An Improved Neural Baseline for Temporal Relation Extraction
This provides a strong baseline for temporal relation extraction, addressing a challenging natural language understanding task with limited high-quality training data.
The paper tackles the problem of determining temporal relations between events in natural language by proposing a new neural system that achieves about 10% absolute improvement in accuracy (25% error reduction) over the previous best system on two benchmark datasets.
Determining temporal relations (e.g., before or after) between events has been a challenging natural language understanding task, partly due to the difficulty to generate large amounts of high-quality training data. Consequently, neural approaches have not been widely used on it, or showed only moderate improvements. This paper proposes a new neural system that achieves about 10% absolute improvement in accuracy over the previous best system (25% error reduction) on two benchmark datasets. The proposed system is trained on the state-of-the-art MATRES dataset and applies contextualized word embeddings, a Siamese encoder of a temporal common sense knowledge base, and global inference via integer linear programming (ILP). We suggest that the new approach could serve as a strong baseline for future research in this area.