Users and developers interested in the semantic web attended a community event at HP Labs Bristol, UK on Tuesday 11 November 2008. This was an opportunity for users and developers to meet and to share experiences with semantic web applications. The objective was to allow people to share experiences of using the semantic web. The following talks were given:
Graham Klyne, Oxford University
The FlyWeb Project is building a web of genomic data on the fruit fly (Drosophila), a key model organism for biomedical research. We're also developing biomedical research tools that exploit this Web of data. We're trying to follow an agile approach, where user stories captured from scientists drive design of data, services and applications in the Web. So far we've deployed SPARQL endpoints for a subset of data from three separate genomic databases. We've also developed a small Javascript widget library for interacting with these data sources, and composed these widgets into a simple Web applicaton which allows a researcher to compare spatial patterns of gene expression in different organs.
Our experience overall has been positive, and we've found that the SPARQL/AJAX pattern fits well with an agile approach. Ideas we'd like to share and discuss include strategies for ensuring robustness and availability of SPARQL endpoints; using SPARQL to probe & test the content of a triplestore; incremental strategies for data integration via RDF/SPARQL; and design patterns for RDF and SPARQL in Web apps.
Laurian Gridinoc, Knowledge Media Institute, The Open University
The primary goal of PowerMagpie is to bring semantic interpretation to web browsing, without involving important effort from the user. This is done by bringing and relating knowledge from the semantic web to a regular web page. The first version demonstrates how this can be achieved in almost real-time. However, there are important shortcomings in presenting this semantic information to a regular user. Hence, we will outline the approach towards the next version which will use a different knowledge representation layer (inspired from Gärdenfors' Conceptual Spaces). We will demo the first version as well as the prototype of the second version which would illustrate our new approach.
Hugh Williams, OpenLink Software
This talk covered the alleviation of challenges associated with conceptual level data access, across heterogeneous data sources; using Virtuoso's RDF Views, Data Virtualization, Native RDF storage, and Linked Data deployment technology.
Leigh Dodds, Talis
As we move past the boot-strapping phase of the linked data cloud and begin to explore the potential of this huge and growing data set, we need to begin discussing what other features we need to support sustainable and commercial usage of the semantic web. What services can be offered over linked data? Do we need to consider quality of service, stability, longevity, or does the sheer scale of the web make these moot points?
This talk will explore these issues and look at how the Talis Platform is hoping to support the flow of data around the semantic web.
Jeni Tennison
Microformats and RDFa encode semantics within web pages, but methods for accessing those semantics generally operate on static documents, with no strong relationship between an element and the RDF triples it encodes. Browser-based applications need Javascript libraries that provide easy access to and manipulation of these semantics and that reflect the current state of the changing document. jQuery is a Javascript library that is both popular and extensible. In this talk, I'll discuss some jQuery plugins for accessing and querying the semantics within an HTML page, and how they can be used to enhance semantic browser-based applications.
Damian Steer, ILRT
This talk introduces “Caboto”, a RESTful RDF-based annotation system designed to meet the requirements of three projects involving the Universities of Bristol and Manchester. Annotations might be comments and criticisms about a presentation, linking to a relevant paper or resource such as a photo or rating an event.
The talk will explore issues around implementing a RESTful, SPARQL-backed service using Jena SDB, ARQ, Spring and JSR 311 (Jersey). The novel part of the system lies in the privacy layer, which uses named graphs as the basic unit of access. The talk will discuss issues surrounding authentication (OpenID) and authorization since the system matches restful URL requests to RDF graphs and applies security to them so there is the concept of public and private annotations.
Chris Dollin, Hewlett-Packard Laboratories
The Jena 2 rules language has proved useful both for the implementation of RDFS/OWL reasoning and for more general computation, but its implementation doesn't handle large datasets well and it has no convenient features for aggregation. The Jena 3 rules language builds on the concepts of Jena 2 language and the syntax of SPARQL, adds support for aggregation, and will have a scalable implementation. This talk sketches the language and proposed implementation tactics.
Eamonn Neylon, Oxon SWIG