Extracting Paraphrases from Aligned Corpora

The Problem: The expressiveness of human language allows people to express the same idea in many different ways; they may use different words to refer to the same entity or employ different phrases to describe the same concept. Thus, an effective information retrieval (IR) and question answering (QA) system must be equipped to handle these variations, both when processing documents and when fielding queries. While there are many resources that help systems deal with single word synonyms, there are few resources for multiple word paraphrases.