Additionally, we further validate the robustness of our method by the adversarial examples of AddSent and AddOneSent. Moreover, the overall performance of our proposed model surpasses the state-of-the-art on both datasets. This is the set of steps we would follow: Setup the module. The results show that the proposed MPGE is effective for both types of representation. We will try to use sentence embeddings to find out similar sentences from a given corpus. We conduct extensive experiments on two datasets, WikiQA and SQuAD. In contrast to list-learning experiments. By utilizing MPGE as a module, we construct two answer sentence selection models which are based on traditional representation and pre-trained representation, respectively. The present research determined that set-size effects are found when words are encoded in sentence contexts. Unlike previous work, which only models the relation between the question and each candidate sentence, we propose Multi-Perspective Graph Encoder (MPGE) to take the relations among the candidate sentences into account and capture the relations from multiple perspectives.
This paper focuses on the answer sentence selection task.
0 kommentar(er)
