PIRE

Probabilistic IR Engine

Contact Persons:

• Dipl.-Inform. Henrik Nottelmann
• Prof. Dr.-Ing. Norbert Fuhr

PIRE is a flexible, extensible IR engine which is based on the combination of logics and probability theory. As part of our java-unidu project, it is available under an open source licence (Apache Licence 2.0), which allows easy integration into other projects.

PIRE has a public mailing list for discussing everything about PIRE and advertising new releases.

The publications listed below give an insight into PIRE. The ECIR 2005 paper is a shorter version of the technical report. The latter does not only explain the theoretical background but also the code in detail. In addition, also a HTML version of the implementation-specific parts of the technical reports is available.

Examples of PIRE are included in the de.unidu.is.retrieval.pire.examples package.

Publications

H. Nottelmann (2005).

PIRE: An extensible IR engine based on probabilistic Datalog. In ECIR:05

Henrik Nottelmann (2005).

Inside PIRE: An extensible, open-source IR engine based on probabilistic logics. Technical Report, University of Duisburg-Essen

H. Nottelmann; N. Fuhr (2003).

From uncertain inference to probability of relevance for advanced IR applications. In ECIR:03

Diploma, master and bachelor theses

Only in german!

Information Retrieval im Semantic Web

Finished diploma thesis

Software

PIRE in java-unidu

PIRE version is included in the latest java-unidu release.