Existing information extraction IE systems extract entities from text such as people, organizations and locations and deposit them into a structured data representation capable of being queried and manipulated. IE systems use statistical techniques to identify entities and relationships found in a sentence and these entities and relationships can then be used to update a knowledge-base (KB). In contrast NLP systems parse sentences to reveal lexical relationships as a method of identifying constructs such as subject-verb-object for a sentence, apply a grammar and generate output that ultimately can update a knowledge base (KB).
IE and NLP systems have a common thread. They use a two-phased approach. The first phase is syntactic (e.g.; apply a part-of-speech tagger and a syntactic parser) and the second phase is semantic. It takes the output of the first phase and updates a KB. The details of how the second phase works for an NLP system is very different form that of a IE system, but it still is a second phase.
Our approach uses a single pass to produce an ontological representation of the meaning of the sentence. The same approach is used to interpret non-linguistic phenomena. It is based on a combinatory ontological grammar.
For more information, contact ethomsenATetworks.org