Social media analytics

This week's ebiquity meeting will focus on social media and two research efforts that are part of our Relief Social Media project.

Mohit Kewalramani will present the topic that he is addressing in his MS research. An important task in analyzing highly networked information sources like Twitter is to identify communities that are formed. A community can be defined as a group of nodes that have more links within the set than outside it. We plan to present a technique for detecting communities in Twitter using link structure, folksonomy, text and other available metadata of the tweets. The link structure can be determined using the Twitter notion of followers, being followed and the @Mentions and @RT tags in tweets. We intend to use the normalized cut algorithm along with dimensionality reduction for clustering. The communities thus created can be characterized using the dominant tags, sentiment associated with these tags and other meta data used for clustering.

Akshaya Iyengar will describe the topic of her MS research on integrating information from multiple social media streams. Over time the information found in Twitter status messages has changed. From simple text messages, we now observe increasing messages with links to images. We have tweets which essentially similar in content but point to different images. If the information linked with these images were different, then they would be classified in different groups. The idea is to apply topic modeling on tweets and use it to group images, using it to improve classification of images. Also by implementing it in reverse we can group images and then using that information to better classify Twitter messages.