THIS IS A WHITESMOKE PRESS RELEASE
Subscribe to the WhiteSmoke Press Feed:

---

English Writing Software Hits a New Stride

WhiteSmoke Announces Second Generation Grammar Engine

NEWARK, Delaware, June 10, 2007 / WhiteSmoke Inc. - This morning WhiteSmoke announced the release of their second generation grammar engine for WhiteSmoke Writing Software. The new grammar engine furthers WhiteSmoke's commitment to provide users with the most advanced grammar checker available. Three years of intensive research and development have gone into enhanced detection algorithms for missing words, agreement errors, contextual spelling and verb conjugation. This release also features faster analysis times, improving overall usability.

"We do not target the easily marketable or computationally-simple grammar checking features. We know the types of mistakes that people actually make, and constantly refine our algorithms to better detect them," says company co-founder Liran Brenner. "For this release, a lot of hard work has gone into correcting grammar in complex sentence constructions, something which has eluded grammar checking software until now." The new algorithms developed by the WhiteSmoke linguists and programmers were created in tandem with Dr. Ido Dagan, a world expert in Natural Language Processing (NLP). Statistical information drawn from texts written by native speakers enabled the design of complex algorithms that detect obscure and hard to identify grammar errors, and build the best predicted result.

One of the key areas in the research and development of this release was that of missing words within a sentence. WhiteSmoke found that over a third of all mistakes in written English result from missing words. In many cases, prepositions, determiners and auxiliaries are left out of sentences. To retain the meaning of a sentence with missing words, the grammar correction must take into account the sentence context. WhiteSmoke can now accurately detect many unique cases of missing words, even in complex sentence constructions, and suggest the appropriate solution.

Another area of research for this release was agreement-related problems within a sentence. The second generation grammar engine features enhanced detection algorithms for subject-verb agreement, number (determiner-noun) agreement and auxiliary-verb agreement. On analyzing a sentence, the extensive database of written English examples is cross-referenced for related entries. Then, by situating the appropriate parts of speech in the context of the sentence or phrase, agreement problems are corrected with great accuracy and speed.

Combined with a number of other minor enhancements, these features take the WhiteSmoke grammar engine's artificial intelligence to an unprecedented level of accuracy in grammar correction. The second generation grammar engine is effective immediately for all current WhiteSmoke users.