|
Title:
|
FINDEX: PROPERTIES OF TWO WEB SEARCH RESULT CATEGORIZATION ALGORITHMS |
|
Author(s):
|
Mika Käki |
|
ISBN:
|
972-8924-02-X |
|
Editors:
|
Pedro Isaías and Miguel Baptista Nunes |
|
Year:
|
2005 |
|
Edition:
|
1 |
|
Keywords:
|
Web search, algorithm, categorization, clustering, information access. |
|
Type:
|
Full Paper |
|
First Page:
|
93 |
|
Last Page:
|
100 |
|
Language:
|
English |
|
Cover:
|
|
|
Full Contents:
|
click to dowload 
|
|
Paper Abstract:
|
The vast amount of information in the web combined with the short queries submitted to the web search engines may cause the result ranking methods to fail. Result categorization methods have become popular in solving the problem. We have developed Findex search user interface and two simple algorithms for this purpose and we have showed their utility with experiments. However, the details of the algorithms have not been discussed. The algorithms are based on word frequencies in the result summaries (snippets) the web search engines typically return. After extracting the most frequent words and phrases the two algorithms use different ways to filter out uninteresting candidates and to merge the similar ones. We evaluated the algorithms heuristically and empirically. The results show that the algorithm utilizing the query term contexts produce more understandable category names, but the simpler statistical approach offers better coverage and performance. Both algorithms perform well under certain conditions and help users in accessing the search results. |
|
|
|
|
|
|
