Today was another meeting with my supervisor. She suggested some minor corrections which i will implement within the next days. After that, the chapters Background and State-of-the-Art are finished.

During the last weeks i created a data set that is going to be used for the evaluation task. I used a freely available data set consisting of around 4411 tweets that where manually annotated based on their sentiment. I enhanced the data with scores from the TweetCred project that provides a ‘per-tweet’ credibility value. This data set will be used to evaluate user perception of tweet credibility. The next step is to send out the link to the volunteers that are going to annotate around 250 tweets and define the individual percepted credibility. For this task i created an evaluation platform based on SailsJS, AngularJS and MongoDB.

Currently i am working on a method to provide a baseline for the evaluation. I’d like to create a simple ‘naive’ feature set to easily classify tweets as credible or not.

Results will follow…