Return very few or no duplicates minimum false posives

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: 20 Sangmi Lee Pallickara € CS480 Principles of Data Management Spring 2013 CS480 Principles of Data Management Spring 2013 Optimizing Precision and Recall •  Ul.mate goal •  Op.mizing Precision –  Find only relevant document and all relevant documents –  Play safe! –  Return very few (or no) duplicates –  Minimum false posi.ves •  Precision •  Op.mizing Recall –  Correctness –  How accurate is the detec.on? –  Maximum recall: n2 candidate pairs •  Recall –  Completeness –  How complete is the detec.on? •  There are tradeoffs between op.mizing precision and recall. 21 Sangmi Lee Pallickara CS480 Principles of Data Management Spring 2013 CS480 Principles of Data Management Arithmetic Mean vs. Harmonic Mean •  F ­measure –  Harmonic mean of precision and recall: Harmonic Mean Arithme.c Mean ll ca Re ll ca Re Spring 2013 Finding Tradeoff between precision and recall Harmonic Mean Precision 22 Sangmi Lee Pallickara F − measure = 2 × recall × precision recall + precision Precision Arithme5c Mean € Sangmi Lee Pallickara 23 Sangmi Lee Pallickara 24 4 3/1/13 CS480 Principles of Data Management Spring 2013 Recall-Precision Diagram Sangmi Lee Pallickara CS480 Principles of Data Management Spring 2013 Recall-Precision-F-measure Diagram 25 Sangmi Lee Pallickara 26 5...
View Full Document

Ask a homework question - tutors are online