Hi guys Just stumbled across this project http://www.s3db.org/ .
Was wandering whether anyone has seen it before and maybe has already thought about its similarities/differences to openehr. Many goals (interoperability, distributed systems, explicit/separate domain models, collaborative, bottom up, ..) and some components and implementation ideas (query language, form generator) seem similar. Obviously S3DB has a different technology focus (semantic web: RDF, SPARQL,...) and origin (life science research/clinical trials). My quick assessment is that S3DB uses a more decentralized/loose/WWWish/adhoc paradigm whereas openehr tries to manage/govern it's central domain models (archetypes) a-priori in a flexible (through reusability, archetype vs template) yet controlled way (via CKM). IMO this difference is due to the different main target areas: clinical medicine (imprecision/wrong data can be lethal) vs translational research (some imprecision is irrelevant in large enough data sets). The following paragraph from one of the S3DB papers (http://europepmc.org/articles/PMC3071752) describes a similar circumstance when comparing S3DB with caBIG. "4.2. Combining approaches to query federation The cancer Biomedical Informatics Grid (caBIG?) is a project aimed at enabling the sharing of cancer-related data using a federated query model, whereby different institutions working on related problems can share data either by adapting their local repositories to a set of data models provided by caBIG or by adopting one of the caBIG applications [39]. The various data sources in caBIG can then be queried simultaneously using the caGrid query language [40]. The caBIG approach offers the advantage of facilitating query assembly because it relies on a set of common data structures; this approach is therefore indicated for knowledge fields that are very well established. The linked data approach [9] does not impose a data model before integration is possible; instead data can be integrated using SPARQL queries as soon as an RDF representation is available. Although the latter approach requires that datasets be linked through the use of common terminologies before the assembly of SPARQL queries, it is better suited for knowledge areas that evolve quickly as it benefits from having novel data immediately available for integration. The two approaches could therefore greatly benefit from each other; indeed, a call for Semantic Web opportunities has been launched by the caBIG community [41]. The semCDI [42] and Corvus [43] projects, for example, have already developed extensive work towards modeling and integrating the various data models available at caBIG including the availability of SPARQL engines. The architecture described in this report could thus be easily integrated with caBIG datasets that are made available as RDF or if a SPARQL endpoint is provided. Indeed, one of the steps in that direction was mapping the terms within TCGA S3DB representation to NCI Thesaurus, also widely used by the caBIG community [44,45]." Keen to hear your thoughts or otherwise just to spread an interesting link... Cheers, Thilo -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.openehr.org/pipermail/openehr-technical_lists.openehr.org/attachments/20121112/a716cd51/attachment.html>
