Bayesian Stratified Sampling to Assess Corpus Utility

by: Judit Hochberg, Clint Scovel, Timothy Thomas, Sam Hall

edited by: Eugene Charniak

(15-16 August 1998), pp. 1-8.

View FullText article

http://www.aclweb.org/anthology/W98-1101

Reviews [Write a review of this article]

There are no reviews of this article

Find related articles from these CiteULike users

bkhl, STP (group), kvorg

Find related articles with these CiteULike tags

bayesian, bayesian-sampling, corpora, linguistics, nlp, statistics

Abstract

This paper describes a method for asking statistical questions about a large text corpus. We exemplify the method by addressing the question, "What percentage of Federal Register documents are real documents, of possible interest to a text researcher or analyst?" We estimate an answer to this question by evaluating 200 documents selected from a corpus of 45,820 Federal Register documents. Stratified sampling is used to reduce the sampling uncertainty of the estimate from over 3100 documents to fewer than 1000. The stratification is based on observed characteristics of real documents, while the sampling procedure incorporates a Bayesian version of Neyman allocation. A possible application of the method is to establish baseline statistics to estimate recall rates for information retrieval systems.

@inproceedings{Hall98, abstract = {This paper describes a method for asking statistical questions about a large text corpus. We exemplify the method by addressing the question, "What percentage of Federal Register documents are real documents, of possible interest to a text researcher or analyst?" We estimate an answer to this question by evaluating 200 documents selected from a corpus of 45,820 Federal Register documents. Stratified sampling is used to reduce the sampling uncertainty of the estimate from over 3100 documents to fewer than 1000. The stratification is based on observed characteristics of real documents, while the sampling procedure incorporates a Bayesian version of Neyman allocation. A possible application of the method is to establish baseline statistics to estimate recall rates for information retrieval systems.}, author = {Hochberg, Judit and Scovel, Clint and Thomas, Timothy and Hall, Sam }, citeulike-article-id = {161657}, editor = {Charniak, Eugene }, keywords = {linguistics, statistics}, month = {August}, organization = {The Association of Computational Linguistics}, pages = {1--8}, posted-at = {2005-04-15 10:46:41}, priority = {0}, title = {Bayesian Stratified Sampling to Assess Corpus Utility}, url = {http://www.aclweb.org/anthology/W98-1101}, year = {1998} }

RIS record

TY - CONF ID - Hall98 L3 - citeulike-article-id:161657 TI - Bayesian Stratified Sampling to Assess Corpus Utility SP - 1 EP - 8 OR - The Association of Computational Linguistics N2 - This paper describes a method for asking statistical questions about a large text corpus. We exemplify the method by addressing the question, "What percentage of Federal Register documents are real documents, of possible interest to a text researcher or analyst?" We estimate an answer to this question by evaluating 200 documents selected from a corpus of 45,820 Federal Register documents. Stratified sampling is used to reduce the sampling uncertainty of the estimate from over 3100 documents to fewer than 1000. The stratification is based on observed characteristics of real documents, while the sampling procedure incorporates a Bayesian version of Neyman allocation. A possible application of the method is to establish baseline statistics to estimate recall rates for information retrieval systems. KW - linguistics KW - statistics AU - Hochberg, Judit AU - Scovel, Clint AU - Thomas, Timothy AU - Hall, Sam A2 - Charniak, Eugene PY - 1998/08/15-16/ UR - http://www.aclweb.org/anthology/W98-1101 ER -

RIS BibTeX