Powerset Natural Language Engineering
The Natural Language Engineering (NLE) team at Powerset (part of Microsoft’s Bing Search) develops the ParGram English grammar and semantics for search applications, including the Bing reference vertical.
Applications include:
- Semantic indexing for Wikipedia and other reference content
- Tuples extraction from natural language text for answers-like products (e.g. Factz)
- Improved caption (snippet) choice and readability
ParGram/ParSem related projects include:
- General grammar, morphology, tokenizer, and named entity detection development
- Domain adaptation
- Stochastic ranking
- Semantics development
- Word sense disambiguation
- Training data creation and component evaluation
Contributors:
- Manager: Tracy Holloway King
- Text Processing for Semantic Applications: Dick Crouch (lead), Martin van den Berg, Ascander Dost, Martin Forst, Olya Gurevich, Stuart Robinson, Rajesh Venkataraman, Scott Waterman, Yingwei Zhang
- Statistical Text Processing and Classification: Valerie Nygaard (lead), Chris Biemann, Joanna Chan, Will Fitzgerald, Riccardo Turchetto, Robert Voyer
- Semantic Services Platform: Jim Firby (lead), Dave Fayram, Jim Kellerman, Michael Stack, Justin Trobec
- Corpus Development and Evaluation: Chris Walker (lead), Alexis Baird, Hannah Copperman
- Other Microsoft contributors: Ron Kaplan, Livia Polanyi