High speed text retrieval from large databases on a massively parallel processor

摘要

A word-based text document retrieval system is being implemented by AMT (Active Memory Technology) on the AMT DAP series of massively parallel computers. For a 20 G Byte database of 10 million documents a system with 2 DAP 610s can sustain around 70 Boolean queries per second. Any of these parameters can be extended. In the initial application the addition of two DAP 610s to a cluster of four large VAXes will increase the performance by several hundred times.