OTSD: 58,000+ comments in Online Community for Target-level Sentiment Analysis
- Title
- OTSD: 58,000+ comments in Online Community for Target-level Sentiment Analysis
- Authors
- 현동민
- Date Issued
- 2017
- Publisher
- 포항공과대학교
- Abstract
- Target-level sentiment analysis (TLSA) is to extract sentiments toward tar-
gets in text. Since target-level sentiments play an important role in develop-
ing marketing strategies for companies, several researchers have actively studied
TLSA. However, existing datasets do not contain contextual information such
as user, dependent-text. In this paper, we build and release Online community
Target-level Sentiment analysis Dataset (OTSD) that contains 58,000+ online-
community comments with rich contextual information: user ID, dependent-text,
time and thumbs-up. With the real-world datasets, we observe that sentiments
toward targets in text is closely related to not only text itself but also contextual
information. To investigate the effectiveness of the contextual information, we
build baseline models based on convolutional neural network that additionally
utilize user and dependent-text information as the contextual information. Our
experimental evaluations on the OTSD show that our baseline models outperform
the state-of-the-art models that do not use the contextual information.
- URI
- http://postech.dcollection.net/jsp/common/DcLoOrgPer.jsp?sItemId=000002378152
https://oasis.postech.ac.kr/handle/2014.oak/93556
- Article Type
- Thesis
- Files in This Item:
- There are no files associated with this item.
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.