Sentiment Classification of Unstructured Data Using Lexical Based Techniques

Abstract

Sentiment analysis is the computational study of people’s opinion or feedback, attitudes, and emotions toward entities, individuals, issues, events, topics and their attributes. There are many research conducted for other languages such as English, Spanish, French, and German. However, lack of research is conducted to harvest the information in Malay words and structure them into a meaningful data. The objective of this paper is to introduce a lexical based method in analysing sentiment of Facebook comments in Malay. Three types of lexical based techniques are implemented in order to identify the sentiment of Facebook comments. The techniques used are term counting, term score summation and average on comments. The comparison of accuracy, precision and recall for all techniques are computed. The result shows that the average on comments method outperforms the other two techniques.

Rule-based Text Normalization for Malay Social Media Texts

“…Additionally, the studies conducted by [14], [15], [16], and [17] are using only four pre-processing techniques [10]. The authors in [14]'s used pre-processing methods such as capitalization, spam elimination, emoticon elimination, and symbol removal.…”

“…Additionally, the studies conducted by [14], [15], [16], and [17] are using only four pre-processing techniques [10]. The authors in [14]'s used pre-processing methods such as capitalization, spam elimination, emoticon elimination, and symbol removal. [15]'s analysis used capitalization methods, selecting only tweets written or containing the language of analysis, filtering tweets that did not suit the report's features, and selecting only tweets that could be encoded and decoded.…”

Sentiment Analysis of Noisy Malay Text: State of Art, Challenges and Future Work

“…In another study by [2], they introduced a lexical based method to analyze sentiment of Facebook comments in Malay language. They created two types of score dictionary which are also known as sentiment lexicon namely Malay Adjective score dictionary which only contained the adjective of Malay words and Malay-English (BM-ENG) score dictionary where the English words in it were the translation of the Malay adjectives words.…”

“…The Malay Adjective score created by [2] has been updated by [14] in their study which classified the sentiment of Facebook comments in Malay language based on the adjectives, verbs, adverbs, and negation. They used the term ''score dictionary'' for their version of sentiment lexicon.…”

“…by the users are very valuable for the high-level administration in any organizations to study and understand on how to improve their services. For example, the higher educational institutions can know the students feedback on their new programmes [1] or the high level administration of any local government agencies can know how the public responds to whatever move that they made [2]. Hence, to identify whether the user's responds are neutral, negative or positive, the implementation of Sentiment Analysis (SA) is needed.…”