Home | Activities | Resources & Tools | Portal | Publications | Events | People | Links


Ita-Chronos is a rule-based system for the automatic recognition and normalization of temporal expressions appearing in an Italian written text.

The system, written in Lisp, takes a plain text as input (see Fig.1), and returns an in-line annotated text where the recognized time expressions are normalized according to the TIMEX2 annotation standard (see Fig.2).

Designed following a modular approach, ITA-Chronos resorts to a total of 981 rules, organized into separated sets in charge of dealing with a specific aspect of the problem. More specifically, while 409 basic detection rules are in charge of recognizing all possible time expression appearing in the input text, 572 normalization rules, divided into 10 specialized sets, are run to determine the value of each TIMEx2 attribute.

Fig. 1: Plain text

Fig. 2: Annotated text

ITA-Chronos participated in the Temporal Expression Recognition and Normalization Task at EVALITA 2007, obtaining the best scores both in the recognition (Precision: 95.7%, Recall: 89.8%, F-Measure: 92.6%) and normalization (Precision: 68.5%, Recall: 66.3%, F-Measure: 67.4%) tasks.


Maintainer: bentivofbk.eu
Last modified: Tue Aug 28 2007