Supporting Constructive Navigation of Web Space
Romain Zeiliger, GATE-CNRS
E-mail : zeiliger@gate.cnrs.fr
Home page : http://www.gate.cnrs.fr/~zeiliger
Navigation is the main interaction mode between human and computer in hypermedia structured information space. As so, navigation is the main interaction mode in Web space and it is likely to become as well an important interaction mode on the desktop (through IE4 active desktop feature). From a more human centred point of view, navigation is just another computer mediated activity. Working within the framework brought on by the Activity Theory [Vygotsky, Leontiev, Kaptelinin, Nardi], we will draw on three basic principles to derive an approach to supporting both personal and social navigation in a constructive way : namely i) the hierarchical structure of activity (activity/action/operation), ii) the principle of internalisation/externalisation and iii) the principle of tool mediation.
At the operation level, standard browsers are quite easy to use and suffer only from a lack of (and some misleading) features which induce user problems such as : the difficulty of backward navigation, the hub and spoke problem, the "far from home" problem. At this level, the availability of a dynamic graphical representation (a map) of the user's visited subspace can efficiently ease up basic navigation difficulties.
Our approach will mainly address the problems at the action level : standard browsers offer almost no support to directed navigation. Current practices have not yet been accumulated in tools through design. Dealing with content categorisation and traversal anticipation, users have difficult decisions to make to navigate toward their goal. In this respect, due to its openness and extreme heterogeneousness in content, design, and control WWW (a non-moderated space) is more challenging than pre-web hypermedia. Situated as well as plan-based navigation is required.
There is much to say about supporting the intentional upper level of the navigation activity, but this issue will not be addressed in this paper.
With focus on the level of actions (as defined by activity theory) our research favours a constructive approach - both personalised and social - to navigation support : users are given means and are encouraged in the production and sharing of intermediary navigational objects and representations whose expected role is hypothetised as follows :
a) externalising self experience helps shaping of mental representations. Mental representations, in turn, shape the external ones thus allowing experience to be capitalised .
b) production of intermediary objects generates and structures navigational activity (providing an activity framework for the internalisation/externalisation process to take place).
c) production of navigational representations promotes a constructive learning of navigational abilities ( users become designers, readers become writers).
d) production and use of navigation representations promote plan-based navigation.
e) sharing of navigational representations with others (allowing for asynchronous collaborative navigation) helps establishing a common scheme of use, and ultimately a personal scheme of use (both necessary components in the process of "tool instrumenting" according to Rabardel).
f) sharing of navigational objects also allows for the social construction of information space and the social accumulation of knowledge.
It is proposed that - as a start - navigational objects and representations encompass collections of such things as annotations, maps, paths, concept-maps, and guides (with a loose definition). Objects are coined "intermediary" because they are not considered as mere navigation tools but rather as means for representing the user's self navigational experience. This approach inspires also from equivalent practices drawn from the study of design (the user is viewed as the designer of her own activity). Danger exists than this approach would ultimately bring more obstacles to navigation and no facilitation.
To explore this approach we have developed a prototype browser (called NESTOR) whose features implement the above hypothesis and can be demonstrated.
NESTOR has been used in several projects in 1997 : the french regional ORION project supported by the ARASSH programme of Rigion Rhtne-Alpes, and the European TELEDU project (ELECTRA in ALMA) part of the Telematics Application Programme for Education and Learning. Data have been already collected and analysed. More developments are underway, particularly in the course of the European LEARN-NET programme.