1 / 10

Precision and Recall in Information Retrieval: Understanding Their Importance

Learn about precision and recall metrics in information retrieval, how they are calculated, and their significance for evaluating search results. Discover the relationship between precision and recall and how they affect the quality of retrieved documents. Explore interpolation, extrapolation, and precision/recall curves. Understand F-measure and normalized recall to assess the accuracy of retrieval rankings.

kasimir-wyatt
Download Presentation

Precision and Recall in Information Retrieval: Understanding Their Importance

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Precision and Recall Reminder: Precision : % of # of retrieved documents that are relevant Recall : % of all relevant documents that are retrieved

  2. Not Relevant Relevant Not Rel Not Ret Rel but Not Ret A Ret but Not Rel C Rel and Ret B Not Retrieved Relevant Not Relevant Retrieved

  3. Computing Precision and Recall Theoretically : a continuous relationship (Precision value for every level for recall) 1.0 Precision Recall 1.0 In practice : Can only estimate at a resolution  to the # of relevant docs in the collection 1.0 Precision Recall 1.0

  4. Interpolation of Precision/Recall .9 P1 Precision DP P2 .7 .20 (DR) R1 R2 .125 .20 .25 Recall Interpolation

  5. Extrapolation of Precision/Recall ? ( for Recall less than ) 1.0 ? .9 P1 Precision P2 .7 R1 R2 .125 .20 .25 Recall

  6. Precision/Recall Curves 1.0 .9 .8 .7 Precision .6 .5 .4 .3 .2 .1 0 1/8 2/8 3/8 4/8 5/8 6/8 7/8 8/8 .125 .25 .375 .5 .625 .75 .875 1.0 Recall

  7. Precision/Recall Curves (1/1) (2/2) (3/3) 1.0 .9 .8 .7 .667(4/6) Precision .6 .5 .4 .312(5/16) .3 .2 .111(6/54) .030(7/230) .1 .003(8/2664) 0 1/8 2/8 3/8 4/8 5/8 6/8 7/8 8/8 .125 .25 .375 .5 .625 .75 .875 1.0 Recall

  8. F-measure Harmonic mean between Precision and Recall for R = Recall P = Precision for any relevance-scored retrieved set and for Recall and Precision at j-th document in an ordered ranking

  9. E-measure • Van Rijsbergen(1979) • relative importance of Precision/Recall weighted • by user given parameter  if  = 1, Ej is complement to harmonic mean if  > 1, Precision more important than Recall if  < 1, Recall more important than Precision

  10. Normalized Recall How closely do the ranks of the retrieved documents (e.g. 1,2,4,6,13) match the ideal ranking of true retrieved documents(1,2,3,4,5)? Ideal ranks(1,2,3,4,5) 1 4/5 Recall 3/5 Actual ranks(1,2,4,6,13) 2/5 Worst ranks (196,197,198,199,200) 1/5 0 1 5 10 15 195 Ranks of retrieved documents

More Related