|
RDF-F3
Navy SBIR FY2010.2
| Sol No.: |
Navy SBIR FY2010.2 |
| Topic No.: |
N102-176 |
| Topic Title: |
RDF-F3 |
| Proposal No.: |
N102-176-0106 |
| Firm: |
Modus Operandi, Inc.
709 South Harbor City Blvd., Suite 400
Melbourne, Florida 32901-1936 |
| Contact: |
Kent Bimson |
| Phone: |
(321) 473-1446 |
| Web Site: |
http://www.modusoperandi.com |
| Abstract: |
Navy operators are being inundated with intelligence information, especially text. Better methods and tools are needed to help them identify the mission-relevant "needles" within these information "haystacks." Researchers have made progress using ontologies--represented as RDF triples--to guide text extraction algorithms and as a reference model for transforming extracted information into RDF triples corresponding to the ontology. One of the major hurdles encountered in this research, however, has been identifying and eliminating redundant statements in the knowledge base. Our goal in this Phase I Project is to design a methodology and architecture for preventing the assertion of redundant RDF statements to knowledge bases based on comparing the lexical semantics of RDF statements with the lexical semantics of text statements and preventing redundant assertions to the knowledge base, based on synonymy. The concept of operations can be simplified to three steps: (1) Find mission-relevant information in text, based on ontology semantics; (2) Format the extracted information and RDF triples as lexicalized; and (3) Filter statements by comparing lexicalized triples to lexicalized text, preventing redundant RDF triples from being asserted to the knowledge base. We call this the RDF Redundancy Find, Format, and Filter process, or RDF-F3. |
| Benefits: |
RDF-F3 will benefit the Navy by helping analysts prevent redundant knowledge, in the form of RDF triples, from being asserted to the knowledge base, especially when that new knowledge has been extracted from text sources. It will benefit science by providing a way to map lexical semantics to the concepts and relations of an ontology, helping to solve the problem of mapping one relation in an ontology (or RDF triple) to the many different synonymous words and phrases that can represent the same meaning in natural language. The one-to-many ontology:natural-language semantic mapping challenge is a significant hurdle to the use of ontologies in the extraction of essential elements of information from text sources. RDF-F3 provides a way to change that to a one-to-one lexical and semantic mapping problem. RDF-F3 provides a formal definition of redundancy and metrics to measure it, whereby redundancy is based on the semantics of synonomy, rather than on graph theory or logic. |
Return
|