Databases Reference
In-Depth Information
5 Experimental Results
LSS clustering organizes the returned Web pages hierarchically from a user
query on-the-fly over two remote search engines (PubMed and GOOGLE).
As we already known, GOOGLE provides a flat list of search result snippets
and PubMed returns a summarized list of the abstracts of medical literatures
associated with MeSH terms. In our experiments, all the PRIMITIVE CON-
CEPTs are bound a threshold, since the simplexes in the lower dimension is
highly overlapped with several IDEA. In our system, the lower dimension the
PRIMITIVE CONCEPTs are near to the root of hierarchy, and versus visa.
5.1 LSS System
The system is depicted in Fig. 5 that demonstrates the search results from
PubMed for a search query “pain”. A hierarchy of clusters is built on the
return results. The abstract (unstructure) and MeSH terms (semi-structure)
are in use to cluster them. The same term “pain” has been taken to retrieve
information from GOOGLE. The returned result snippets are grouped into
several clusters as shown in Fig. 6.
5.2 Results
The experimental evaluation of document clustering approaches usually mea-
sures their
effectiveness
rather than their
e
ciency
[22], in the other word,
the ability of an approach to make a
right
categorization. Entropy is involved
Fig. 5.
LSS System oversearch engine PubMed
