SpiderLink: a Keyword Search Algorithm

Studenteropgave: Kandidatspeciale og HD afgangsprojekt

  • Linas Butenas
The increasing need for a keyword-based search systems on relational data-bases motivated us to develop the SpiderLink search engine. It uses a k-tree data structure to find the connections between given keywords. As a result SpiderLink returns the sequences of tuples relating the tuples where keywords were found. In paper we first define a k-tree data structure. It has several important properties: i) it works on hierarchical, parallel and single relationships in a database; ii) it is minimal; iii) it is finite; iv) it can be implemented as a hash-table allowing to use it most efficiently. Later we present the SpiderLink search algorithm and diagram in the example how it actually works. We have implemented SpiderLink and it is a fully developed keyword search engine. The tests done on the databases support all our theoretical assumptions.
Udgivelsesdatojun. 2003
ID: 61058203