Connecting Web Resources with Deep Hyperlinking

In the field of Humanities Computing, the idea that related resources should be interlinked not merely via the relatively superficial mechanism of the hyperlink, but more fundamentally intertwined in a more structured and meaningful way, has been espoused for many years. Indeed, one of the most rewarding aspects of work in this field is that - as projects progress - the more academics begin to understand more about humanities computing
and the implications it has for their research and
interests, the more imaginatively they begin to perceive
connections and relationships within their source materials
which were never before apparent. This process of
scholarly enquiry and discovery often means that the practice of creating digital humanities resources often has to accommodate considerable change as project
outcomes are remodelled and refined better to reflect the true nature of the source materials as it emerges. Of course
it is not only within the context of a single project that connections arise: most projects begin with knowledge of related resources and the potential for partnerships and collaboration.
There have traditionally been two approaches to
integrating separate digital resources: firstly, by
supplying an overarching portal fabric to tie the resources
together and present the data in a unified manner; or
secondly to insist on the separation of the resources and allow superficial cross linking between them. The most successful portal environments have traditionally been found in the context of the digital library, where they are almost always focussed on information retrieval by search: OCLC’s FirstSearch service, for example, provides
a unified environment for querying numerous digital
resources although it is not well suited to a more
browsing-oriented model of usage.
Between humanities computing projects, the desire for formal relationships with related resources - expressed through data connections – certainly exists, although as yet often remains unexpressed: over and above simple hyperlinks, close integration with other projects is often simply not feasible. There are several reasons for this. Firstly the Web Services approach, as it has largely so far been seen, necessitates a rather different view of a finished project to the traditional, digital library model
preferred by many funding bodies who (albeit with
notable exceptions) remain attached to the idea of a
‘deliverable’, a unique resource with a distinct identity and presence. Secondly, even when funding does allow for more flexibility, the lingering metaphor that digital
resources are somehow like giant books whose intellectual property must be protected often precludes projects from even considering the possibility of low level connections with other resources and the considerable scope for rich and rewarding collaboration this would afford.
One possible explanation for this situation is that whilst
there are established technical standards for data
interchange that fit perfectly the model of distributed
knowledge – viz. Web Services – there are no generic standards or agreements in place to deal with the broader
implications of connecting academic resources to one another, across institutional and national boundaries. In the UK, whilst there are certainly funded initiatives
looking into the establishment of such a standard, we are not close to an answer, and what work is being done is focused more on the needs of the scientific community
than of the humanities. Until a standard is agreed upon, we will be left in a situation where, for the most part, projects are developed in relative isolation and we must therefore continue to try and integrate resources as best we can, after the event. Moreover the thousands of rich web resources which were developed before the advent of technologies such as WDSL cannot be ignored simply because their technical implementation is outdated.
One solution to this is to start thinking about how
projects can be more seamlessly integrated at a higher level – within the presentation layer. In other words: if hyperlinking is the only viable option for linking one resource to another, how can it be improved upon as a concept? The obvious answer seems at first glance to be a simple one: using hyperlinks to target low level data, in context, within other sites, i.e. creating connections from
points deep within one site to deep within another.
Although implementation may require some sort of low level access to the target site – such as analysis of the core database or XML repository so that authority lists and links can be constructed – the cost implications would nevertheless be trivial. Whilst linking resources in this way does nothing actually to integrate them, each
individual project remains free to maintain an independent
framework of secondary and supporting material, and in fact it may be beneficial to a user to see related data in a number of different contexts - as long as the transition
from one context to another is sympathetic enough to
allow the data to remain intelligible in each environment.
This last clause reveals why realising a ‘deep
hyperlinking’ solution is far from simple: allowing users to jump between resources with separate visual identities, interfaces and structures presents potentially massive problems in terms of usability and, more fundamentally,
HCI. If one humanities web resource is to rely, heavily,
on the content of another, yet the two must retain
individual branding, how will the user perceive the
relationship between them? How should this interconnection
be presented? The biggest difficulty lies in the fact that, allowing for the fact that users may be sent between a number of very obviously different digital resources, at all times the visual environment must keep the user aware of where he is (not only to the extent of indicating where
he is within a resource, but also which resource he is
actually viewing), how he got there, and what his options are (i.e. explore the current site; return to the previous
site; explore a related resource on a third site, etc). There are equally tricky subsidiary issues to consider: how should
URLs be managed? Should sites share a common
vocabulary, extending the scope of ontologies to include
consistent labels for special pages, functions and
navigational concepts across connected websites? Could digital library technologies such as OpenURL be brought to bear on the problem?
Of course, from any user’s point of view, what is important
is not so much how interconnection between two or
digital resources is achieved as what the experience of following connections will be. Put simply, any user must to be able to follow cross references and view related resources as simply and smoothly as possible. What is needed here is some way of creating a ‘seamless seam’: making it obvious to the user that they have switched
into another resource, whilst adhering to the HCI
concepts of context and orientation.
One possible, low cost approach to this problem is
to think about setting up, for each project, a special
‘reception’ or ‘landing pad’ area specifically to cater for people visiting the resource from another site, built upon middleware able to detect and reassure an itinerant
visitor from another resource a) where they have come from, b) how they can get back, and c) what their
options are. This is not per se a new concept: commercial
websites have for some time used query strings in urls to post data between one another to allow a site to
offer customised content specific to users coming from a
certain context. The challenge from the perspective of
interconnected humanities computing resources is, however, that users may well want to return to the context from which they have come, and pick up exactly where they left off; equally of course they may decide that they wish to stay in the context of the ‘new’ resource in which they find themselves.
In this paper I shall explore these issues further, discussing how we at CCH have approached such problems across a number of related projects all oriented around the field of Anglo Saxon studies. Although we are now striving to interlink these projects to a considerable degree, this sort of functionality was never a significant part of the way they were originally conceived and so each project has been developed to have a ‘standalone’ existence of the type I have described here. Low-level interconnects were not possible for many of the reasons outlined above, but, because of the involvement of CCH as a key stakeholder in each project, we have been in a unique position to be able to tackle experimentally some of the difficulties with deep hyperlinks, taking users from a deep context within one site directly to a related deep context in another. I will describe the HCI strategy and technical system we are developing to allow for ‘context sensitivity’ in our newest projects, specifically to accommodate peripatetic users as they move between one resource and the next.