Via links in the Tables of Contents, every chapter of our thousands of online books can be skimmed with Active Skim, and can be explored further by clicking on key terms from the chapter.

The basic skim presents the most significant single chunk of text from every page in the chapter, as identified by our programs, enabling a reader/researcher to get the gist of a chapter, and to focus in on particular pages of interest.

We also provide the 30 most significant terms derived from that chapter in the right-hand column. Click on a term, and you will see, in context, the chunks from any page from that chapter containing the term. From there, you can open the exact page for more further, contextualized, reading. Again, though imperfect, our textual analysis approach does a reasonably good job at identifying key ideas and themes from a chapter.

We believe the Active Skim tool has the potential to significantly assist researchers, students, specialists, and others more rapidly make use of our online resources. We'd be delighted to hear from you at searchfeedback@nas.edu, with your comments and suggestions about how to improve this interface even further.

Note: some of our publications only have rough OCRed text -- machine-read from page images -- which may lead to unpredictable typos and errors. Also, we have found that an early version of Internet Explorer for the Mac (no longer supported by Microsoft) responds poorly to the Javascript upon which our active links depend. All other Mac browsers (Safari, Konqueror, Firefox) have no trouble.

Backend Info:
  • Top terms for each chapter are identified and evaluated for significance within the chapter algorithmically, from ASCII text
  • Skim chunks algorithmically extracted on the fly from ASCII text, based on preponderance of high-value terms
  • Metadata contained in book-specific XML files
Frontend Info:
  • Perl (on a Linux box) prepares and presents javascript-enhanced HTML pages via CGI.
  • Each Active Skim page is generally between 50 and 100k. Once loaded all processing happens within the client's browser via javascript, rather than on a server.
Most of the Active Skim architecture and approach was invented by Michael Jensen based on lexical analysis systems currently under patent review. Visual look designed by Sara Sandhu; usability design by Megan Ellinger.
Visit The Academies