The following new six page issue brief linked here and embedded below is published by the Association for Research Libraries (ARL) and written by Krista Cox, director of public policy initiatives at ARL.
From the Background Section of the Document:
No researcher can read all relevant research articles that are published in her field of interest. Even if she could, she would not be able to detect patterns in the research results that emerge only from large-scale computational analysis, known as text and data mining (TDM). Researchers who want to perform TDM on copyrighted research articles might seek clarity about whether they need permission from journal publishers or whether copyright’s fair use doctrine permits TDM on accessible articles. In almost all cases, performing TDM on accessible articles is a fair use. As long as the researcher is not bound by a contract that forfeits her fair use rights, she may proceed with TDM so long as her results do not make the full text, or substantial portions, of the underlying
articles publicly available.
Numerous courts in the United States have upheld the reproduction necessary to perform TDM as fair use, even though the content being copied into the database is copyrighted.
Fair use is a flexible limitation and exception that allows copyright law to adapt to changing circumstance s and new technologies and helps ensure a balanced copyright system. Thus, while the United States does not have a specific limitation or exception to explicitly allow TDM, fair use has accommodated the creation and growth of TDM as a new research tool.