Improving Match-LSTM for Machine Comprehension
نویسندگان
چکیده
Machine comprehension is a critical problem that lies on the frontier of natural language processing. The Stanford Question Answering Dataset (SQuAD), offers a set of questions and their answers created by humans through crowdsourcing. We implemented an end-to-end neural architecture for the task based on MatchLSTM and Pointer Net, inspired by previous work done by Wang and Jiang in Machine Comprehension Using Match-LSTM and Answer Pointer (2016) as well as Vinyals et al. in Pointer Net, a sequence-to-sequence model (2015). We were able to achieve an F1 score of 0.65 and EM score of 0.53 through our implementation, which leverages a dynamic-programming search to extract the final answer.
منابع مشابه
Machine Comprehension Using Match-LSTM and Answer Pointer
Machine comprehension of text is an important problem in natural language processing. A recently released dataset, the Stanford Question Answering Dataset (SQuAD), offers a large number of real questions and their answers created by humans through crowdsourcing. SQuAD provides a challenging testbed for evaluating machine comprehension algorithms, partly because compared with previous datasets, ...
متن کاملReading Comprehension Using Modified Match-LSTM
In this paper we detail our attemtps to replicate an extended Match-LSTM models that uses a dynamic loss function and a boundary based model with extensions based on observed parts of ynamic Coattention Models and a decoder that mirrors an AttentiveReader model.
متن کاملImplementation and Improvement of Match-LSTM in Question-Answering System
In this paper, we tackle the popular machine comprehension task derived from Stanford Question Answering Dataset (SQuAD), which consists of more than 100 thousand questions whose answers are segments of text from Wikipedia articles. We implemented the Machine comprehension system based on the model of[2], and made two extensions on top of that: 1. incorporate the word-level similarity in calcul...
متن کاملSing M Atch - Lstm and a Nswer P Ointer
Machine comprehension of text is an important problem in natural language processing. A recently released dataset, the Stanford Question Answering Dataset (SQuAD), offers a large number of real questions and their answers created by humans through crowdsourcing. SQuAD provides a challenging testbed for evaluating machine comprehension algorithms, partly because compared with previous datasets, ...
متن کاملAttention-based Recurrent Neural Networks for Question Answering
Machine Comprehension (MC) of text is an important problem in Natural Language Processing (NLP) research, and the task of Question Answering (QA) is a major way of assessing MC outcomes. One QA dataset that has gained immense popularity recently is the Stanford Question Answering Dataset (SQuAD). Successful models for SQuAD have all involved the use of Recurrent Neural Network (RNN), and most o...
متن کامل