SIRIUS-LTG-UiO at SemEval-2018 Task 7: Convolutional Neural Networks with Shortest Dependency Paths for Semantic Relation Extraction and Classification in Scientific Papers

TL;DR

This paper introduces a CNN-based system that uses shortest dependency paths for semantic relation extraction and classification in scientific papers, achieving competitive F1 scores and ranking third in the SemEval-2018 Task 7.

Contribution

The novel use of shortest dependency path embeddings with CNNs for relation extraction and classification in scientific texts.

Findings

Achieved F1 scores of 76.7 and 83.2 on clean and noisy data.

Ranked 3rd in all three sub-tasks of SemEval-2018 Task 7.

Demonstrated effectiveness of dependency path-based CNNs for scientific relation tasks.

Abstract

This article presents the SIRIUS-LTG-UiO system for the SemEval 2018 Task 7 on Semantic Relation Extraction and Classification in Scientific Papers. First we extract the shortest dependency path (sdp) between two entities, then we introduce a convolutional neural network (CNN) which takes the shortest dependency path embeddings as input and performs relation classification with differing objectives for each subtask of the shared task. This approach achieved overall F1 scores of 76.7 and 83.2 for relation classification on clean and noisy data, respectively. Furthermore, for combined relation extraction and classification on clean data, it obtained F1 scores of 37.4 and 33.6 for each phase. Our system ranks 3rd in all three sub-tasks of the shared task.