Powerset is building a consumer search engine based on natural language processing.

Unlike other conventional search engines that index keywords, Powerset claims to do a deep linguistic analysis on every sentence it reads.

Powerset extracts and indexes facts about every sentence in every page, making its index the first truly robust semantic index of the web. Powerset will offer a transformative search experience, with better, more relevant results. Powerlabs is an online community that will contribute to the development of Powerset and allow the company a direct line of communication with consumers to better understand the potential, possibilities, and challenges of building a large-scale search engine. [Powerset]

Using Powerset’s semantic search engine, a user could ask, “When did earthquakes hit San Francisco? ” and potentially get more on-point search results than from typing in mere keywords such as “San Francisco earthquakes.” Another advantage of the Powerset approach is that it gets to the meaning of words, so queries for one word will fetch results that contain words which are synonyms.

The project is backed by a team comprised by industry leaders from companies such as Altavista, Apple, Ask.com, BBN, Digital, IDEO, IBM, Microsoft, NASA, PARC, Promptu, SRI, Tellme, Whizbang! Labs, and Yahoo!.

For now, Powerset’s is currently under development. The search engine is testing the system based only on Wikipedia content before they go for the huge expense of indexing the entire Web.