Database Support for Path Query Functions

Extending relational database functionality to include data mining primitives is one step towards the greater goal of more closely integrated database and mining systems. This paper describes one such extension, where database technology is used to implement path queries over a graph view of relational data. Partial-path information is pre-computed and stored in a compressed binary format in an SQL data type. Path querying is implemented using SQL table functions, thus enabling the retrieved path tables to be manipulated within SQL queries in the same way as standard relational tables. The functions are evaluated with particular reference to response time, storage requirements and shortest-path optimality, using road system data representing relationships between over 2.8 million entities.

• Type: Conference
• Year: 2004
• Where: BNCOD
• Authors: Rachel Hamill, Nigel J. Martin
• Keywords: BNCOD 2004, BNCOD 2007, Data Mining Primitives, Relational Database Functionality, SQL Data Type