Content area
Full Text
Abstract
Sentiment analysis (SA) is a technique of textual data that uses Natural Language Processing (NLP) and Machine Learning (ML) to evaluate text automatically for the writer's feelings (positive, negative, and neutral). The lexicon-based approach is used to extracting sentiment from text and user reviews. In the sentiment analysis task, the sentiment lexicon, which offers sentiment polarity in terms, plays an important role. Most sentiment lexicons currently have only one polarity of sentiment for each word and disregard sentimental complexity. The problem of Sentiment Analysis was well studied and two main approaches were developed namely corpus-based and lexicon-based approaches. This paper discusses lexiconbased approaches to sentiment analysis. Contextual words, Acronyms, and emoticons are the major problems in sentiment analysis. The proposed techniques to improve the accuracy of sentiment analysis and also analyze the contextual words, acronyms, and emoticons.
Keywords: Sentiment Analysis, Lexicon-Based Approaches, Acronyms, Emoticons, Contextual Words, Natural Language Processing.
1.Introduction
Sentiment analysis is an evolving area of processing the natural language based on the interaction between humans and computers, extraction of information, and distillation of feelings from ever-increasing online social data. It includes recognizing the words or phrases indicating a positive, negative or neutral attitude in the underlying text. Sentiment analysis generally extracts various characteristics from structured or unstructured textual data and analyzes them to get thoughts, opinions, and feelings out of it. In this internet era, it is relatively easy to get the voice from customers or stakeholders through various channels, such as blogs, online forms, social media, customer service, and many more [1]. Typically, up to three different levels can be used in the sentiment classification namely (i) Document-level classification, (ii) Sentence-level classification, and (iii) Aspect-level classification. This research work performed a sentence-level sentiment classification in the research experiment. In several reviews, in a single product or service review, people express more than one opinion, typically distributed in various sentences.
In addition, lexicon-based methods are also popular in the study of sentiment, which takes into account the semantic orientation of words in a text and calculates sentiment. In this strategy, a dictionary of positive and negative terms is created where a sentiment value is assigned to each positive or negative word. These values are added to the text of the analysis,...