RDF::RDFWeb Perl Tools: Current Status

This document describes the status of various features under development for our Perl RDF applications. Specifically, the annotatation tools needed for the MedCERTAIN project, and the API features used in the Harmony ABC work.

See also: README.html

RDFWeb Perl: Annotation tools

These Perl tools are being developed for a variety of applications, in particular including ILRT's work on the Harmony and MedCERTAIN metadata projects. The MedCERTAIN work builds on prior ILRT research in the ROADS and DESIRE projects, and is based on the idea of a domain-neutral platform for networked quality ratings.

MedCERTAIN needs tools which facilitate... "Provision of simple mechanisms (and tools for raters to rate/annotate websites using bookmarklets, what's related menu, bookmark uploads and remote bookmark storag) and technology to aggregate these data"

We aim to provide tools that allow Web developers to Develop ratings query APIs, search/browse interfaces to RDF annotation stores, what's-related services, HTTP proxy add-ons, and PICS-based annotation services.

Recent work

During Nov 2000 - June 2001 we have worked on development and testing in the following areas:

• development of a Perl API to support retrieval, parsing, aggregation, querying of RDF/XML datasources (including APIs suitable for generation of HTML for end-user interfaces)
• Implementation of simple developer-centric RDF interface; implementation of Mozilla RDF API in Perl
• Implementation of indexed in-memory RDF database, RDF::RDFWeb::MemDB
• incorporation of multiple parsers for reading RDF/XML syntax (examples code demonstrates RSS newsfeed datasources)
• experimentation with and development of different flavours of querying over the stored (aggregated) RDF datasources, including a wrapper for W3C's RDF Perllib codebase with translators to support the Squish RDF query language
• Re-usability of codebase demonstrated through ILRT's SOSIG service, which has developed thesaurus tools using the same code
• implementation of a wrapper for the Redland RDF database system

Functionality Overview

- DATASOURCES
- API has been developed so datafiles can be loaded directly in RDF or from a source such as an RSS newsfeed, located by URL. 
(Q: "live" dataloading possibilities have been explored??)

- FILTERING
Sources can be classified according to common schemes they support, such as DC for example.
(Q: how is this useful specifically to Medcertain?)

- DATASOURCE PARSING -> RDF
- you can plugin various RDF parsers (eg Cara can be installed, or use W3C etc) - RDF parsers of choice can be plugged in via reference to their URL, thus returning an RDF graph (from RDF/XML) for aggregation into the API.


DATA AGGREGATION
RDF "graphs" returned from parsers can be combined in terms of their overlapping nodes and arcs.  Thesaurus issues are relevant here
(Q has there been any further development of thesaurus server - or do we have to say that in lieu of a sufficiently developed medical ratings vocabulary, alternative testbeds - s.a. SOSIG data - are (going to be) being used for proof of concept? 
Q Where does OMNI fit in?).


SERVER REQUESTS
- send out requests to servers for info? Any work on this? Or so far we rely on exposure of RSS channels as datasources? 

USER INTERFACE
- web view of the RDF - weblet? stuff

MEDCERTAIN APPLICATION
- note what Phil said about offering up ratings on sites - one of the medcert quality ratings vocab elements is on 1-5 scoring, Q for Dan: how can the RDFAPI tools be handy here?


PROGRESS?/FUTURE WORK??
- testing incomplete
- aggregation of ratings, recommendations to be developed
- query mechanisms under continuing development
- case studies to be done re deployability of software
- user interface under development
- addons - signing PGP on sources eg usergroups