Seminar Transcript, 2005-05-09


Picture of the week
Cube with Penrose triangles
Wrong Cube
Article of the week (on Ajax or Rich Web Applications)
Finally, Sisyphus, There's Help for Those Internet Forms
Our Skype IDs
davis.b, extropy, kalvis.apsitis, uldis.bojars.
Geography and time
Raiņa bulvaris 29; 2005-05-09 17:30


Ontology clearly is at the heart for most SW applications, so we can ask what are the other essential things, and how they are interconnected. The view on how a typical SW application looks has evolved over time. Some software and tools for practical work with SW were discussed. Another topic concerned the recent activities by RIXC - as told by Dāvis. At the end we discussed a nice picture (uldis.bojars has it), on how P2P, Ontologies, People Networks, Desktop Applications can do the join-semilattice thing.

The Current View on SW applications (presented by Uldis)

  1. Ontology - it is created anew, or some set of existing ontologies is used. (For example Dublin Core metadata is one frequently used ontology. FOAF and its friends could be another.
  2. Data sources - they can be in people's heads, but should be represented into some preliminary format to be usable by a SW application
  3. Data wrapper/exporter to RDF - can transform legacy data and various non-RDF data (which are used by non-SW systems) to RDF format.
  4. RDF transport - since SW (by definition) is distributed knowledge management, there should be either a crawler/robot gathering all RDF data accordingly to some criterion, or a way to query various data sources and wrappers. Latter solution is practical, if remote RDF data is huge and/or is updated frequently.
  5. RDF repository - a database or similar amenity, where the RDF data is stored.
  6. [Inference engine and other SW application logic layer - make various requests to repository, sort and process the answers]
  7. User interface - a friendly way to browse and query the knowledge

The application logic layer [in square brackets] was not mentioned in the original list, but we believe that it should be a separate layer, as it may happen that user of a SW application needs something more than just querying RDF repository accordingly to some standard query langage. There may be an additional logic, where the queries issued by the user are fed back in the knowledge management system. I.e. each user action may add some knowledge to the SW system.

Old View on SW applications

Here is version, what Kalvis remembers from some AIFB Karlsruhe paper. The publication itself has not been found yet.

If you wander, what are the ovals and rectangles; visit some resources on data flow diagrams (see references) or agile modeling. Or enjoy Microsoft Visio with its Gane/Sarson diagrams ;)

Stack of RDF-related data

If the lower layers in the Web are represented by HTTP protocol, URIs and byte/unicode streams, then XML, RDF, etc. are elaborations of this Web thing. Here is the stack of SW data accordingly to the paper: Sudhir Agarwal, Rudi Studer, Raphael Volz The Semantic Web - Methods, Application and Tools (see

Stack of RDF-data

Tools and approaches to deal with SW processes

  • Knowledge management on a desktop machine can be described with Protege
  • RDF repository - Jena (in Java) or some alternative in Python (ask Uldis, he may know)
  • Metadata browsing - Simile&Fresnel
  • Sharing bookmarks and pictures -,
  • Searching SW in a consistent way - Hannes Gassert and his references to SWSE (SW Search Engine)


Why do we need a reference to Valdis Krebs? He can be found with Google, but his relation to SW is obscure.


  2. - introduction to REST
  3. - YARS (Yet Another RDF Store)
  4. - on Gane&Sarson data flow diagrams
  5. - create public lists of links to resources, access them by tags.
  6. - Closed World Machine in Python
  7. - SIMILE, 'faceted' browsing of metadata
  8. - SPARQL, a query language
  9. - a way to share and browse photos
  10. - Protege, a way to build ontologies
  11. - HP Jena framework to deal with RDF data
  12. - Semantic Web search engine
  13. - blogs by Hannes Gassert

Probably need to gather all the links scattered in all the text ...

Lapa mainīta 2005-05-15 23:20:16