Keyword Competition Approach In Ranked Document Retrieval
due to the availability of huge storage spaces, multiple storage devices and different storage media. The rapid growth of data in the database will eventually render the data unmanageable and cause problems in retrieval, where the users are unable to retrieve the right document. This is one of...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2010
|
| Subjects: | |
| Online Access: | http://eprints.usm.my/43051/1/Poltak_Sihombing24.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | due to the availability of huge storage spaces, multiple storage devices and different
storage media. The rapid growth of data in the database will eventually render the data
unmanageable and cause problems in retrieval, where the users are unable to retrieve the
right document. This is one of the most important problems in IRS. The use of keywords
is one of the methods in IRS which can solve this problem. In this thesis, we propose a
methodology in GA (Genetic Algorithm) which is known as Keyword Competition (KC)
approach. KC is a competition scheme in finding the best keyword, known as ‘keyword
solution’ (KS), among the available keywords. The keyword solution is then matched to
the document collection in the database in order to retrieve the most relevant document.
In this research, the collection of proceedings of BADAN TENAGA ATOM NASIONAL
(BATAN) Indonesia, presented by University of Indonesia (UI), Jakarta was used as a
standard dataset. We also propose a keyword based ranking scheme aimed to better rank
the retrieved document in the spirit of presenting the most relevant document to the users.
Keyword based ranking scheme consists of two (2) main phases; namely keyword
solution matching and similarity percentage formulation. In the keyword matching
process, the system will match those KS by finding the same words in the title, abstract
& keyword of each document collection in the database. The similarity percentage
formulation is used to rank the retrieved document based on the similarity value. The
scheme was tested with two different fitness formulations, i.e. Jaccard’s function and
Cosine’s function. We then compare the result of KC to the similarity level in Hopfield
method. A prototype called Journal Browser System (JBS) based on this scheme was
developed. The results collected from JBS provide the evidence that KC approach and
keyword based ranking scheme give better performance compared to Hopfield method. |
|---|