HRI-JP’s entry into the SemEval 2015 shared task workshop was selected as the best paper for the SpaceEval spatial role labeling shared task, and the authors were invited to give a special presentation on their approach at the workshop.
SemEval is a workshop where teams from around the world gather to compete on various shared tasks to advance the state of the art in the field of natural language processing. SemEval 2015 included 17 different shared tasks with close to one hundred participating teams.
Spatial role labeling is the task of identifying spatial relations, such as location and motion, and their participants in texts. It is essential for understanding human language about location and motion and is important for many applications including robotics, navigation systems, and wearable computing.
By effectively applying word vectors and other lexico-syntactic information in a framework combining sequential labeling and multi-class classification, HRI-JP constructed one of the world’s best spatial role labeling systems, achieving top performance on real world evaluation with unannotated texts.
For more information, see the following publication:
Eric Nichols and Fadi Botros.
SpRL-CWW: Spatial Relation Classification with Independent Multi-class Models.
Proceedings of the 9th International Workshop on Semantic Evaluation (SemEval 2015).