Talmudic hypertext browser/formatter

Updates

Introduction to the concept

Often, one can get ideas for new systems from a careful examination of older or ideas drawn from different but related media. This final project idea is a very rich one, and is an idea that has not been explored before, to my knowledge.

Briefly and (probably a little) inaccurately, the Talmud is a series of Jewish theological works and commentaries on the Torah and religious practice, developed over hundreds of years of rabbinical tradition. If you examine a page of a printed edition of the Talmud, or look at an old manuscript, you will notice an interesting text layout.

Each page is divided into several areas (sometimes in different languages, sometimes translated). Each area of the page is devoted to a particular text, and since the texts are commentaries, each page tries to contain all of the relevant comments in a single page.

From a hypertext point of view we can analyze this book as a set of interlinked texts: annotation links connect commentaries to the "main text" (or to other commentaries, as there are commentaries on commentaries in the Talmud). As much of the time as the typesetter can manage, however, these links are represented implicitly by placing both ends of the linked text on the same page. Since each document, for the most part, has links to only a few of the other documents, this presentation as parallel texts works well at gathering the information that a reader wants into a single place; it also reduces the number of explicit cross-references needed, since references to the same page can be omitted or greatly simplified.

The project
:

The project idea is to create a system that would use a layout mechanism like that used in Talmud to browse hypertext documents. Given a specification of page areas, and their possible size ranges, the system would attempt to create single-page display of several linked documents by filling the page areas so that the page is full, and there are as few explicit links needed as possible.

Unlike a Talmud, this display could be dynamic, so that a page could be generated starting from any point in a single document. Decisions need to be made about how scrolling and the following of navigational links might behave.

The layout problem of filling n variably sized text areas so that no inter-document links point off-page is clearly insoluble, at least in general. Optimizing page like this may well be NP-complete (the problem seems harder than optimal table formatting, which has already been shown to be NP-complete.

However, we can extend the principle of the dynamic line-breaking technique to deal with this problem, by ordering the text frames in a sequence based on link dependencies. For each frame, we will assume that all previous frame breaks have been considered, thus reducing the number of cases to be considered.

Data and Testing for Talmudic formatter
:

Commented texts are pretty widely available, but may not be needed for a project like this, because the linked texts could be composites of other texts, retrieved either via explicit links or by a search mechanism.