Textual Feature Ensemble-Based Sarcasm Detection in Twitter Data
| Autor: | Anandhakumar Palanisamy, J. Vijay Saravana, Karthik Sundararajan |
|---|---|
| Rok vydání: | 2020 |
| Předmět: |
Sarcasm
Computer science business.industry media_common.quotation_subject Sentiment analysis Feature extraction Context (language use) computer.software_genre Ensemble learning Feature (machine learning) Emoticon Artificial intelligence business computer Natural language processing Sentence media_common |
| Zdroj: | Intelligence in Big Data Technologies—Beyond the Hype ISBN: 9789811552847 |
| Popis: | Sarcasm is one of the most popular ways of expressing one’s opinion on a specific subject. Sarcasm is an act of conveying a viewpoint with an opposite emotion. Identifying such statements among the textual information is an important challenge present in sentiment analysis. Often, sarcastic statements are misleading. They mislead the whole context of the sentence; hence, detecting sarcasm is important in order to understand the actual meaning involved. The main objective of this paper is to propose a feature ensemble-based learning algorithm for sarcasm detection. Most of the existing works focused on using lexical features for identifying sarcasm. In the proposed work apart from the traditional lexical features, hyperbolic and pragmatic features have been extracted. After extracting the features, a model for detecting sarcasm has been proposed based on stacking-based feature ensemble algorithm. Experimental results show that the feature-based approaches when considered individually, Internet slang-based features attain an accuracy of 62%; interjection features obtained an accuracy of 49%, and emoticon-based features attained 59%, respectively. However, the proposed ensemble learning algorithm attains an overall accuracy of 83%, which is much better than the feature set-based approaches. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|