Throwing Shoes In the Machinery of the World
Thursday, June 21st, 2007Michael Gorman is still showing his ass to the world over at Britannica Blog. This time he demonstrates that he hasn’t got a clue how Google works:
Information retrieval systems have been studied for many decades. In the course of that study two important criteria have been developed to evaluate such systems—those criteria are recall and relevance. The first measures the percentage of pertinent documents retrieved from a database (for example, if there are 100 documents on Zambian agriculture in a database and a search on that topic retrieves 76 of them, the recall is 76%). The second measures the supposed appropriateness of the documents that have been retrieved (for example, if you retrieve 100 documents when searching for Zambian agriculture and 76 of them are actually about Zambian agriculture, the relevance is 76%).
Information retrieval systems achieve high recall and relevance rates by the use of controlled vocabularies (indexing terms, etc.) and present the results of complex searches in a meaningful and usable order. By any of these criteria, Google and its like are miserable failures. A search on those engines on anything but the most minutely detailed topic will yield many thousands of “results” in no useful order and with wretched recall and relevance ratios. However, even when the documents retrieved by a search engine are on the subject sought, the quality of the material - often community-generated material that pops up high on a hit list because the material is free and easily accessible — is shoddy or irresponsible.
Let’s unpack some of the misconceptions that Gorman is, once again spreading heedlessly.
