期刊名称：International Journal of Advances in Soft Computing and Its Applications
印刷版ISSN：2074-8523
出版年度：2019
卷号：11
期号：1
页码：112-130
出版社：International Center for Scientific Research and Studies
摘要：Combining of local and foreign language in single utterance has become a norm in multi-ethnic region. This phenomenon is known as code-switching. Code-switching has become a new challenge in sentiment analysis when the Internet users express their opinion in blogs, reviews and social network sites. The resources to process code-switching text in sentiment analysis is scarce especially annotated corpus. This paper develops a guideline to build a code-switching subjectivity corpus for a mix of Malay and English language known as MY-EN-CS. The guideline is suitable for any code-switching textual document. This paper built a new MY-EN-CS to demonstrate the guideline. The corpus consists of opinionated and factual sentences that are constructed from combination of words from these the languages. The sentences were retrieved from blogs and MY-EN-CS sentences are identified and annotated either as opinionated or factual. The annotated task yields 0.83 Kappa value rate that indicates the reliability of this corpus.
关键词：Annotation guideline;code-switching corpus;sentiment analysis;subjectivity corpus
