Algorithms and Tools for Big Data Sentiment Analysis
Abstract
Business and organization constantly looking for ways to improve their performances and to extend their customer reaches. In the age of social media where people communicate and share their thoughts in all aspects ranging from politic to personal perferences in the form of blogs, posts, video and etc. Such artifacts which are publicly accessible can be harnessed with the power of big data analytic so that more insight of how those people collectively think about some issues or topics of interest could be collected.
By following series of academic researches, many business enterprises began to realize the benefit of understanding regarding crowd's sentiment especially toward their products and services. Consequencely, this caused disruptive ripples that affect corporates all over the world where, traditionally, most products and services are developed by experienced sales and marketing personels who might unconsciously introduce biases into the final outcomes. In contrast, adopting more scientific approaches to such development in combining with new technologies such as data science and big data analytic would make new product/service to be more aligned with the real need or more desirable to their prospects. Hence, learning how to analyze people's opinion and feeling is by all mean mandatory for those who seek real insight.
However, journeying on the learning path is quite a bumpy road. First of all, one must find reliable sources of opinion data which usually acquired from social media or ecommerce sites. However, these sites usually enforce some form of security related requirements such as registration, access policy compliances and communication protocols that prove to be non-trivia obstracle from time to time.
Moreover, after data acquisition, one still must carefully choose the right set of suitable algorithms and tools to perform analysis on such data. This is definitely not an easy task since understanding some particular algorithms takes time and experience. Even with savvy data scientists might have to try out different algorithms to find the right one or ones.
All those aforementioned reasons led to this compilation of useful information on sentimental analysis such that new comers would take less effort on their learning journeys. We start by introducing what are text mining and sentimental analysis. From there we continue to explore primary sourcees of data such as those from popular social media and e-commerce sites in addition to some details on how to access them. Next, we describe and summarize some interesting algorithms and useful tools. Finally, we conclude with discussion on how to further improvement analytic process's flow by incorporating new technologies such as cloud and cluster computing.