ISR10 The Query-Document Matching Function chapter Joseph John Rocchio Harvard University Gerard Salton Use, reproduction, or publication, in whole or in part, is permitted for any purpose of the United States Government. Doc. No. Corr. Doc. No. Corr. Doc. No. Corr. Doc. No. Corr. 52 1.00 52 .71 52 .71. 123 .71 311 .57 *[OCRerr]177 .70 264 .69 177 .70 77 .49 264 .69 171 .67 264 .70 53 .48 171 .67 123 .66 171 .68 264 .36 123 .66 77 .62 127 .67 171 .34 77 .62 174 .62 263 .66 174 .33 174 .62 127 .58 259 .63 123 .33 * 263 .61 259 .58 52 .63 259 .32 * 127 .58 53 .58 225 .62 47 .32 259 .58 47 .57 361 .60 309 .31 53 .58 225 .55 174 .60 236 .30Th * 361 .58 311 .54 403 .59 cutoff 47 .57 403 .49 77 .58 225 .55 336 .40 47 .57 340 .54 345 .38 (cutoff) 311 . .54 . a 53 .51 S [OCRerr] 336 .47 __________ b.. dl cutoff 1 345 .45 g * - previously ne 311 .44 clustered Initial clas- b) Correl. distri- c) Partition d) Correl. distri- sification subset. bution of vector class of class. bution of class. Correl. of doc.#52. formed from subset vector formed vector formed from of part (a). from subset of subset of part (c). part (a). Progression of Categories and Correlation Distributions Figure 4.7 t)1' I,