The Web is filled with acronyms, and you can’t get away from them. Many queries searched for on the Web contain acronyms.
Often people will search using an acronym, and then perform another search with the acronym expanded. Can a search engine learn from such queries by looking at pairs of searches like that, first for an acronym, and then for the expansion of the acronym?
In the top ten results of a search on Google for [BAC], I come up with:
- Bank of America
- Blood alcohol content
- International Union of Bricklayers and Allied Craftworkers (BAC)
- Boston Architectural College
- Baltimore Aircoil
If I were searching for the [Boston Architectural College], and used the acronym for it first, I might be tempted to expand the acronym to the full name of the school, in a new search.
If a search engine looks at how people expand acronyms like that, might it be helpful in determining what results to show first when someone does search using an acronym?
A paper from Alpa Jain of Columbia and Microsoft’s Silviu Cucerzan and Saliha Azzam, Acronym-Expansion Recognition and Ranking on the Web (pdf), aims at building a collection of acronyms and expansions to go with them.
It looks at information about acronyms gathered from three different sources – Web crawls, search engine logs (looking at those pairs of expansion sessions), and search results.
While this paper doesn’t go into detail on what one might do with a collection of acronyms and expansions, what is interesting is the interpretation by the researchers on the value of expanding acronyms from these three different sources.
Information about acronyms from a Web crawl is more likely to provide a wider range of expansions for acronyms. Information gathered from search engine logs and search results is more likely to show what current searchers might be interested in finding when they perform a search using a specific acronym.