Mining on Social Media

Introduction

Social media mining is the process of representing, analyzing, and extracting actionable patterns and trends from raw social media data. Social media is the “group of internet-based applications that build on the ideological and technological foundations of Web 2.0, and that allow the creation and exchange of user-generated content|”. There are many categories of social media including, but not limited to, social networking (Face book or LinkedIn), micro blogging (Twitter), photo sharing (Flickr, Photo bucket, or Picasa), news aggregation (Google reader, Stumble Upon, or Feed burner), video sharing (YouTube, MetaCafe), livecasting (Ustream or Twitch.tv), virtual worlds (Kaneva), social gaming (World ofWarcraft), social search (Google, Bing, or Ask.com), and instant messaging (Google Talk, Skype, or Yahoo! messenger) proposed in Jiliang and Tang, (2014). Since social media is for everyone and it became such a massive part of their life. Since it is available all over the world, social media sites had become very much popular now a day's, such as twitter, face book, YouTube, LinkedIn etc. This provides a great platform for students to express their views, emotions, opinions, joy, struggle and feelings. Everyday student’s discuss and share their encounters in formal and informal way on different social media sites. Student’s tweets or comments on particular posts provide large amount of implicit knowledge and a whole new perspective for the educational and institutional researchers, users and practitioners to understand the student’s behavior outside the controlled classroom environment. This understandings are useful for taking the decision at institutional level in taking the consideration of student’s point of view for their success. Even though social media data provides a lots of opportunities to understand student’s behavior, but still there are some methodological difficulties in making sense of the social media data for educational purposes (Hu, Xia, Tang and & Huan 2013). There are number of methods used by educational and institutional researchers such as surveys, focus groups, and classroom activities to collect data related to the student’s behavior. These methods are usually very time-consuming and not very frequent. Considering these drawbacks of existing system, a new system was proposed. In proposed system a qualitative analysis using classification algorithm instead of the sentiment analysis is performed because sentiment analysis considers the opinion of the user about a system and categorizes it into 3 different levels namely neutral, negative or positive mood but in the proposed system, we searched the information of the student’s based on the keywords such as engineer, student, campus, class, professors and labs in the twitter data as per the geo location, keyword and search id (Vorvoreanu & Q. Clark, 2010). One of the hardest task is to search keywords in twitter because of the diversity of the languages and the Internet slang used are different. By exploring more advanced information retrieval methods there are two ways of extracting data. One among them is semantic based information retrieval in which it uses semantic information to understand the documents and queries. The other method is machine learning based method which is used to reorganize Web documents such as classification and clustering. In this paper we mainly discuss on the improvement of information retrieval based on machine learning. So that we use a multi label classification algorithm which is implemented to analyze the content as per the category and the results will be reported to a decision maker which helps the person to get the overview of student’s problems and their experience in learning process, so that an institution can make a proper decision making policy to improve the performance of the students as they are the future of society.