TakeLab participated in two tasks at this years international workshop on semantic evaluation (Semeval). Our systems did fairly well (within top 5!) on both paraphrase detection in twitter and medical text analysis.
The twitter paraphrase detection task consists of recognizing whether a given pair of tweets are semantically equivalent. You can find more details and examples on the Semeval 2015: Task 1 page. Our system (TweetingJay) ranked 4th out of 16 systems. The system is also freely available for download.
The medical text analysis task consisted of finding medical named entities in text and mapping them to identifiers in a medical thesaurus. A detailed description of the task can be found on the Semeval 2015: Task 14 page. Our submission (MINERAL) ranked 4th. Be sure to check out the online demo of the system.