This paper is an interest-type piece from a 2005 IEEE publication. It is an interest-type piece in as much as it is not written in academic-prose but instead in a more reader-friendly and informal manner. This being said, it is highly informative and provides great insight into the lingo algorithm for clustering search results.
The authors (Osinki & Weiss, 2005) identify that in popular search engines search results match the searcher's question, rather than attempt to answer the search question themselves. This notion forms the premise for the rest of the paper, in which they describe the algorithm (the Lingo Algorithm) that facilitates the separation of search results into meaningful groups.
The Lingo Algorithm has a number of well-defined phases:
1. Preprocessing;
2. Phrase Extraction;
3. Cluster-Label Induction; and
4. Cluster-Content Allocation.
Finally, Osinki & Weiss evaluate the Lingo Algorithm through emperical and analytical evaluation. They explain experiments carried out and summarise the results of these experiments.
No comments:
Post a Comment