Using statistical testing in the evaluation of retrieval experiments Export

@inproceedings{citeulike:4269852, abstract = {The standard strategies for evaluation based on precision and recall are examined and their relative advantages and disadvantages are discussed. In particular, it is suggested that relevance feedback be evaluated from the perspective of the user. A number of different statistical tests are described for determining if differences in performance between retrieval methods are significant. These tests have often been ignored in the past because most are based on an assumption of normality which is not strictly valid for the standard performance measures. However, one can test this assumption using simple diagnostic plots, and if it is a poor approximation, there are a number of non-parametric alternatives.}, address = {New York, NY, USA}, author = {Hull, David}, booktitle = {SIGIR '93: Proceedings of the 16th annual international ACM SIGIR conference on Research and development in information retrieval}, citeulike-article-id = {4269852}, citeulike-linkout-0 = {http://portal.acm.org/citation.cfm?id=160758}, citeulike-linkout-1 = {http://dx.doi.org/10.1145/160688.160758}, doi = {10.1145/160688.160758}, isbn = {0-89791-605-0}, keywords = {evaluation, representative, statistics, study}, location = {Pittsburgh, Pennsylvania, United States}, pages = {329--338}, posted-at = {2009-07-07 11:22:17}, priority = {2}, publisher = {ACM}, title = {Using statistical testing in the evaluation of retrieval experiments}, url = {http://dx.doi.org/10.1145/160688.160758}, year = {1993} }

TY - CONF ID - citeulike:4269852 L3 - citeulike-article-id:4269852 N2 - The standard strategies for evaluation based on precision and recall are examined and their relative advantages and disadvantages are discussed. In particular, it is suggested that relevance feedback be evaluated from the perspective of the user. A number of different statistical tests are described for determining if differences in performance between retrieval methods are significant. These tests have often been ignored in the past because most are based on an assumption of normality which is not strictly valid for the standard performance measures. However, one can test this assumption using simple diagnostic plots, and if it is a poor approximation, there are a number of non-parametric alternatives. CY - Pittsburgh, Pennsylvania, United States SN - 0-89791-605-0 AD - New York, NY, USA EP - 338 TI - Using statistical testing in the evaluation of retrieval experiments PB - ACM T2 - SIGIR '93: Proceedings of the 16th annual international ACM SIGIR conference on Research and development in information retrieval SP - 329 KW - evaluation KW - representative KW - statistics KW - study AU - Hull, David PY - 1993/// UR - http://dx.doi.org/10.1145/160688.160758 ER -