Gentics Blogosphere

I’m currently attending the 2nd IKS workshop in rome. Follow my tweed or #iks for live updates!

The IKS (intelligent knowledge stack) project’s goal is to bring semantics to european content managment systems by working together with CMS vendors on these topics, and build an open source implementation of the stack.

The 1st workshop (may in salzburg) was about requirements. This one is about semantic search, and will cover 11 vendor demos within 2 days.

Peter Mika, Yahoo! Research

Peter Mika, researcher and data architecht of Yahoo! started the workshop with a presentation on basics and facts about semantic search.

Yahoo! research is working on ambiguouity, multimedia search, imprecise or over precise search (jim/james), search for description (product search or search for yourself without using your name), searches that require aggregation (world temperature 2020)

He liked “Zemanta”, a plugin for popular blogging plattforms processing and enriching the content the user writes.

Some metadata (rdfa, erdf, tags, hcard, adr, hatom, xfn, geo, hreview) statistics: currently in average one of the top 10 search result has some metadata. If youtube had a microformat, it would be the 5th most important.

A semantic interpretation of todays keyword queries:
40% entity queries (starbucks palo alto)
12% type query (plubmers barcelona)
5% attribute query (zip code waterville maine)
0.7% relation query (chris brown rihanna)
36% other keyword query (nightlife barcelona)
6% uninterpretable

Some current semantic search engines:
search open web data: sindice, sigma, google squared
search closed world: wolfram alpha
web search: yahoo’s search monkey (boss, yql), google rich snipplet’s

Yahoo! has a discussion group on their current semantic work.
pmikaatyahoo-inc.com
search monkey website, suggestions

Let’s get started with the Demos!

Deri - Aexel Polleres and Jürgen Umbrich

Install drupal, and get semantics for free with these modules:
rdfcck, evoc, sparql_ep, rdfproxy.
You can autocreate rdf auto of drupal content objects, and use sparql queries based on this data.

There is an online prototype

The jury honoured the concept, but missed the user perspective and benefits.

Trialox - Reto Bachmann-Gmür

is a radically RDF based system. Clerezza is an open source foundation to easily build resfull semantic web applications. Based on OSGI, Sesame, TDB, Mulgara, JAX RS.

The demo showed the developers point of view, sparql queries, but no enduser experience either. The jury said, SPARQL is good, but should be under the hood. It’s like trying to teach endusers SQL 20 years ago.

Kiwi - Rolf Sint

Knowledge In A Wiki - an EU project. Salzburger Nachrichten news have been imported, geotagged and displayed in clusters on google maps. The wiki displays incoming and outgoing links, user activities and features personalized search results.

I didn’t see much semantic aspects here. Neither did the jury.

Stay tuned for day 2 …

Laurin Herlt