KI 2009 Mashup Challenge

AI 2009 Mashup Challenge Contestants

Below you’ll find the list of mashups competing for our prizes. Have a look and try them out! The winner of the mashup challenge will be determined by popular vote on the KI 2009 conference. If you are attending this conference you’ll receive the ballot paper with your conference papers.

AISearch Firefox Plugin

The AIsearch Firefox add-on performs on-the-fly categorization of search results obtained when searching the Web with Google. The add-on marks a first step toward making browsers intelligent tools to access the Web since the category formation is performed entirely within the browser. As a basis for the add-on we have developed the first JavaScript library for information retrieval which includes a selection of well-known retrieval models, clustering algorithms, cluster validity measures, and topic identification algorithms.


EpiSPIDER stands for Semantic Processing and Integration of Distributed Electronic Resources for Epidemics and Disasters. It is an application that:

  1. Receives unstructured data from news sources: moderated list servers such as ProMED Mail, authoritative health news sources such as the World Health Organization (WHO) and the World Animal Health Information Database (WAHID) web sites, search engines such as Google and news aggregators such as Reuters, Moreover and DayPI and web 2.0 applications such as Twitter about epidemics and on emerging infectious diseases and disasters
  2. Collects country-related data from the United Nations Development Programme and CIA Factbook and topic specific data from Wikipedia. This enables EpiSPIDER to mashup news articles with country-specific data (e.g., human development index, GDP per capita income, etc.)
  3. Classifies unstructured data using the Digramic Bayesian Classifier into event, nonevent and spam – This provides EpiSPIDER with internal AI functionality.
  4. Passes on the the unstructured event and nonevent data to OpenCalais NLP web service to extract structured data (medical terms and locations) – This web service courtesy of Thomson-Reuters OpenCalais provides EpiSPIDER with the external AI functionality. Before OpenCalais, EpiSPIDER used custom developed natural language processing scripts to extract structured data from news articles. In addition, Google Maps API, Yahoo Maps API and Geonames provide EpiSPIDER with geolocation capabilities.
  5. Passes on the structured data to the U.S. National Library of Medicine Unified Medical Language System (UMLS) Knowledge Source Server web service API (experimental) to annotate news articles with concept unique identifiers.
  6. Converts structured information to visualization formats: JSON for SIMILE Exhibit Widgets and Google data format for Google Visualization API – This provides EpiSPIDER with mashup visualization functionality.
  7. Converts structured information to other formats for redistribution: KML for Google Earth, RSS and GeoRSS feeds, Twitter feed and mobile phone SMS.
  8. Links the structured information to other data sources such as Wikipedia and PubMED articles using a custom NLM web service. The latter links topics of news articles to their scientific article counterparts.

EpiSPIDER is currently used by health departments, local and international public health agencies, research institutions, federal agencies and academic centers to track emerging infectious disease epidemics worldwide.

IBIS – Intelligent Band Information System

The IBIS (Intelligent Band Information Service) platform is a mashup of different services, which provides band-related information, with special attention to the intelligent aggregation on the one hand and also be customizable by the user on the other hand.

The services currently included are: DBPedia – an extraction of semantic data from Wikipedia, YouTube, Amazon and Twitter. The user has a dashboard and can decide which of these services he wants to add. On some services such as DBPedia the user can also decide which information he wants to have, such as the current band members or the abstract of the band. The user can also decide how this information should be displayed. Services can be positioned freely on the dashboard and the components size is also editable. All these changes can be saved persistent on the IBIS server or on the own computer.


For the JOBAD demo, you need a recent Firefox browser (3.0 recommended, 3.5 is even better), as that is the only browser offering comprehensive MathML support.

  • Entry point to JOBAD-enriched computer science lecture notes. Try e.g. basic naive set theory.
    • you can look up the definitions of most symbols by right-clicking on the respective symbol and selecting “lookup definition” from the context menu
    • the context menu is known to work on Linux and Windows. Sometimes you may have to click more than once to activate it. It may not work on Mac OS. There, please right-click or Ctrl+click on a symbol and then hit Ctrl+1 to look up its definition
    • the close button of the definition popup is in its upper right corner – a bit hard to see, we’re working on that
  • Unit conversion (and also some definition lookup), powered by a different server backend
    • right-click on any number or unit, and you can convert the value to any unit available from the context menu (or hit Ctrl+4, Ctrl+5, etc.) – this can be repeated for further units, and undone
  • client-only scripts (no interaction with web services) that demonstrate subterm folding – fold subterms into …, and unfold them again

Abstract (PDF)

Try it here!


The goal of JobTweet is two enable the use of microblogging-technology for the job market. This is done through the identification of the ever increasing number of job offerings that are posted on twitter. JobTweet has gone live in March of 2009 and support the Twitter job search in German, English, French and Russian.

JobTweet is conceived as a mashup build on top of the free public services Twitter, Yahoo Pipes and WordPress. The actual result presentation is done with JavaScript.

JobTweet is an AI-Mashup since it makes use of DataMining technology to filter job postings from the stream of all twitter postings.

Try it here!

Leave a Comment

Previous post:

Next post: