Proposed Approach for Sarcasm Detection in Twitter
| Autor: | Shubhodip Saha, Prabhat Ranjan, Jainath Yadav |
|---|---|
| Rok vydání: | 2017 |
| Předmět: |
Multidisciplinary
Parsing Stop words Sarcasm Computer science business.industry Part-of-speech tagging media_common.quotation_subject 02 engineering and technology computer.software_genre Machine learning Support vector machine Naive Bayes classifier Tokenization (data security) 020204 information systems 0202 electrical engineering electronic engineering information engineering 020201 artificial intelligence & image processing Artificial intelligence business computer Natural language media_common |
| Zdroj: | Indian Journal of Science and Technology. 10:1-8 |
| ISSN: | 0974-5645 0974-6846 |
| Popis: | Background: Sarcasm detection in twitter is a very important task as it had helped in the analysis of tweets. With the help of sarcasm detection, companies could analyze the feelings of user about their products. This is helpful for companies, as the companies could improve their quality of product. Methods: For preprocessing of data TextBlob is used. TextBlob is a package installed in Natural Language Toolkit. The preprocessing steps include tokenization, part of speech tagging and parsing. The stop words are removed using python programming. The stop words corpus which consist of 2400 stop words and which is distributed with NLTK have been used. RapidMiner is used for finding polarity and subjectivity of tweets. TextBlob is used for finding the polarity and subjectivity confidence. Weka is used for calculating the accuracy of tweets based on Naive Bayes classifier and SVM classifiers. Findings: The paper provides the polarity of tweets which include whether the tweet is positive, negative or neutral. Polarity confidence and subjectivity confidence are also found. Accuracy of tweets are found using Naive Bayes and SVM classifiers. Applications: Sarcasm Detection could be helpful in analyzing the exact opinion of the user about a certain thing. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|