Date: Wed, 10 Jul 1996 18:50:39 -0500 (EST) From: John Fieber <jfieber@indiana.edu> To: James Raynard <fdocs@jraynard.demon.co.uk> Cc: roberto@keltia.freenix.fr, jhs@freebsd.org, hoek@freenet.hamilton.on.ca, doc@freebsd.org Subject: Re: share/doc/FAQ/obj/freebsd-faq.html Message-ID: <Pine.BSF.3.94.960708164808.24187A-100000@Fieber-John.campusview.indiana.edu> In-Reply-To: <199607081710.RAA01482@jraynard.demon.co.uk>
next in thread | previous in thread | raw e-mail | index | archive | help
[playing e-mail catchup...] On Mon, 8 Jul 1996, James Raynard wrote: > Sorry for the delay in replying to this. There will probably be more > suggestions later, but two that strike me immediately are:- > > 1. The TOC entries should be separated from the question and answer With appropriate tagging of the document, any of a variety of TOC structures can be generated. > Also, this would allow the same question to appear in different > forms in the TOC (eg ``How do I get a PS/2 mouse to work with > FreeBSD?'' and ``How do I get my laptop's trackerball to work with > FreeBSD?'' would both lead to the instructions on configuring a > kernel with PS/2 support). Now this is interesting. It basically amounts to tagging an FAQ answer with multiple questions. This amounts to a specialized form of cross reference. > 2. I don't know how feasible this is, but it would be really good to > have some way of identifying keywords for an indexing program. For > example, in the above example, ``laptop'', ``mouse'' and ``kernel'' The controlled vocabulary rears it ugly head! If we add keywords then we have to establish and maintain a controlled vocabulary and ensure that it gets consistently applied. If we don't do that, a keyword field is meaningless and can actually be worse than no keyword field if the user and/or the searching software assumes it is a controlled vocabulary. Basically, for something on the scale of the FAQ a combination of free-text searching and rich cross references between related topics would be more effective. Where hypertext typically fails is in helping the user choose a place to dive in. Here free-text searching is a help but a map of some sort is essential. This is where Sean Kelly's nested FAQ structure comes into play. At first, this scheme strikes me as being inverted. By that I mean that the topics contain the questions, rather than the questions containing the topics. This bugs me because it favors the construction of a mono-hierarchy, which decades of classification research have discredited as an effective scheme. Additionally, it could result in a deep hierarchy which is also regarded (more recently) as a Bad Thing for usability. Subject classification schemes, such as LC, have traditionally been fairly deep largely do the phisical structure of the card catalog. With the advent of electronic catalogs, there has been a great flattening, even to the point where catalog software actually takes apart hierarchical subject headings to make them searchable by any component in the hierarchy. Two places where hierarchy still has a role are in filing physical materials, which can only exist in a single place in a one dimensional filing system, and in browsing. For a printed version of the FAQ, both issues concern us. For online, only the latter. So, we still have the task of coming up with some hierarchy, although I don't think it should be more than two levels. I suspect this could be easily constructed with a simple sectioning element. -john == jfieber@indiana.edu =========================================== == http://fallout.campusview.indiana.edu/~jfieber ================
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSF.3.94.960708164808.24187A-100000>