Use of Co-occurrences for Temporal Expressions Annotation
Authors
Abstract
The annotation or extraction of temporal information from text documents is becoming increasingly important in many natural language processingapplications such as text summarization, information retrieval, question
answering, etc.. This paper presents an original method for easy recognition
of temporal expressions in text documents. The method creates semantically classified temporal patterns, using word co-occurrences obtained from training corpora and a pre-defined seed keywords set, derived from the used language temporal references. A participation on a Portuguese named entity evaluation contest showed promising effectiveness and efficiency results. This approach can be adapted to recognize other type of expressions or languages, within other contexts, by defining the suitable word sets and training corpora.