Virginia Tech. Department of Computer Science. Digital Library Research LaboratoryFrance, Robert K.2015-05-292015-05-291995France, Robert K. "When Stopping Rules Don't Stop." Internal Report, Virginia Tech, 1995.http://hdl.handle.net/10919/52844Performing ranked retrieval on large document collections can be slow. The method of stopping rules has been proposed to make it more efficient. Stopping rules, which terminate search when the highest ranked documents have been determined to some degree of likelihood, are attractive and have proven useful in clustering, but have not worked well in retrieval experiments. This paper presents a statistical analysis of why they have failed and where they can be expected to continue failing.23 pagesapplication/pdfen-USIn CopyrightStopping rulesDocument collectionsRanked retrievalTechnical reporthttp://www.dlib.vt.edu/reports/StoppingRules.pdf