Result: Oracle at TREC 10: Filtering and Question-Answering

Oracle’s objective in TREC-10 was to study the behavior of Oracle information retrieval in previously unexplored application areas. The software used was Oracle9i Text[1], Oracle’s full-text retrieval engine integrated with the Oracle relational database management system, and the Oracle PL/SQL procedural programming language. Runs were submitted in filtering and Q/A tracks. For the filtering track we submitted three runs, in adaptive filtering, batch filtering and routing. By comparing the TREC results, we found that the concepts (themes) extracted by Oracle Text can be used to aggregate document information content to simplify statistical processing. Oracle's Q/A system integrated information retrieval (IR) and information extraction (IE). The Q/A system relied on a combination of document and sentence ranking in IR, named entity tagging in IE and shallow parsing based classification of questions into pre-defined categories. 1. Filtering based on Theme Signature As a first time filtering track participant, Oracle submitted runs for adaptive filtering, batch filtering and routing this year. Only linear-utility optimized runs were submitted for adaptive filtering and batch filtering. The filtering system is built based on the Oracle 9i database with PL/SQL- an Oracle supported database access language. Since the routing sub-task outputs the top 1000 ranked documents per category, and the training process and similarity score calculation algorithm are the same for batch filtering and routing, we will focus our discussion on batch filtering and adaptive filtering. The filtering system can be divided into three parts based on functionality