论文部分内容阅读
问答技术的重要挑战之一就是解决问题与答案之间的语义空白。机器翻译模型已经被证明能有效的提升解决问题与答案之间的语义空白。本文提出了一种基于注意机制的深度神经网络模型来解决问答系统中的答案选择任务。该模型采用了基于双向长短时记忆(Long short-term memory,LSTM)的编码解码模型,编码解码模型是一个被证明再机器翻译领域取得了突出的成绩。我们还在模型中应用了注意力机制来提升模型的效果。本文在一个公开数据集上验证了实验的有效性,同时通过结合该模型显著提高了问答系统的性能在TREC 2015 liveQA的任务中。
One of the key challenges of Q & A technology is to solve the semantic gap between questions and answers. Machine translation models have been shown to effectively improve the semantic gap between problem solving and answer. This paper presents a depth neural network model based on attention mechanism to solve the question answering system in the task of selecting answers. The model uses a code-decoding model based on Long short-term memory (LSTM), and the codec model is a proven achievement in machine-to-machine translation. We also apply attentional mechanisms in the model to improve the model’s effectiveness. This paper validates the experiment on a public dataset, and at the same time significantly improves the performance of the Q & A system in the mission of TREC 2015 liveQA by incorporating this model.