| 講演名 | 2015-01-26 Sentiment Analysis in Twitter for Multiple Topics : How to Detect the Polarity of Tweets Regardless of Their Topic , |
|---|---|
| PDFダウンロードページ |  PDFダウンロードページへ |
| 抄録(和) | |
| 抄録(英) | Being one of the biggest web destinations for people to express their opinions, share their experience and report real-time events, Twitter is attracting significant interests from the research community over the past few years. Tweets, the short messages posted on the Twitter website, have been subject to contextual processing by many researchers for opinion mining and sentiment analysis. Opinion mining and sentiment analysis refer to searching throughout a huge amount of data in the digital space and automatically identify and aggregate attitudes and opinions expressed by users. This is of a great use when it comes to summarizing opinions and reviews, and providing statistics and primary analysis of the level of satisfaction of customers about a product or a service. However, due to the limitation in terms of characters (i.e. 140 characters per tweet) and the use of informal language, applying the conventional techniques to extract sentiments and opinions from a tweet is not applicable, and the accuracy of the classification is in general lower than that when they are applied on long texts (e.g. reviews collected from movie review websites). On the other hand, classifying texts dealing with various topics or themes remains a challenging task: texts dealing with different topics use different lexicons, and some vocabularies might have different meanings depending on the context. Therefore, most of the existing approaches usually classify tweets dealing with one topic. In this report we propose a new method for sentiment analysis that overcomes this limitation and classifies tweets regardless of their topic. We provide a new set of features to be used by machine learning algorithms to classify tweets belonging to different topics. We analyze the importance of each of the proposed features and its contribution to enhance the accuracy of the classification. |
| キーワード(和) | |
| キーワード(英) | Twitter / sentiment analysis / machine learning |
| 資料番号 | ASN2014-122 |
| 発行日 |
| 研究会情報 | |
| 研究会 | ASN |
|---|---|
| 開催期間 | 2015/1/19(から1日開催) |
| 開催地（和） | |
| 開催地（英） | |
| テーマ（和） | |
| テーマ（英） | |
| 委員長氏名（和） | |
| 委員長氏名（英） | |
| 副委員長氏名（和） | |
| 副委員長氏名（英） | |
| 幹事氏名（和） | |
| 幹事氏名（英） | |
| 幹事補佐氏名（和） | |
| 幹事補佐氏名（英） | |
| 講演論文情報詳細 | |
| 申込み研究会 | Ambient intelligence and Sensor Networks(ASN) |
|---|---|
| 本文の言語 | ENG |
| タイトル（和） | |
| サブタイトル（和） | |
| タイトル（英） | Sentiment Analysis in Twitter for Multiple Topics : How to Detect the Polarity of Tweets Regardless of Their Topic |
| サブタイトル（和） | |
| キーワード(1)（和/英） | |
| 第 1 著者 氏名（和/英） | / Mondher BOUAZIZI |
| 第 1 著者 所属（和/英） | Graduate School of Science and Technology, Keio University |
| 発表年月日 | 2015-01-26 |
| 資料番号 | ASN2014-122 |
| 巻番号（vol） | vol.114 |
| 号番号（no） | 418 |
| ページ範囲 | pp.- |
| ページ数 | 6 |
| 発行日 | |