Sentiment analysis of twitter data in Malay language (Bahasa Melayu)

Nawi, Nurnajwa Hazwani (2017) Sentiment analysis of twitter data in Malay language (Bahasa Melayu). Project Report. Universiti Teknikal Malaysia Melaka, Melaka, Malaysia. (Submitted)

Text (Full text) Sentiment analysis of twitter data in Malay language (Bahasa Melayu).pdf - Submitted Version Download (1MB)

Abstract

The main motivation of this project is to identify the sentiment values of Twitter data whether it is positive, neutral or negative. Firstly, a set of tweets are labelled manually (using human interpretation) with their sentiments and considered as training data. Then, another set of tweets that is live streaming, are collected based on the text mining on Twitter Streaming API (Application Programming Interface) and python. The tweets are retrieved and saved as a text file and later will be used as a testing set. Testing data will learn from training data’s calculation to predict sentiment value. The problem statements of this project are, there are no Twitter dataset corpus available in Malay language with labelled sentiment values. Next, finding a filter to search tweets in Malay language that is from Malaysia. Then, finding a classifier to categorize tweets into positive, neutral or negative. Major challenge of this project is to collect a labelled corpus as a training set. Since there is no labelled Twitter corpus available in Malay language, a database of sentences is manually labelled with sentiments using human interpretation and uses tweet’s geolocation to search for tweets posted in Malaysia. At the end of this project, Twitter corpus using Twitter Streaming API able to be collected. Secondly, tweets from Malaysia collected by using tweet’s geo location able to be obtained. Thirdly, there will be a Malay dataset, using the decision tree classifier can be categorized according to its sentiment value which are positive, neutral and negative.

Actions (login required)

View Item