|
There is an excellent query expansion technique called Local Context Analysis
and it is based on the use of expansion concepts, which are noun phrases (group
of nouns with of one, two or three adjacent nouns). Concepts are extracted and ranked by their co-occurrence with terms in top ranked documents. Then the highest ranked ones are used for query expansion. Why Nouns? Because research suggests they provide more possibilities for expanding queries than other type of terms, and they also are more informative.
Business - List of business/finance/loan/mortgage resources
more 1 2 3 4 5
Computers - List of computer hardware/software/peripheral resources
more 1 2 3 4 5
Internet - List of webhosting/webdesign/internet marketing resources
more 1 2 3 4 5
Software - List of software resources
more 1 2 3 4 5
Web Design - List of web design/development resources
more 1 2 3 4 5
Web Hosting - List of web hosting resources
more 1 2 3 4 5
Web Promotion - List of search engine optimization/internet marketing resources
more 1 2 3 4 5
Web Resources - List of other web resources
more 1 2 3 4 5
Recreation - List of travel/hotel/cruise resources
more 1 2 3 4 5
Casino - List of online gambling/poker/blackjack/roulette resources
more 1 2 3 4 5
Health - List of online pharmacy/hospital/health resources
more 1 2 3 4 5
Shopping - List of online shopping/gift resources
more 1 2 3 4 5
Miscellaneous - List of all other resources not stated above
more 1 2 3 4 5
But there is a downside. Analysis of a certain block of words without transition margins can affect the presentation and flow of semantics. So, some prefer to use semantic segmentation techniques. According to semantic structure and flow, and based on document punctuation, it involves partitioning a document into topics
or sub-topics.
An alternative to the previous structures could be the use of a pre-qualified list of terms, or keyword list. Terms are typically qualified by popularity, most frequently searched terms, most frequently clicked terms, search logs or similar metrics. It is widely used, and it can be an excellent tool for finding terms.
|