Abstract
Based on theories of associations, the Semantic Associative Network for Text Analysis (SANTA) has been developed. Nodes in the network represent words and links between nodes represent the association strengths between them. The links are adjusted by a learning function based on the co-occurrence of the words. Given textual samples, SANTA can determine which words have greater association. Using this information, semantic keywords for text can be suggested and textual similarity can be determined. As the number of samples increases, so does the need for finding relevant documents. In line with this need, the ability to find relevant documents increases since better associations are learned given the additional samples. Ostensive retrieval is the goal in that searching for documents is initiated by first presenting a relevant item. Keyword searching is also available. Evaluations of SANTA suggest much promise for this approach.
Airhart, Robert William (2000). Semantic associative network for text analysis (SANTA). Master's thesis, Texas A&M University. Available electronically from
https : / /hdl .handle .net /1969 .1 /ETD -TAMU -2000 -THESIS -A385.