment connections multiclass SVM feature extraction Sentiment analysis pattern classification social networking (online) TASC-t ensemble classifiers topic-sensitive task color gradation intensity supervised classifiers universal classifier dynamic tweets text analysis mixed labeled data nontext features extraction cross-domain Google river graph topic-adaptive unlabeled data sentiment classification Twitter Product reviews Xueqi Cheng oth Fuxin Li oth Fangtao Li oth Enthalten in IEEE transactions on knowledge and data engineering New York, NY : IEEE, 1989 27(2015), 6, Seite 1696-1709 (DE-627)130765732 (DE-600)1001468-8 (DE-576)023036060 1041-4347 nnns volume:27 year:2015 number:6 pages:1696-1709 http://dx.doi.org/10.1109/TKDE.2014.2382600 Volltext http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=6990628 http://search.proquest.com/docview/1677626558 GBV_USEFLAG_A SYSFLAG_A GBV_OLC SSG-OLC-TEC SSG-OLC-PHY SSG-OLC-MAT SSG-OPC-BBI SA 5571 54.00 AVZ AR 27 2015 6 1696-1709 |
| allfieldsGer |
10.1109/TKDE.2014.2382600 doi PQ20160617 (DE-627)OLC1958213837 (DE-599)GBVOLC1958213837 (PRQ)c2119-bb3b284f9fc675eb1420da6ebed2aa58c7d2a180319a1ffa14762a17bd5720070 (KEY)0175400920150000027000601696tasctopicadaptivesentimentclassificationondynamict DE-627 ger DE-627 rakwb eng 620 530 000 DNB SA 5571 AVZ rvk 54.00 bkl Shenghua Liu verfasserin aut TASC:Topic-Adaptive Sentiment Classification on Dynamic Tweets 2015 Text txt rdacontent ohne Hilfsmittel zu benutzen n rdamedia Band nc rdacarrier Sentiment classification is a topic-sensitive task, i.e., a classifier trained from one topic will perform worse on another. This is especially a problem for the tweets sentiment analysis. Since the topics in Twitter are very diverse, it is impossible to train a universal classifier for all topics. Moreover, compared to product review, Twitter lacks data labeling and a rating mechanism to acquire sentiment labels. The extremely sparse text of tweets also brings down the performance of a sentiment classifier. In this paper, we propose a semi-supervised topic-adaptive sentiment classification (TASC) model, which starts with a classifier built on common features and mixed labeled data from various topics. It minimizes the hinge loss to adapt to unlabeled data and features including topic-related sentiment words, authors' sentiments and sentiment connections derived from"" mentions of tweets, named as topic-adaptive features. Text and non-text features are extracted and naturally split into two views for co-training. The TASC learning algorithm updates topic-adaptive features based on the collaborative selection of unlabeled data, which in turn helps to select more reliable tweets to boost the performance. We also design the adapting model along a timeline (TASC-t) for dynamic tweets. An experiment on 6 topics from published tweet corpuses demonstrates that TASC outperforms other well-known supervised and ensemble classifiers. It also beats those semi-supervised learning methods without feature adaption. Meanwhile, TASC-t can also achieve impressive accuracy and F-score. Finally, with timeline visualization of "river" graph, people can intuitively grasp the ups and downs of sentiments' evolvement, and the intensity by color gradation. learning (artificial intelligence) Adaptation models collaborative selection semisupervised learning methods F-score sparse text tweets sentiment analysis Support vector machines adaptive feature sentiment labels topic-adaptive features timeline visualization TASC learning algorithm Data visualization topic-related sentiment words TASC model semisupervised topic-adaptive sentiment classification sentiment connections multiclass SVM feature extraction Sentiment analysis pattern classification social networking (online) TASC-t ensemble classifiers topic-sensitive task color gradation intensity supervised classifiers universal classifier dynamic tweets text analysis mixed labeled data nontext features extraction cross-domain Google river graph topic-adaptive unlabeled data sentiment classification Twitter Product reviews Xueqi Cheng oth Fuxin Li oth Fangtao Li oth Enthalten in IEEE transactions on knowledge and data engineering New York, NY : IEEE, 1989 27(2015), 6, Seite 1696-1709 (DE-627)130765732 (DE-600)1001468-8 (DE-576)023036060 1041-4347 nnns volume:27 year:2015 number:6 pages:1696-1709 http://dx.doi.org/10.1109/TKDE.2014.2382600 Volltext http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=6990628 http://search.proquest.com/docview/1677626558 GBV_USEFLAG_A SYSFLAG_A GBV_OLC SSG-OLC-TEC SSG-OLC-PHY SSG-OLC-MAT SSG-OPC-BBI SA 5571 54.00 AVZ AR 27 2015 6 1696-1709 |
| allfieldsSound |
10.1109/TKDE.2014.2382600 doi PQ20160617 (DE-627)OLC1958213837 (DE-599)GBVOLC1958213837 (PRQ)c2119-bb3b284f9fc675eb1420da6ebed2aa58c7d2a180319a1ffa14762a17bd5720070 (KEY)0175400920150000027000601696tasctopicadaptivesentimentclassificationondynamict DE-627 ger DE-627 rakwb eng 620 530 000 DNB SA 5571 AVZ rvk 54.00 bkl Shenghua Liu verfasserin aut TASC:Topic-Adaptive Sentiment Classification on Dynamic Tweets 2015 Text txt rdacontent ohne Hilfsmittel zu benutzen n rdamedia Band nc rdacarrier Sentiment classification is a topic-sensitive task, i.e., a classifier trained from one topic will perform worse on another. This is especially a problem for the tweets sentiment analysis. Since the topics in Twitter are very diverse, it is impossible to train a universal classifier for all topics. Moreover, compared to product review, Twitter lacks data labeling and a rating mechanism to acquire sentiment labels. The extremely sparse text of tweets also brings down the performance of a sentiment classifier. In this paper, we propose a semi-supervised topic-adaptive sentiment classification (TASC) model, which starts with a classifier built on common features and mixed labeled data from various topics. It minimizes the hinge loss to adapt to unlabeled data and features including topic-related sentiment words, authors' sentiments and sentiment connections derived from"" mentions of tweets, named as topic-adaptive features. Text and non-text features are extracted and naturally split into two views for co-training. The TASC learning algorithm updates topic-adaptive features based on the collaborative selection of unlabeled data, which in turn helps to select more reliable tweets to boost the performance. We also design the adapting model along a timeline (TASC-t) for dynamic tweets. An experiment on 6 topics from published tweet corpuses demonstrates that TASC outperforms other well-known supervised and ensemble classifiers. It also beats those semi-supervised learning methods without feature adaption. Meanwhile, TASC-t can also achieve impressive accuracy and F-score. Finally, with timeline visualization of "river" graph, people can intuitively grasp the ups and downs of sentiments' evolvement, and the intensity by color gradation. learning (artificial intelligence) Adaptation models collaborative selection semisupervised learning methods F-score sparse text tweets sentiment analysis Support vector machines adaptive feature sentiment labels topic-adaptive features timeline visualization TASC learning algorithm Data visualization topic-related sentiment words TASC model semisupervised topic-adaptive sentiment classification sentiment connections multiclass SVM feature extraction Sentiment analysis pattern classification social networking (online) TASC-t ensemble classifiers topic-sensitive task color gradation intensity supervised classifiers universal classifier dynamic tweets text analysis mixed labeled data nontext features extraction cross-domain Google river graph topic-adaptive unlabeled data sentiment classification Twitter Product reviews Xueqi Cheng oth Fuxin Li oth Fangtao Li oth Enthalten in IEEE transactions on knowledge and data engineering New York, NY : IEEE, 1989 27(2015), 6, Seite 1696-1709 (DE-627)130765732 (DE-600)1001468-8 (DE-576)023036060 1041-4347 nnns volume:27 year:2015 number:6 pages:1696-1709 http://dx.doi.org/10.1109/TKDE.2014.2382600 Volltext http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=6990628 http://search.proquest.com/docview/1677626558 GBV_USEFLAG_A SYSFLAG_A GBV_OLC SSG-OLC-TEC SSG-OLC-PHY SSG-OLC-MAT SSG-OPC-BBI SA 5571 54.00 AVZ AR 27 2015 6 1696-1709 |
| language |
English |
| source |
Enthalten in IEEE transactions on knowledge and data engineering 27(2015), 6, Seite 1696-1709 volume:27 year:2015 number:6 pages:1696-1709 |
| sourceStr |
Enthalten in IEEE transactions on knowledge and data engineering 27(2015), 6, Seite 1696-1709 volume:27 year:2015 number:6 pages:1696-1709 |
| format_phy_str_mv |
Article |
| institution |
findex.gbv.de |
| topic_facet |
learning (artificial intelligence) Adaptation models collaborative selection semisupervised learning methods F-score sparse text tweets sentiment analysis Support vector machines adaptive feature sentiment labels topic-adaptive features timeline visualization TASC learning algorithm Data visualization topic-related sentiment words TASC model semisupervised topic-adaptive sentiment classification sentiment connections multiclass SVM feature extraction Sentiment analysis pattern classification social networking (online) TASC-t ensemble classifiers topic-sensitive task color gradation intensity supervised classifiers universal classifier dynamic tweets text analysis mixed labeled data nontext features extraction cross-domain river graph topic-adaptive unlabeled data sentiment classification Product reviews |
| dewey-raw |
620 |
| isfreeaccess_bool |
false |
| container_title |
IEEE transactions on knowledge and data engineering |
| authorswithroles_txt_mv |
Shenghua Liu @@aut@@ Xueqi Cheng @@oth@@ Fuxin Li @@oth@@ Fangtao Li @@oth@@ |
| publishDateDaySort_date |
2015-01-01T00:00:00Z |
| hierarchy_top_id |
130765732 |
| dewey-sort |
3620 |
| id |
OLC1958213837 |
| language_de |
englisch |
| fullrecord |
<?xml version="1.0" encoding="UTF-8"?><collection xmlns="http://www.loc.gov/MARC21/slim"><record><leader>01000caa a2200265 4500</leader><controlfield tag="001">OLC1958213837</controlfield><controlfield tag="003">DE-627</controlfield><controlfield tag="005">20230714144306.0</controlfield><controlfield tag="007">tu</controlfield><controlfield tag="008">160206s2015 xx ||||| 00| ||eng c</controlfield><datafield tag="024" ind1="7" ind2=" "><subfield code="a">10.1109/TKDE.2014.2382600</subfield><subfield code="2">doi</subfield></datafield><datafield tag="028" ind1="5" ind2="2"><subfield code="a">PQ20160617</subfield></datafield><datafield tag="035" ind1=" " ind2=" "><subfield code="a">(DE-627)OLC1958213837</subfield></datafield><datafield tag="035" ind1=" " ind2=" "><subfield code="a">(DE-599)GBVOLC1958213837</subfield></datafield><datafield tag="035" ind1=" " ind2=" "><subfield code="a">(PRQ)c2119-bb3b284f9fc675eb1420da6ebed2aa58c7d2a180319a1ffa14762a17bd5720070</subfield></datafield><datafield tag="035" ind1=" " ind2=" "><subfield code="a">(KEY)0175400920150000027000601696tasctopicadaptivesentimentclassificationondynamict</subfield></datafield><datafield tag="040" ind1=" " ind2=" "><subfield code="a">DE-627</subfield><subfield code="b">ger</subfield><subfield code="c">DE-627</subfield><subfield code="e">rakwb</subfield></datafield><datafield tag="041" ind1=" " ind2=" "><subfield code="a">eng</subfield></datafield><datafield tag="082" ind1="0" ind2="4"><subfield code="a">620</subfield><subfield code="a">530</subfield><subfield code="a">000</subfield><subfield code="q">DNB</subfield></datafield><datafield tag="084" ind1=" " ind2=" "><subfield code="a">SA 5571</subfield><subfield code="q">AVZ</subfield><subfield code="2">rvk</subfield></datafield><datafield tag="084" ind1=" " ind2=" "><subfield code="a">54.00</subfield><subfield code="2">bkl</subfield></datafield><datafield tag="100" ind1="0" ind2=" "><subfield code="a">Shenghua Liu</subfield><subfield code="e">verfasserin</subfield><subfield code="4">aut</subfield></datafield><datafield tag="245" ind1="1" ind2="0"><subfield code="a">TASC:Topic-Adaptive Sentiment Classification on Dynamic Tweets</subfield></datafield><datafield tag="264" ind1=" " ind2="1"><subfield code="c">2015</subfield></datafield><datafield tag="336" ind1=" " ind2=" "><subfield code="a">Text</subfield><subfield code="b">txt</subfield><subfield code="2">rdacontent</subfield></datafield><datafield tag="337" ind1=" " ind2=" "><subfield code="a">ohne Hilfsmittel zu benutzen</subfield><subfield code="b">n</subfield><subfield code="2">rdamedia</subfield></datafield><datafield tag="338" ind1=" " ind2=" "><subfield code="a">Band</subfield><subfield code="b">nc</subfield><subfield code="2">rdacarrier</subfield></datafield><datafield tag="520" ind1=" " ind2=" "><subfield code="a">Sentiment classification is a topic-sensitive task, i.e., a classifier trained from one topic will perform worse on another. This is especially a problem for the tweets sentiment analysis. Since the topics in Twitter are very diverse, it is impossible to train a universal classifier for all topics. Moreover, compared to product review, Twitter lacks data labeling and a rating mechanism to acquire sentiment labels. The extremely sparse text of tweets also brings down the performance of a sentiment classifier. In this paper, we propose a semi-supervised topic-adaptive sentiment classification (TASC) model, which starts with a classifier built on common features and mixed labeled data from various topics. It minimizes the hinge loss to adapt to unlabeled data and features including topic-related sentiment words, authors' sentiments and sentiment connections derived from"" mentions of tweets, named as topic-adaptive features. Text and non-text features are extracted and naturally split into two views for co-training. The TASC learning algorithm updates topic-adaptive features based on the collaborative selection of unlabeled data, which in turn helps to select more reliable tweets to boost the performance. We also design the adapting model along a timeline (TASC-t) for dynamic tweets. An experiment on 6 topics from published tweet corpuses demonstrates that TASC outperforms other well-known supervised and ensemble classifiers. It also beats those semi-supervised learning methods without feature adaption. Meanwhile, TASC-t can also achieve impressive accuracy and F-score. Finally, with timeline visualization of "river" graph, people can intuitively grasp the ups and downs of sentiments' evolvement, and the intensity by color gradation.</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">learning (artificial intelligence)</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">Adaptation models</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">collaborative selection</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">semisupervised learning methods</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">F-score</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">sparse text</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">tweets sentiment analysis</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">Support vector machines</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">adaptive feature</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">sentiment labels</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">topic-adaptive features</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">timeline visualization</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">TASC learning algorithm</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">Data visualization</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">topic-related sentiment words</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">TASC model</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">semisupervised topic-adaptive sentiment classification</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">sentiment connections</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">multiclass SVM</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">feature extraction</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">Sentiment analysis</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">pattern classification</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">social networking (online)</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">TASC-t</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">ensemble classifiers</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">topic-sensitive task</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">color gradation intensity</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">supervised classifiers</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">universal classifier</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">dynamic tweets</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">text analysis</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">mixed labeled data</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">nontext features extraction</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">cross-domain</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">Google</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">river graph</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">topic-adaptive</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">unlabeled data</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">sentiment classification</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">Twitter</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">Product reviews</subfield></datafield><datafield tag="700" ind1="0" ind2=" "><subfield code="a">Xueqi Cheng</subfield><subfield code="4">oth</subfield></datafield><datafield tag="700" ind1="0" ind2=" "><subfield code="a">Fuxin Li</subfield><subfield code="4">oth</subfield></datafield><datafield tag="700" ind1="0" ind2=" "><subfield code="a">Fangtao Li</subfield><subfield code="4">oth</subfield></datafield><datafield tag="773" ind1="0" ind2="8"><subfield code="i">Enthalten in</subfield><subfield code="t">IEEE transactions on knowledge and data engineering</subfield><subfield code="d">New York, NY : IEEE, 1989</subfield><subfield code="g">27(2015), 6, Seite 1696-1709</subfield><subfield code="w">(DE-627)130765732</subfield><subfield code="w">(DE-600)1001468-8</subfield><subfield code="w">(DE-576)023036060</subfield><subfield code="x">1041-4347</subfield><subfield code="7">nnns</subfield></datafield><datafield tag="773" ind1="1" ind2="8"><subfield code="g">volume:27</subfield><subfield code="g">year:2015</subfield><subfield code="g">number:6</subfield><subfield code="g">pages:1696-1709</subfield></datafield><datafield tag="856" ind1="4" ind2="1"><subfield code="u">http://dx.doi.org/10.1109/TKDE.2014.2382600</subfield><subfield code="3">Volltext</subfield></datafield><datafield tag="856" ind1="4" ind2="2"><subfield code="u">http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=6990628</subfield></datafield><datafield tag="856" ind1="4" ind2="2"><subfield code="u">http://search.proquest.com/docview/1677626558</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_USEFLAG_A</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">SYSFLAG_A</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_OLC</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">SSG-OLC-TEC</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">SSG-OLC-PHY</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">SSG-OLC-MAT</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">SSG-OPC-BBI</subfield></datafield><datafield tag="936" ind1="r" ind2="v"><subfield code="a">SA 5571</subfield></datafield><datafield tag="936" ind1="b" ind2="k"><subfield code="a">54.00</subfield><subfield code="q">AVZ</subfield></datafield><datafield tag="951" ind1=" " ind2=" "><subfield code="a">AR</subfield></datafield><datafield tag="952" ind1=" " ind2=" "><subfield code="d">27</subfield><subfield code="j">2015</subfield><subfield code="e">6</subfield><subfield code="h">1696-1709</subfield></datafield></record></collection>
|
| author |
Shenghua Liu |
| spellingShingle |
Shenghua Liu ddc 620 rvk SA 5571 bkl 54.00 misc learning (artificial intelligence) misc Adaptation models misc collaborative selection misc semisupervised learning methods misc F-score misc sparse text misc tweets sentiment analysis misc Support vector machines misc adaptive feature misc sentiment labels misc topic-adaptive features misc timeline visualization misc TASC learning algorithm misc Data visualization misc topic-related sentiment words misc TASC model misc semisupervised topic-adaptive sentiment classification misc sentiment connections misc multiclass SVM misc feature extraction misc Sentiment analysis misc pattern classification misc social networking (online) misc TASC-t misc ensemble classifiers misc topic-sensitive task misc color gradation intensity misc supervised classifiers misc universal classifier misc dynamic tweets misc text analysis misc mixed labeled data misc nontext features extraction misc cross-domain misc Google misc river graph misc topic-adaptive misc unlabeled data misc sentiment classification misc Twitter misc Product reviews TASC:Topic-Adaptive Sentiment Classification on Dynamic Tweets |
| authorStr |
Shenghua Liu |
| ppnlink_with_tag_str_mv |
@@773@@(DE-627)130765732 |
| format |
Article |
| dewey-ones |
620 - Engineering & allied operations 530 - Physics 000 - Computer science, information & general works |
| delete_txt_mv |
keep |
| author_role |
aut |
| collection |
OLC |
| remote_str |
false |
| illustrated |
Not Illustrated |
| issn |
1041-4347 |
| topic_title |
620 530 000 DNB SA 5571 AVZ rvk 54.00 bkl TASC:Topic-Adaptive Sentiment Classification on Dynamic Tweets learning (artificial intelligence) Adaptation models collaborative selection semisupervised learning methods F-score sparse text tweets sentiment analysis Support vector machines adaptive feature sentiment labels topic-adaptive features timeline visualization TASC learning algorithm Data visualization topic-related sentiment words TASC model semisupervised topic-adaptive sentiment classification sentiment connections multiclass SVM feature extraction Sentiment analysis pattern classification social networking (online) TASC-t ensemble classifiers topic-sensitive task color gradation intensity supervised classifiers universal classifier dynamic tweets text analysis mixed labeled data nontext features extraction cross-domain river graph topic-adaptive unlabeled data sentiment classification Product reviews |
| topic |
ddc 620 rvk SA 5571 bkl 54.00 misc learning (artificial intelligence) misc Adaptation models misc collaborative selection misc semisupervised learning methods misc F-score misc sparse text misc tweets sentiment analysis misc Support vector machines misc adaptive feature misc sentiment labels misc topic-adaptive features misc timeline visualization misc TASC learning algorithm misc Data visualization misc topic-related sentiment words misc TASC model misc semisupervised topic-adaptive sentiment classification misc sentiment connections misc multiclass SVM misc feature extraction misc Sentiment analysis misc pattern classification misc social networking (online) misc TASC-t misc ensemble classifiers misc topic-sensitive task misc color gradation intensity misc supervised classifiers misc universal classifier misc dynamic tweets misc text analysis misc mixed labeled data misc nontext features extraction misc cross-domain misc Google misc river graph misc topic-adaptive misc unlabeled data misc sentiment classification misc Twitter misc Product reviews |
| topic_unstemmed |
ddc 620 rvk SA 5571 bkl 54.00 misc learning (artificial intelligence) misc Adaptation models misc collaborative selection misc semisupervised learning methods misc F-score misc sparse text misc tweets sentiment analysis misc Support vector machines misc adaptive feature misc sentiment labels misc topic-adaptive features misc timeline visualization misc TASC learning algorithm misc Data visualization misc topic-related sentiment words misc TASC model misc semisupervised topic-adaptive sentiment classification misc sentiment connections misc multiclass SVM misc feature extraction misc Sentiment analysis misc pattern classification misc social networking (online) misc TASC-t misc ensemble classifiers misc topic-sensitive task misc color gradation intensity misc supervised classifiers misc universal classifier misc dynamic tweets misc text analysis misc mixed labeled data misc nontext features extraction misc cross-domain misc Google misc river graph misc topic-adaptive misc unlabeled data misc sentiment classification misc Twitter misc Product reviews |
| topic_browse |
ddc 620 rvk SA 5571 bkl 54.00 misc learning (artificial intelligence) misc Adaptation models misc collaborative selection misc semisupervised learning methods misc F-score misc sparse text misc tweets sentiment analysis misc Support vector machines misc adaptive feature misc sentiment labels misc topic-adaptive features misc timeline visualization misc TASC learning algorithm misc Data visualization misc topic-related sentiment words misc TASC model misc semisupervised topic-adaptive sentiment classification misc sentiment connections misc multiclass SVM misc feature extraction misc Sentiment analysis misc pattern classification misc social networking (online) misc TASC-t misc ensemble classifiers misc topic-sensitive task misc color gradation intensity misc supervised classifiers misc universal classifier misc dynamic tweets misc text analysis misc mixed labeled data misc nontext features extraction misc cross-domain misc Google misc river graph misc topic-adaptive misc unlabeled data misc sentiment classification misc Twitter misc Product reviews |
| format_facet |
Aufsätze Gedruckte Aufsätze |
| format_main_str_mv |
Text Zeitschrift/Artikel |
| carriertype_str_mv |
nc |
| author2_variant |
x c xc f l fl f l fl |
| hierarchy_parent_title |
IEEE transactions on knowledge and data engineering |
| hierarchy_parent_id |
130765732 |
| dewey-tens |
620 - Engineering 530 - Physics 000 - Computer science, knowledge & systems |
| hierarchy_top_title |
IEEE transactions on knowledge and data engineering |
| isfreeaccess_txt |
false |
| familylinks_str_mv |
(DE-627)130765732 (DE-600)1001468-8 (DE-576)023036060 |
| title |
TASC:Topic-Adaptive Sentiment Classification on Dynamic Tweets |
| ctrlnum |
(DE-627)OLC1958213837 (DE-599)GBVOLC1958213837 (PRQ)c2119-bb3b284f9fc675eb1420da6ebed2aa58c7d2a180319a1ffa14762a17bd5720070 (KEY)0175400920150000027000601696tasctopicadaptivesentimentclassificationondynamict |
| title_full |
TASC:Topic-Adaptive Sentiment Classification on Dynamic Tweets |
| author_sort |
Shenghua Liu |
| journal |
IEEE transactions on knowledge and data engineering |
| journalStr |
IEEE transactions on knowledge and data engineering |
| lang_code |
eng |
| isOA_bool |
false |
| dewey-hundreds |
600 - Technology 500 - Science 000 - Computer science, information & general works |
| recordtype |
marc |
| publishDateSort |
2015 |
| contenttype_str_mv |
txt |
| container_start_page |
1696 |
| author_browse |
Shenghua Liu |
| container_volume |
27 |
| class |
620 530 000 DNB SA 5571 AVZ rvk 54.00 bkl |
| format_se |
Aufsätze |
| author-letter |
Shenghua Liu |
| doi_str_mv |
10.1109/TKDE.2014.2382600 |
| dewey-full |
620 530 000 |
| title_sort |
tasc:topic-adaptive sentiment classification on dynamic tweets |
| title_auth |
TASC:Topic-Adaptive Sentiment Classification on Dynamic Tweets |
| abstract |
Sentiment classification is a topic-sensitive task, i.e., a classifier trained from one topic will perform worse on another. This is especially a problem for the tweets sentiment analysis. Since the topics in Twitter are very diverse, it is impossible to train a universal classifier for all topics. Moreover, compared to product review, Twitter lacks data labeling and a rating mechanism to acquire sentiment labels. The extremely sparse text of tweets also brings down the performance of a sentiment classifier. In this paper, we propose a semi-supervised topic-adaptive sentiment classification (TASC) model, which starts with a classifier built on common features and mixed labeled data from various topics. It minimizes the hinge loss to adapt to unlabeled data and features including topic-related sentiment words, authors' sentiments and sentiment connections derived from"" mentions of tweets, named as topic-adaptive features. Text and non-text features are extracted and naturally split into two views for co-training. The TASC learning algorithm updates topic-adaptive features based on the collaborative selection of unlabeled data, which in turn helps to select more reliable tweets to boost the performance. We also design the adapting model along a timeline (TASC-t) for dynamic tweets. An experiment on 6 topics from published tweet corpuses demonstrates that TASC outperforms other well-known supervised and ensemble classifiers. It also beats those semi-supervised learning methods without feature adaption. Meanwhile, TASC-t can also achieve impressive accuracy and F-score. Finally, with timeline visualization of "river" graph, people can intuitively grasp the ups and downs of sentiments' evolvement, and the intensity by color gradation. |
| abstractGer |
Sentiment classification is a topic-sensitive task, i.e., a classifier trained from one topic will perform worse on another. This is especially a problem for the tweets sentiment analysis. Since the topics in Twitter are very diverse, it is impossible to train a universal classifier for all topics. Moreover, compared to product review, Twitter lacks data labeling and a rating mechanism to acquire sentiment labels. The extremely sparse text of tweets also brings down the performance of a sentiment classifier. In this paper, we propose a semi-supervised topic-adaptive sentiment classification (TASC) model, which starts with a classifier built on common features and mixed labeled data from various topics. It minimizes the hinge loss to adapt to unlabeled data and features including topic-related sentiment words, authors' sentiments and sentiment connections derived from"" mentions of tweets, named as topic-adaptive features. Text and non-text features are extracted and naturally split into two views for co-training. The TASC learning algorithm updates topic-adaptive features based on the collaborative selection of unlabeled data, which in turn helps to select more reliable tweets to boost the performance. We also design the adapting model along a timeline (TASC-t) for dynamic tweets. An experiment on 6 topics from published tweet corpuses demonstrates that TASC outperforms other well-known supervised and ensemble classifiers. It also beats those semi-supervised learning methods without feature adaption. Meanwhile, TASC-t can also achieve impressive accuracy and F-score. Finally, with timeline visualization of "river" graph, people can intuitively grasp the ups and downs of sentiments' evolvement, and the intensity by color gradation. |
| abstract_unstemmed |
Sentiment classification is a topic-sensitive task, i.e., a classifier trained from one topic will perform worse on another. This is especially a problem for the tweets sentiment analysis. Since the topics in Twitter are very diverse, it is impossible to train a universal classifier for all topics. Moreover, compared to product review, Twitter lacks data labeling and a rating mechanism to acquire sentiment labels. The extremely sparse text of tweets also brings down the performance of a sentiment classifier. In this paper, we propose a semi-supervised topic-adaptive sentiment classification (TASC) model, which starts with a classifier built on common features and mixed labeled data from various topics. It minimizes the hinge loss to adapt to unlabeled data and features including topic-related sentiment words, authors' sentiments and sentiment connections derived from"" mentions of tweets, named as topic-adaptive features. Text and non-text features are extracted and naturally split into two views for co-training. The TASC learning algorithm updates topic-adaptive features based on the collaborative selection of unlabeled data, which in turn helps to select more reliable tweets to boost the performance. We also design the adapting model along a timeline (TASC-t) for dynamic tweets. An experiment on 6 topics from published tweet corpuses demonstrates that TASC outperforms other well-known supervised and ensemble classifiers. It also beats those semi-supervised learning methods without feature adaption. Meanwhile, TASC-t can also achieve impressive accuracy and F-score. Finally, with timeline visualization of "river" graph, people can intuitively grasp the ups and downs of sentiments' evolvement, and the intensity by color gradation. |
| collection_details |
GBV_USEFLAG_A SYSFLAG_A GBV_OLC SSG-OLC-TEC SSG-OLC-PHY SSG-OLC-MAT SSG-OPC-BBI |
| container_issue |
6 |
| title_short |
TASC:Topic-Adaptive Sentiment Classification on Dynamic Tweets |
| url |
http://dx.doi.org/10.1109/TKDE.2014.2382600 http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=6990628 http://search.proquest.com/docview/1677626558 |
| remote_bool |
false |
| author2 |
Xueqi Cheng Fuxin Li Fangtao Li |
| author2Str |
Xueqi Cheng Fuxin Li Fangtao Li |
| ppnlink |
130765732 |
| mediatype_str_mv |
n |
| isOA_txt |
false |
| hochschulschrift_bool |
false |
| author2_role |
oth oth oth |
| doi_str |
10.1109/TKDE.2014.2382600 |
| up_date |
2024-07-04T02:22:13.876Z |
| _version_ |
1803613355510857728 |
| fullrecord_marcxml |
<?xml version="1.0" encoding="UTF-8"?><collection xmlns="http://www.loc.gov/MARC21/slim"><record><leader>01000caa a2200265 4500</leader><controlfield tag="001">OLC1958213837</controlfield><controlfield tag="003">DE-627</controlfield><controlfield tag="005">20230714144306.0</controlfield><controlfield tag="007">tu</controlfield><controlfield tag="008">160206s2015 xx ||||| 00| ||eng c</controlfield><datafield tag="024" ind1="7" ind2=" "><subfield code="a">10.1109/TKDE.2014.2382600</subfield><subfield code="2">doi</subfield></datafield><datafield tag="028" ind1="5" ind2="2"><subfield code="a">PQ20160617</subfield></datafield><datafield tag="035" ind1=" " ind2=" "><subfield code="a">(DE-627)OLC1958213837</subfield></datafield><datafield tag="035" ind1=" " ind2=" "><subfield code="a">(DE-599)GBVOLC1958213837</subfield></datafield><datafield tag="035" ind1=" " ind2=" "><subfield code="a">(PRQ)c2119-bb3b284f9fc675eb1420da6ebed2aa58c7d2a180319a1ffa14762a17bd5720070</subfield></datafield><datafield tag="035" ind1=" " ind2=" "><subfield code="a">(KEY)0175400920150000027000601696tasctopicadaptivesentimentclassificationondynamict</subfield></datafield><datafield tag="040" ind1=" " ind2=" "><subfield code="a">DE-627</subfield><subfield code="b">ger</subfield><subfield code="c">DE-627</subfield><subfield code="e">rakwb</subfield></datafield><datafield tag="041" ind1=" " ind2=" "><subfield code="a">eng</subfield></datafield><datafield tag="082" ind1="0" ind2="4"><subfield code="a">620</subfield><subfield code="a">530</subfield><subfield code="a">000</subfield><subfield code="q">DNB</subfield></datafield><datafield tag="084" ind1=" " ind2=" "><subfield code="a">SA 5571</subfield><subfield code="q">AVZ</subfield><subfield code="2">rvk</subfield></datafield><datafield tag="084" ind1=" " ind2=" "><subfield code="a">54.00</subfield><subfield code="2">bkl</subfield></datafield><datafield tag="100" ind1="0" ind2=" "><subfield code="a">Shenghua Liu</subfield><subfield code="e">verfasserin</subfield><subfield code="4">aut</subfield></datafield><datafield tag="245" ind1="1" ind2="0"><subfield code="a">TASC:Topic-Adaptive Sentiment Classification on Dynamic Tweets</subfield></datafield><datafield tag="264" ind1=" " ind2="1"><subfield code="c">2015</subfield></datafield><datafield tag="336" ind1=" " ind2=" "><subfield code="a">Text</subfield><subfield code="b">txt</subfield><subfield code="2">rdacontent</subfield></datafield><datafield tag="337" ind1=" " ind2=" "><subfield code="a">ohne Hilfsmittel zu benutzen</subfield><subfield code="b">n</subfield><subfield code="2">rdamedia</subfield></datafield><datafield tag="338" ind1=" " ind2=" "><subfield code="a">Band</subfield><subfield code="b">nc</subfield><subfield code="2">rdacarrier</subfield></datafield><datafield tag="520" ind1=" " ind2=" "><subfield code="a">Sentiment classification is a topic-sensitive task, i.e., a classifier trained from one topic will perform worse on another. This is especially a problem for the tweets sentiment analysis. Since the topics in Twitter are very diverse, it is impossible to train a universal classifier for all topics. Moreover, compared to product review, Twitter lacks data labeling and a rating mechanism to acquire sentiment labels. The extremely sparse text of tweets also brings down the performance of a sentiment classifier. In this paper, we propose a semi-supervised topic-adaptive sentiment classification (TASC) model, which starts with a classifier built on common features and mixed labeled data from various topics. It minimizes the hinge loss to adapt to unlabeled data and features including topic-related sentiment words, authors' sentiments and sentiment connections derived from"" mentions of tweets, named as topic-adaptive features. Text and non-text features are extracted and naturally split into two views for co-training. The TASC learning algorithm updates topic-adaptive features based on the collaborative selection of unlabeled data, which in turn helps to select more reliable tweets to boost the performance. We also design the adapting model along a timeline (TASC-t) for dynamic tweets. An experiment on 6 topics from published tweet corpuses demonstrates that TASC outperforms other well-known supervised and ensemble classifiers. It also beats those semi-supervised learning methods without feature adaption. Meanwhile, TASC-t can also achieve impressive accuracy and F-score. Finally, with timeline visualization of "river" graph, people can intuitively grasp the ups and downs of sentiments' evolvement, and the intensity by color gradation.</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">learning (artificial intelligence)</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">Adaptation models</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">collaborative selection</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">semisupervised learning methods</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">F-score</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">sparse text</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">tweets sentiment analysis</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">Support vector machines</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">adaptive feature</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">sentiment labels</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">topic-adaptive features</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">timeline visualization</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">TASC learning algorithm</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">Data visualization</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">topic-related sentiment words</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">TASC model</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">semisupervised topic-adaptive sentiment classification</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">sentiment connections</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">multiclass SVM</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">feature extraction</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">Sentiment analysis</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">pattern classification</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">social networking (online)</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">TASC-t</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">ensemble classifiers</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">topic-sensitive task</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">color gradation intensity</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">supervised classifiers</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">universal classifier</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">dynamic tweets</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">text analysis</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">mixed labeled data</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">nontext features extraction</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">cross-domain</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">Google</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">river graph</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">topic-adaptive</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">unlabeled data</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">sentiment classification</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">Twitter</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">Product reviews</subfield></datafield><datafield tag="700" ind1="0" ind2=" "><subfield code="a">Xueqi Cheng</subfield><subfield code="4">oth</subfield></datafield><datafield tag="700" ind1="0" ind2=" "><subfield code="a">Fuxin Li</subfield><subfield code="4">oth</subfield></datafield><datafield tag="700" ind1="0" ind2=" "><subfield code="a">Fangtao Li</subfield><subfield code="4">oth</subfield></datafield><datafield tag="773" ind1="0" ind2="8"><subfield code="i">Enthalten in</subfield><subfield code="t">IEEE transactions on knowledge and data engineering</subfield><subfield code="d">New York, NY : IEEE, 1989</subfield><subfield code="g">27(2015), 6, Seite 1696-1709</subfield><subfield code="w">(DE-627)130765732</subfield><subfield code="w">(DE-600)1001468-8</subfield><subfield code="w">(DE-576)023036060</subfield><subfield code="x">1041-4347</subfield><subfield code="7">nnns</subfield></datafield><datafield tag="773" ind1="1" ind2="8"><subfield code="g">volume:27</subfield><subfield code="g">year:2015</subfield><subfield code="g">number:6</subfield><subfield code="g">pages:1696-1709</subfield></datafield><datafield tag="856" ind1="4" ind2="1"><subfield code="u">http://dx.doi.org/10.1109/TKDE.2014.2382600</subfield><subfield code="3">Volltext</subfield></datafield><datafield tag="856" ind1="4" ind2="2"><subfield code="u">http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=6990628</subfield></datafield><datafield tag="856" ind1="4" ind2="2"><subfield code="u">http://search.proquest.com/docview/1677626558</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_USEFLAG_A</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">SYSFLAG_A</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_OLC</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">SSG-OLC-TEC</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">SSG-OLC-PHY</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">SSG-OLC-MAT</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">SSG-OPC-BBI</subfield></datafield><datafield tag="936" ind1="r" ind2="v"><subfield code="a">SA 5571</subfield></datafield><datafield tag="936" ind1="b" ind2="k"><subfield code="a">54.00</subfield><subfield code="q">AVZ</subfield></datafield><datafield tag="951" ind1=" " ind2=" "><subfield code="a">AR</subfield></datafield><datafield tag="952" ind1=" " ind2=" "><subfield code="d">27</subfield><subfield code="j">2015</subfield><subfield code="e">6</subfield><subfield code="h">1696-1709</subfield></datafield></record></collection>
|
| score |
7.3993654 |