Description:
In this video from Microsoft Research,Chris Biemann, a natural language scientist from Powerset, summarizes Powerset's endeavor to set up a flexible and data-driven approach to handling word senses. In a traditional keyword search engine setting, word sense disambiguation is believed to play a subordinate role. While a keyword query tends to disambiguate itself through the presence of other keywords, e.g. “flying jets” vs. “ny jets,” this is not the case in an index expansion setting, where wrong sense expansions lead to spurious semantic matches.
Speaker(s):
Chris Biemann, natural language scientist, Powerset
|
