Combining Graph-based Dependency Features with Convolutional Neural Network for Answer Triggering
This work addresses the problem of selecting the best answer from candidates for question answering, representing an incremental improvement in a domain-specific area.
The paper tackles the answer triggering task by proposing a hybrid deep learning model that combines graph-based dependency features with CNN embeddings, achieving a 5.86% absolute F-score improvement on the WikiQA dataset compared to previous state-of-the-art models.
Answer triggering is the task of selecting the best-suited answer for a given question from a set of candidate answers if exists. In this paper, we present a hybrid deep learning model for answer triggering, which combines several dependency graph based alignment features, namely graph edit distance, graph-based similarity and dependency graph coverage, with dense vector embeddings from a Convolutional Neural Network (CNN). Our experiments on the WikiQA dataset show that such a combination can more accurately trigger a candidate answer compared to the previous state-of-the-art models. Comparative study on WikiQA dataset shows 5.86% absolute F-score improvement at the question level.