Literature review on sentiment analysis (Research Paper Sample)

Literature Review on Sentiment Analysis of Twitter Data on 2012-2013 Cyprus financial crisis Abstract Sentiment analysis has attracted a lot of research in recent years. Sentiment analysis over twitter has now offered organizations and governmental institutions a fast and effective means of monitoring public feelings and emotions towards their business, services, brand, employee, and so on. In this paper, we examine twitter sentiments regarding the 2012-2013 Cyprus financial crisis and we shall measure the correlation between negative and positive sentiments Literature Review Sentiment analysis is also called opinion mining (Liu, 2012) and it is the field of study that collects and analyzes people sentiments (opinions, evaluations, beliefs, attitudes, and emotions) towards such entities as organizations, political topics, personalities, products and services, issues, events, and many other things. Sentiment analysis, unlike any other medium of market research represents a much larger problem space. Sentiment analysis is also known by other names like opinion extraction, sentiment mining, subjectivity analysis, emotion analysis, review mining, and affect analysis (Liu, 2012). But in most cases, the terms sentiment analysis and opinion mining are always interchanged, especially in academia. Sentiment analysis term first appeared in (Nasukawa & Yi, 2003), while the term opinion mining appeared for the first time in (Dave, Lawrence & Pennock, 2003). From there, the terms have been used in numerous research papers particularly in the field of market research. Are there areas where sentiment analysis have been applied? Of course there are plenty. A good example is (Liu et al., 2007) who used a sentiment model to predict sales performance. Another area where sentiment analysis was used was in (McGlohon, Glance & Reiter, 2010), where reviews were used in ranking products and merchants. (Hong and Skiena, 2010) also used sentiment analysis to establish the relationships between public opinion and the NFL betting line. The first time twitter sentiments were linked with public opinion was in (O'Connor et al., 2010). Sentiment analysis was also shown by (Bar-Haim et al., 2011; Feldman et al., 2011) to prove that it can be used in microblogs to show how stock prices perfomed. In terms of social relations, (Groh & Hauffa, 2011) showed how sentiment analysis can be used to characterize social relations. Twitter network was started in 2006 and since then, it has grown tremendously; in fact, it is among the internet’s fastest growing social network. It now has over 600 million users with over 2 billion tweets per day. Just like any other social network, twitter is one place where people meet to share their stories, experiences, joys, frustrations, and so on. As a results, twitter has become a powerful research place for researchers and marketers. Twitter is basically a microblogging platform that allows users to post or retweet messages of no more than 140 characters in length (Deitrick & Hu, 2013, p20). Users can also send links from other sites through abbreviated URLs, known as short url code. The network also has specific constructs such as replies, retweets, mentions and hashtags. Hashtags are words that denote a specific trending topic and it is preceded by # symbol. All these constructs are meant to increase interactivity among users. Users can opt to follow their favorite twitter users in order to get updates of their posts. Social media forums such as Facebook and Twitter are used by consumers for different engagements in real time. The engagements may range from casual chat to business engagements. This type of interaction has offered an unprecedented business opportunities to companies in terms of marketing and advertising. People across all nationality, gender, race, class, religion, and social status often use the social media to share their experiences and impressions about things that affect them in life. This includes things like political opinions, economic debates, as well as other international debates like the Cyprus Financial crisis of 2012-2013. For instance, people used social media networks like twitter and Facebook to express their concerns, opinions, and views about the financial crisis that hit Cyprus in 2012-2013. Notable among them is Twitter. To police makers, this is a perfect opportunity to analyze the sentiments by mining them and studying the keywords used. As (Kaplan and Haenlein) notes, explosions of user generated content on social sites present a unique challenge in collecting, analyzing, and interpreting textual content since the content/data is dispersed, and disorganized. Thus, sentiment analysis is a tool that has the capability of overcoming such challenges because it helps to systematically extract and analyze online data without consuming a lot of time to manually look for the relevant data. Marketers and policy makers can use sentiment analysis to learn about consumer feelings, public opinion and attitudes in real time regardless of the bulkiness of the data and the data structure. Opinions in social media have been central in human, organizational, and every aspect of human interaction activities as they are key influences of behaviors (Liu, 2012). So, whenever a decision needs to be made, other people’s opinion needs to be considered and this applies across all walks of life and across all business and political environments. Individual customers might also be interested in knowing about the opinions of others such as product reviews. So, social media has given users a venue to express and share their thoughts, feelings, and opinions on all kinds of subjects, topics and events. With its 600 million users and over 250 million tweets per day, twitters has quickly become one area where organizations are turning to in order to mine important data which can be used for sentiment analysis. Business organizations, non-governmental organizations, policy makers, and religious institutions, among others can use twitter to understand the mood of the users and reposition themselves in terms of brand, policy making, among other things. Sentiment analysis is far from being an easy problem. In fact, sentiment analysis is much harder than the normal text review documents (Saif, He & Alan, 2008). According to (Deitrick & Hu, 2013,p1) Sentiment analysis has gone on to become one of the most popular tools for analysis of social networks and that "Sentiment analysis is often formulated as a two-step problem, in which it is first necessary to determine whether a given text is subjective or objective.” Sentiment analysis, sometimes called opinion mining can be applied in a wide variety of real life situations such as business marketing research, political opinion analysis, and general research by scholars. Thus, there has always been a need to develop tools and systems that cannot only simplify work, but offer the most accurate analysis in the shortest time possible. In other words, the system must be able to process subject contents in a document (Liu 2011). (Deitrick & Hu, 2013, p1) postulates that many methods of harvesting critical information from social network’s immense amounts of user generated content has been successfully applied to many of the real life issues and topics such as business marketing and both local and international politics, to name just a few. Sentiment analysis uses sentiment classifications to enhance community detection and community segmentation to allow deeper analysis of sentiment data (Deitrick & Hu, 2013, p1). There two main concerns with using sentiment analysis as a marketing tool. Firstly, academics and marketers as well have realized the profound influence of social networks on consumer behavior (Park, Lee & Han, 2007; Gruen, Osmonbekov & Czaplewski, 2006). However, the only challenge is learning how to handle the enormous amount of data generated. Secondly, recent technological developments have enhanced user friendliness and classification accuracy thus the marketers and academics can collect large amount of data without being obstructed or contaminated by the presence of an external researcher, but such a tool would be costly though sampling errors can be reduced and readability of the results greatly enhanced. In marketing research, there are two basic types of research methods as purported by marketing research literature: Quantitative market research and Qualitative market research (Newman, 2011). Quantitate research methods are normally used by researchers to test hypothesis and it is typically considered to be the more scientific approach to conducting social science research (Tewksbury, 2009, P1). With Quantitative research method, the focus is normally of specific ideas and is concerned with finding out what particular variables and concepts mean. (Creswell, 1994) defines quantitate research methods as an inquiry into a specific identified problem based on a theory to be tested. It is comprises of variables measured with numbers, and analyzed using various statistical techniques. The goal of quantitative method is to determine whether a particular hypothesis/predictive generalization is true. Social scientist find quantitative methods to be very useful because it helps to gain in-depth understanding of a particular phenomenon. Qualitative methods of on the other hand, are used to gain an in-depth understanding of issues that may not be understood using the quantitate methods. Thus, qualitative researchers are interested in knowing the meaning people have made, that is, how people view things and make sense of their world as well as their experiences (Merriam, 2...