SCS Undergraduate Thesis Topics

This project examines the problem of extracting structured knowledge from unstructured free text. The extraction process is modeled after construction grammars, essentially providing a means of putting together form and meaning. The knowledge base is not simply treated as a destination, but also an important partner in the extraction process. In particular, the ideas are implemented as a module closely tied to the Scone Knowledge Base system. I demonstrate that with a reasonable knowledge base and general construction rules, one can easily extract structured knowledge. This project also explores partial matching, word sense disambiguation and generalization in the context of constructions.