SP500215 Appendix B: System Features Appendix National Institute of Standards and Technology D. K. Harman `IA. QUERY CONSTRUCIION -- AUTO[OCRerr]CALLY BUILT QUERIES (ADHOC) `EM NAME [OCRerr] CITY [OCRerr] ERIM [OCRerr] TMC [OCRerr] CUHU J UCLA ] SEC [ ETh Concept strings; jelds Title and Concepts nationality strings All All Title and concepts ______ All 15 min. for initial Dmputer Time to Build queries; 4 hours 10-20 ms .`) < 1 second each - 2 seconds for optimization Negligable < 1 second each querv [OCRerr] Used to Construct None Weighting on topic terms) No Yes, n[OCRerr] Yes (adjacent extraction words) __________ ___________________ _______ No Yes, simple tag ic Parsing _________________ identification No ;ense Disambiguation No Noun Identification No [OCRerr]r No Patterns ic Associations No ion of Queries [OCRerr]revious Constructed [OCRerr]ructure No Structure [OCRerr]ic addition of Boolean tors/Proximity Qperators No Robertson/Sparck- IDF weights Robertson[OCRerr]parck- Jones F4 predictive based on Jones F4 predictive weights (approL collection weights (approL IDF IDF in the absence frequencies in in the absence of of relevance training docs for relevance No ___________________ information). routing db. information).