• Keine Ergebnisse gefunden

Task 1: Precision und Recall

N/A
N/A
Protected

Academic year: 2022

Aktie "Task 1: Precision und Recall"

Copied!
1
0
0

Wird geladen.... (Jetzt Volltext ansehen)

Volltext

(1)

Multimedia Retrieval –HS 2020

Task 1: Precision und Recall

Two search engines A and B perform a search on the same collection. Each engine returns the top 30 documents for a single query in ranked order by their relevance. The following table provides the ranking and denotes with a ‘+’ if the document is relevant and with an empty cell if it is not relevant. The collection contains a total of 12 relevant documents for this single query.

Exercise-1-1

a) Draw the precision-recall graph for both engines.

b) Which engine performs better and why?

The answer depends on how important the precision and recall are for the user. If the user prefers precision, engine A is better than B as it contains many relevant documents among the top results. If the user favors recall, engine B is better than A as it returns more relevant documents and keeps precision higher with growing recall. Finally, considering the system efficiency B outperforms A with 0.42 to 0.37.

0.00 0.20 0.40 0.60 0.80 1.00

0.00 0.20 0.40 0.60 0.80 1.00

Recall

Precision

A B

Exercise 1: Performance Evaluation Solution

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

A

+ + + + + + +

B

+ + + + + + + + + +

Referenzen

ÄHNLICHE DOKUMENTE

OR grade 1 OR grade one OR 2nd-grade* OR second-grade* OR grade 2 OR grade two OR 3rd-grade* OR third-grade* OR grade 3 OR grade three OR 4th-grade* OR fourth-grade* OR grade 4

B Meta-analytic overview of prevalence of lower back WMSDs.. C Meta-analytic overview of prevalence of upper

• On the search results page, you will often see Factz for general topic queries; these Factz are collected from pages across Wikipedia. • On a particular topic page, you can see

Considering multiple instances of such biases identified in relation to matters of gender and race on search engines raises the question of the implications of algorithmic

Two search engines A and B perform a search on the same collection. Each engine returns the top 30 documents for a single query in ranked order by their relevance. The following

Our preliminary findings mix several techniques, such as caching, locality prediction and distributed query processing, that try to maximize the fraction of queries that can be

Our approach is unique in the sense that it combines interactive topic graph ex- traction and exploration on different mobile devices with recently developed tech- nology

Harvest