SweetTweet: A semantic analysis for microblogging environments

Abstract

User collaboration became the key factor in the development of today‟s Internet applications with the emergence of Web 2.0. Users not only consume the services available on the Internet, but also interact with them and collaborate to provide content generation for the services. Microblogs are recently one of the most interesting applications in the Internet. They are rapid, simple and easy to use when compared to the traditional blogs. These properties of microblogs create user interest and increase the popularity of these services. Twitter is the most popular microblog and it has millions of users posting millions of messages every day. The data available on Twitter is massive and it is growing continuously. This massive data contains valuable information. The work done in this M.S. thesis is to provide a methodology to categorize, analyze this data, understand the user contributions made to microblogs and export valuable information. However, microblogs have some limitations, especially on the size of the content. Same situation also applies for the user posts in Twitter, which are also known as “tweets”. This makes the analysis of the data on Twitter more challenging, since the only information we have for performing an analysis are the words in user tweets. First step in our method is to retrieve user tweets and parse them into words. Next, we need to analyze and understand the content of the user posts. To achieve this goal, we utilized Semantic Web resources. DBpedia, which is a central node on Linked Data effort, is selected as Semantic Web resource in this thesis work. DBpedia provides the data on WikiPedia in RDF format and it has an interface that enables us to perform complex SPARQL queries on the data set available on it. The model we proposed in this thesis work takes the words which are used frequently on users‟ posts as input, queries them on Semantic Web resources and finds out the matching categories defined on this resource for these words. At the end of the analysis process, we have a group of category names for the users, which enables us to understand their contributions made to microblogs.