Article Preview
Top1. Introduction
Internet has been considered as a medium to exchange information or knowledge. Due to the rapid growth of information online, almost every activity in human life has become automated. For instance consider a user, who wishes to purchase a music system or digital camera. The user can sit comfortably in his place and browse for the best products through online websites which provides about each products. Such commercial sites allow people to express their opinions on the products that they have purchased. Hearst (1992) and Wiebe (1994) originally proposed the idea of mining direction-based text, namely, text containing opinions, sentiments, affects, and biases.
The current scenario is that, social web-forums allows each individual to post review which may be a discussion groups or blogs which enables the new user to buy product based on the views of expressed by different users. These reviews are useful for both customers and manufacturers. Opinions or sentiments about the products can be analyzed about these products and their performances can be analyzed effectively. Opinions are hidden in long forum posts and blogs, which allows human to find relevant sources, summarize them and organize them into usable forms. Sun et al. (2009) proposed the system for sentimental classification and comparison of products from both subjective and objective perspective on various feature levels.
To make decisions about online reviews, opinions are so important and we should take some special interest in mine such positive and negative opinions. Review comments from the sites are considered as valuable because they cover a lot more products than those formal review sites. Facts and opinions are the categories of textual information. All entities and events in the world have objective statements that are called facts. On contrary subjective statements that reflect people’s sentiments or perceptions are called opinion. Recent research on such opinion mining and sentimental analysis have been performed based on unsupervised learning approach (Popescu & Etzioni, 2005), analyzing lexical features and their relationships (Riloff et al., 2006), sentimental word based scheme (Cai et al., 2008) and frequency of occurrence of product features (Wong & Lam, 2005).
Opinion or sentiment bearing words (e.g., great, amazing, wonderful, bad, and poor) are mainly used in the research on opinion mining. As per the review, work on mining such opinion bearing words and to identify their semantic orientations (i.e., positive or negative) helps both merchants and buyers (Su et al., 2008). Since a product may have hundreds or even thousands of reviews it is difficult for buyers to choose right decision and also for manufacturer to keep track of change if products of different kinds are produced. Hu and Liu (2004) aimed to mine and to summarize all the customer reviews of a product in a system. Only those opinions that are given by customers as positive or negative for each product is summarized; this makes the task different from traditional text summarization.
User generated contents are created and published by the end users who surf the web on a daily basis. Examples of such online documents are blogs, newsgroup postings and discussion forums. There are two types of textual information available on the web namely facts and opinions. Facts are objective expressions about entities, events and their properties. Facts can be expressed with topic keywords. Currently available search engines search only for facts and they are not appropriate for opinion retrieval or search. But the user generated content on the web such as personal experiences and opinions about a product or a movie in the form of reviews play a very important role in business, education, e-commerce, etc. Online review websites allow users to express their opinions for the information they are interested in. So there is a huge amount of information available online, however they fail to provide the knowledge about the products.