Evidential Search Engine
Published in 2006
The search for “Nike” is ambiguous in context, as this is a company that manufactures sports clothes and equipment also this can be the winged goddess of victory.
Searching for this with MS search will normally have Nike the sports brand come in the higher places, due to the search engine works on a social network (site rank) and the sports brand is normally the most popular choice.
The ESE engine does not take the social status of the website (how many other sites link to it, or how popular it is). This engine uses a semantic network to find relating words to the search term’s context.
Then it looks for the existence of these words in the results of the MS search to reorders the results on their context.
The screen shots show The ESE engines top 4 results, the 3 top are to do with the chosen context.
And the MS Search top 8, none are to do with this other context. (results taken in 2006)
Please note that the ESE engine re-sorted a list very biased to the most popular context.
The ESE engine included:
MS search engine Web services. This was the crawler; it finds the pages which could be of significance
An indexer engine, based on the X^2 measure with co-occurrence. This downloaded the pages in the MS search results and generated a keyword list. (Like an index at the back of a book) DBA, this stored the pages indexed information. So they would only have to be indexed once.
The DB Access layer was the business Object level.
Lexical DB, WordNet with the WordNet.Net class library was used as the dictionary to find the relating words.
Engine, this was a simple engine which reordered the MS results by context.
HCI, separated GUI, very simple design, mainly to test the engine, as the GUI was not the project the engine and other parts are.
The UI, was not the purpose of this project, the engine was.
