The simple notion is that, on the Reflex page, you upload a document (or a case name or single citation) from your machine and Reflex, recognizing case names, citations and legislation data, will edit that document by supplying citations (where necessary) and hyperlinks to the appropriate text. You can save the final result as an HTML document (which, of course, you can then convert to other formats as needed). Reflex accepts material in the following formats: .DOC, .DOCX, .WPD, .RTF, .TXT, .HTML.
I gave it a quick test by pulling a patch of prose rife with citations out of a judgment from CanLII at random ((3 paragraphs from Milsom v. Corporate Computers Inc., 2003 ABQB 296 (CanLII), as it happens.)) and submitting it to Reflex after I had deleted the hyperlinks, but left the citations, in paragraph 1 and deleted both the hyperlinks and citations from paragraph 2. The third paragraph contained no hyperlinks in the original so I left it as it was.
Reflex performed well. It supplied missing the hyperlinks in paragraph 1, though it did so by duplicating the citations. I think what happens there is that Reflex reads the case name and the citation as separate instances each requiring the insertion of a hyperlink to the single case; though, I haven’t by any means put Reflex through its paces and don’t pretend to know its limits yet in any sense. In paragraph 2, where, remember, I had deleted the hyperlinks and the citations, Reflex put the hyperlinked citations back in perfectly. Finally, Reflex understood that there were no citations in CanLII for any of the material in paragraph 3, and so left it just as it was.
If you’re interested in seeing the results of my test, you can download the three files: patch1.rtf (the original piece from CanLII); patch2.rtf (the altered piece submitted to Reflex); and patch3.html (the file Reflex gave me back).
There’s more to Reflex than I am able to tell you about now. Apparently you will be able to integrate it with your knowledge management system, for example. And I want to look more closely at the sort of HTML it writes and whether that depends in turn on the format you feed it to begin with.