Long-time readers (really long-time readers) will know of my affinity for similarity - it was similarity that led me to connectionism and networks. To me similarity still forms the semantic basis for categorization, causation and other predictive events. So this sort of article, which outlines a proposed similarity measure for clustering web pages, speaks to me. And it's not a long jump from calculating keyword weights to calculating weights of connections between them. More from the most recent IJICT, just out.
Today: 5 Total: 24 [Direct link] [Share]
View full size
