Data from ILP '98 Paper

Background

Here you'll find the data used for the ILP '98 paper Combining Statistical and Relational Methods for Learning in Hypertext Domains. The data consists of relations suitable for providing to FOIL, as well as the complete text of all the web pages and also of anchors and the text surrounding anchors.

We've provided two views of the data:

The complete directory structure can be inspected from the comfort of your web browser. The gzipped versions of the files required for running FOIL on this data total 26.5 megabytes.
A tar file (of gzipped constituent files) containing all the data can also be obtained. This tar file is nearly 28 megabytes.

The ReadMe file explains all the various components. If you have any questions, please email the authors.

See the Web->KB project home page for more information on this data set and related work.

Seán Slattery

Last modified: Tue Aug 18 18:14:36 EDT 1998