2003-08-29 AHA! meets Auld Linky- Integrating Designed and Free-form Hypertext Systems http://www.ht03.org/papers/pdfs/26.pdf "In this paper we present our efforts to integrate two adaptive hypermedia systems that take very different approaches. The Adaptive Hypermedia Architecture (AHA!) aims to establish a consistently organized, strictly designed form of hypertext while Auld Linky takes an open and potentially sculptural approach, producing more freeform, less deterministic hypertexts. We describe the difficulties in reconciling the two approaches. This leads us to draw a number of conclusions about the benefits and disadvantages of both and the concessions that are required to combine them successfully." two trends in hypertext research: * open hypermedia - separate link structure from content - combined at runtime - separate linbases - popular mechanism on the web - Auld Linky (southampton) * adaptive hypermedia - adapt hyperstructure/content to users' needs - user modelling - methods from ITS [intelligent teaching systems?], AI - AHA! (Eindhoven) this is an experiment to merge these two systems/paradigms. how does structural approach affect the process of adaptation? how can we characterise these systems, how to they relate to each other? AHA!: - generic framework to create hypertext applications - web-based run-time layer - comes with authoring tools (graph tools, etc) - an author creates "page templates" and "concepts" - templates are XHTML with logic that decides whether fragments are shown - concepts are named and related entities that have logic and are triggered by page accessed. these are used to change the *user model* - eg, "Mammals" and "Reptiles" might trigger the concept of "Animals" as an experiment they've made a scripting language for AHA! to automatically compile page templates and concepts. Auld Linky: - OHS-lite in a single executable [?] - serves XML FOHM structures [Open Hypermedia Model] - query via patterns, uses HTTP - results filtered by context - when you query Auld Linky you do within a *context*, a set of key/value attributes that describe the query context. [this is like the position in information/conversation space my IM bots use for static implicature.] then when the hyperstructures are returned, it uses the context to only return the relevant slice types of dynamic hypertext (Mark Bernstein's terms): . caligraphic hypertext - traditional, links added between nodes . sculptural hypertext [this has been mentioned *many times* so far] - all nodes initially linked - links have conditions. based on user model, links are make invisible at a node when viewed - links have actions. when followed, they update user model [good definition!] joining AHA! and Auld Linky... Adaptive Model + Domain Model + User Model [interesting division] => concepts concepts feed into the conceptual engine and template engine (which comprise the adaptive engine) auld linky feeds into this adaptive engine to provide the sculptural hypertext. what do we gain? - sculptural links are looser and may offer serendipitous linking... - adding extra pages, unplanned links *appear* [in traditional (caligraphic) hypertext, everything is deliberate. see also, constructivist internet, paratactic aggregate. the ethics of cyberspace are determined by the deliberate nature of technology and - on top of that - hypertext. we need to have a more sculptural approach, or actually a city approach, where a wall isn't a deliberate *joining* statement, but a *separating* statement that CREATES POSITIVE DISTANCE WITHIN THE SPACE. by blocking out deliberately one thing, you relatively shorten the distance to "everything else". the everything else is undetermined until runtime! like that article about the american constitution that says it deliberately leaves a whole load of things out, emotional, unwritable stuff. very clever. mainly links create negative space - WORMHOLES - these are joining things, and the way of cyberspace at the moment. oooh, lots to think about here. this hypertext conference has mentioned many ways of how to make things further apart in a space. the web, dirk, all our technology isn't very good at that at all. so, some fundamentals: rooms, roads (space); walls, wormholes (distance); garden, knots (activity); levers, people. like voiced and unvoiced...] issue: - a tightly defined system doesn't appreciate anarchic, opportunistic behaviour - if you try to control the behaviour you lose serendipity - if you try to embrace the behaviour you loosen the design [interesting. this is the law of unintended consequences. leave opportunistic behaviour open. weblogs demand the opportunistic web, which goes against the old-school controlled-design paradigm -- or rather, against design itself.] [schraefel says this is a historic presentation in that there are two systems like this that can be constructively compared, and this is a first in the field of adaptive hypermedia.]