NLP lab receives support from Yahoo! to mine microblogs
Prof. Di Eugenio has received a $10000 Yahoo! Faculty Research and Engagement award to support the project "Mining life-stage events and user personas in microblogs”. The project focuses on:
1. Capturing life-stage events. We want to be able to answer questions such as "Is this user engaged / married/single?" (relationship status); "Is this user socially active in the real world (frequent bars, sporting activities etc.)?"; etc. Neither data mining and NLP state-of-the-art techniques are sufficient, since e.g. Twitter messages are highly ambiguous wrt these events and who is the actor of those events (as in the following two public tweets "ackorie Sequari Andrews 4 the record people I am not nor am I anywhere near engaged!!"
Vs " Aftenthurston Aften Thurston Still cannot believe I am engaged!! yfrog.com/h3u2vsqj")
2. Studying the relationship (if any) between user's lifestage events and/or personas and online behavior (webpage view activity, search activity). In the process, we will also build predictive models to identify other such users based on behavioral data alone. Specifically, we will use the mined set of users as our labels to train a classifier to identify other similar users (look-a-like modeling).