IRS13 Scientific Report No. IRS-13 Information Storage and Retrieval Document Length chapter E. M. Keen Harvard University Gerard Salton Use, reproduction, or publication, in whole or in part, is permitted for any purpose of the United States Government. [OCRerr]BSTRACT(A) VERSUS TITLE(T) ABSTRACT(A) VERSUS TEXT(x) ABSTRACT(A) VERSUS IND LECTION DICTIONARY I II III I II flI I II R-2 R-5 R-8 NR NP (NR)(N?) R-2 R-5 R-8 NR NP (NR)(NP) R-2 R-5 R-8 NR NP (NR [OCRerr]RE23 Stem []A LA []A L]AL]A OA OA 3[OCRerr] Thesaurus []A LA LA DADA ([OCRerr] ([OCRerr] [OCRerr]uests -3 [OCRerr]4I)IT Stem ADADADADAOAOA xDxx XX X A 35 Thesaurus T T A LA LA OOA A X X XX XOX [OCRerr]uests -1 ___________________________ _____________________________ [OCRerr] Stem ATLALALAOOA I Li I I I [OCRerr]-l Thesa[OCRerr][OCRerr][OCRerr]s A* A* A* A* A* A* Ii+IIi+IDi 1+ I[OCRerr] I[OCRerr] [OCRerr]uests Stem(Ov'er- lap Logical) T T A A T A T Word Form [OCRerr]IELD (Co-ord. T LT [OCRerr] [OCRerr]l L[OCRerr]]Li I Cutoff) 42 Word Form [OCRerr]sts (Doc. A T A L][] I ___ ________ 7 Cutoff) I Precision at Recall O.2(-2), O.5(R-5), and O.8(R-8). II Normalized Recall(NR) and Normalized Precision(NP) LJ over D = over .05 better. [No Box] = up t bett .05 better. [No Box] = up t bett in Individual Requests based on merit assigned by Normed. Recall (NR) and Normed. Precision (NP. O = over 60% of requests better. (No Circle] = between 50% and 60% better Summary of Performance Results of Document Length Comparisons Figure 39