Site Reference Forums

Seems a silly name for an article that never answers its own question, but I gotta know from those who understand LSI, is Mr. Nisbet right? It seems reasonable, but I couldn't find his old article to compare. Also, found one little flaw in it that left me asking this question --

I have read several LSI papers and reviews, written by mathematical statisticians, that attempt to explain the subject to the layman. This was achieved with extreme difficulty, and I doubt that anyone who is not an expert on semantics fully understands what the term means.

A mathematician is to semantics what a rocket scientist is to an English professor. I understand (as best my untrained mathematical mind allows me) that SERs will always have to be collected using algorithms, but understanding semantics "fully" isn't necessarily helping anyone understand the math involved, nor is it necessary to understand what each of the three words mean -- "latent," "semantic," and "indexing." Mathematics and semantics are two fairly unrelated fields, which now seem to need to be merged.

So, for those who know LSI (or think they know, according to Mr. Nisbet), what do you think of the article?