Whose text, whose mining, and to whose benefit?

Abstract

Scholarly content has become more difficult to find as information retrieval has devolved from bespoke systems that exploit disciplinary ontologies to keyword search on generic search engines. In parallel, more scholarly content is available through open access mechanisms. These trends have failed to converge in ways that would facilitate text data mining, both for information retrieval and as a research method for the quantitative social sciences. Scholarly content has become open to read without becoming open to mine, due both to constraints by publishers and to lack of attention in scholarly communication. The quantity of available text has grown faster than has the quality. Academic dossier systems are among the means to acquire more quality data for mining. Universities, publishers, and private enterprise may be able to mine these data for strategic purposes, however. On the positive front, changes in copyright may allow more data mining. Privacy, intellectual freedom, and access to knowledge are at stake. The next frontier of activism in open access scholarship is control over content for mining as a means to democratize knowledge.