In our research group, there is a need to capture, organize and share resources associated with a domain of exploration. We are building a tool for this task, based on previous experience in the knowledge management domain. In this position paper, we present our thoughts on what works (and what doesn't work), together with details of our initial implementation.
The snippet manager idea is not a new one [
Knowledge management is defined widely, for example achieving a global sharing of
knowledge within a company [
Secondly, the users are technically literate. This means that they are likely to quickly get to grips with a new tool, and may be motivated to make some small changes to behaviour for sufficient added value (a good example of this would be the use of ‘graffiti’ writing on PDAs). However, getting the balance right is not always easy; we note that in our (internal) semantic wiki, people hardly ever use the supplied wiki syntax to add RDF metadata. This may be due to the lack of instant, direct reward for adding metadata, a point which we are trying to address in our work.
Finally, the domain of interest is tightly focused, even more so than that envisaged
by Cayzer [
What is it, then, that the snippet manager is expected to achieve? We are in the early stages of this project, but we have engaged the user community and gathered some initial use cases in order to inform our guiding principles, a few of which are described here. We reiterate that these principles are relevant for our domain of interest – small group, tightly focused, technically literate researchers. We don’t expect the principles to necessarily generalize to the whole of knowledge management, or to the web at large. However, we believe that our problem domain is sufficiently common for these principles to be of value for the semantic web community.
“I need a way to collect evidence (web pages, PDFs, emails, forum posts), and to categorise parts of these so they can be linked together for post-hoc search.” “I need a way of annotating resources with evidence - 'why have you written this'” It should be ludicrously simple to collect snippets, using familiar methods such as bookmarklets or email. Snippet Manager should also pull in snippets from other sources (eg intranet databases) and handle provenance.
I need to create a category or classification for a new area of interest. Now, I need to add new companies, products, documents or links and tag them with this classification. I want to create relationships between these instances - e.g. competitor links. Our users will certainly want to change the ontologies on the fly. Although this sounds like a tall order, in our case both the structure of the ontologies (effectively taxonomies) and the nature of the changes (adding/removing/renaming a node) can simplify the implementation enormously. Of course the UI for such changes may not be trivial.
I want a regular alert showing the results of a web search for a topic. [OR I want to produce a report that shows all relevant products or technologies for a given topic] From a technical point of view, the ability to export (meta)data in a standard, machine readable way is a future-proofing mechanism, intended to prevent the portal becoming yet another information silo. From a user point of view, export in a human-readable form is equally important.
Our experience with the early snippet manager prototype [
There should be an instant reward for the user who adds metadata. The community aspect should be (from the user’s point of view) a beneficial side effect.
We have used the semantic portal [
We have previously built several semantic web applications whose primary function was to demonstrate a particular aspect of the technology. Our focus here is to build a tool. Therefore the simplest way of assessing its success is to measure its usage: 1. 2. 3. 4. 5. 6.
At what rate is new content added to the snippet manager? What proportion of the user group use the snippet manager as a day to day tool How well does the snippet manager integrate with other tools in use? How often is the snippet manager consulted for information or report generation? What is the satisfaction level of the users?
How quickly can new user requirements be integrated into the tool These measures are largely qualitative in nature. Yet they get to the heart of what of means to build a semantic web tool for personal, and group, productivity. We intend to assess our work using these criteria. 5
Simile’s Semantic Bank [
The broader idea of a semantically enabled website is explored in a number of
public portals, notably the Semantic Web Community Portal [
Many people use their weblog as a knowledge management tool and we think that
structuring the content of a post by adding some metadata could be useful for a group
of people. But the chronological view that weblog gives to the content not always it
the best solution to let users move through information. Our solution to this, which
we call semantic blogging [
Wikis are also interesting tools for collaboratively building knowledge, and there
are examples [
In this position paper, we have outlined our thoughts on what it would take to build a snippet manager for small group domain-focused knowledge sharing. We have shared some design principles which we hope will prove generally useful. We have also explained how we are going about building a system using these ideas. We have high hopes that our user-centred approach will function less as an interesting demo and more as a genuinely useful tool.