INDEED: a System for Relational Rule Induction in Domains with Rich Space of Constant Terms
| Authors | |
|---|---|
| Year of publication | 2003 |
| Type | Article in Proceedings |
| Conference | Proceedings of the Work-in-Progress Track at the 13th International Conference on Inductive Logic Programming |
| MU Faculty or unit | |
| Citation | |
| Field | Informatics |
| Keywords | relational rule induction; morphological disambiguation |
| Description | This paper presents INDEED, a new ILP system suitable for relational rule induction in domains with rich and possibly structured space of constant terms. These constants can be arbitrary terms, not only atoms. INDEED efficiently introduces the constants into a body of a clause and is capable of refining them with respect to a given background knowledge and user-defined specification. The primary application field for INDEED is morpho-syntactic tagging of highly inflectional (e.g. Slavic) languages. Nevertheless, INDEED has been applied also to well-known benchmarks in the area of ILP with highly competitive results. This paper presents experiments with tagging the Slovene version of the Multext-East corpus and two structure-activity relationship tasks from molecular biology: mutagenesis and carcinogenesis. |
| Related projects: |