ISR11 Relevance Feedback in an Information Retrieval System chapter W. Riddle T. Horwitz R. Dietz Harvard University Gerard Salton Use, reproduction, or publication, in whole or in part, is permitted for any purpose of the United States Government. Results for Query No.9 OA9ANALYSIS Document Rank No. Name Relevant Correlation 1 39 51ThESEARCH .43301270 2 47 lll6RETRIEVA .42135049 3 11 lll7I[OCRerr]ANNlNG .40209035 4 62 809A MECHANI .39900375 [OCRerr] 12 l1l3THE MATE .34815531 6 2 814A [OCRerr]IEW EFF .34299717 7 4 3O7ROLE OF / .32478490 8 17 3OIDOCUMENT .32075015 9 9 206ANALYS1S .29329425 10 23 815A FACET A .29329423 11 27 50415 REL[OCRerr][OCRerr]VA .28867513 12 21 l209SCIENCE .28724249 13 1 1l04THE IBM .28647316 14 71 lO8SEARCHERS .28517591 15 70 601ENCODING .28306926 Ranking took .120000 seconds Relevant Document Ranks Document __ __T I No. Name Relevant Correlation 1 33 82 3O8MACHINE R X 1 .20701967 ifim 4l8ENGLISH-L x J__.12038585 J vI-49 Normalized Recall = .5061728 Normalized Precision = .1717605 Rank Recall = .0361 Log Precision = .0936 N [OCRerr]uery(l) = Ouery(0) + Alpha [OCRerr] v[OCRerr]IGH[OCRerr](I) * Document of Rank I 1=1 (1) Document Rank 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Weight -1-1 0 0 0 0 0 0 0 0 0 0 0 0 0 N = Number of retrieved documents Alpha = 1 (a) Initial Results Use of Negative Weights Cosine Correlation Function Increasing Alpha Strategy Figure 10