Introduction to the 3rdROMAND Workshop on Robust Methods in Analysis of Natural Language Data

Abstract

Robustness in Computational Linguistics has been recently recognized as a central issue for the design of interactive applications based on natural language communication. If a failure of the system can be acceptable in batch applications requiring a human intervention, an on-line system should be capable of dealing with unforeseen situations in a more flexible way. When we talk about system failure we do not think at inherent program failures like infinite loops or system exception, we consider, rather, failures related to the processing of the input and its assimilation in the system's knowledge base. A failure of this kind means simply that the system does not understand the input. The automated analysis of natural language data has become a central issue in the design of Intelligent Information Systems. Processing unconstrained natural language data is still considered an AI-hard task. However, various analysis techniques have been proposed in order to address specific aspects of natural language. In particular, recent interest has been focused on providing approximate analysis techniques, assuming that when perfect analysis is not possible, partial results may be still very useful.