Neural Responding Machine for Short-Text Conversation (2015-03-08T00:00:00.000000Z)

TL;DR

Empirical study shows that NRM can generate grammatically correct and content-wise appropriate responses to over 75% of the input text, outperforming state-of-the-arts in the same setting, including retrieval-based and SMT-based models.

Abstract

We propose Neural Responding Machine (NRM), a neural network-based response generator for Short-Text Conversation. NRM takes the general encoder-decoder framework: it formalizes the generation of response as a decoding process based on the latent representation of the input text, while both encoding and decoding are realized with recurrent neural networks (RNN). The NRM is trained with a large amount of one-round conversation data collected from a microblogging service. Empirical study shows that NRM can generate grammatically correct and content-wise appropriate responses to over 75% of the input text, outperforming state-of-the-arts in the same setting, including retrieval-based and SMT-based models.

Authors

Lifeng Shang

2 papers

Zhengdong Lu

2 papers

Hang Li

2 papers

TL;DR

Abstract

Authors

References32 items

Neural Machine Translation

A Neural Network Approach to Context-Sensitive Generation of Conversational Responses

LSTM: A Search Space Odyssey

Empirical Evaluation of Gated Recurrent Neural Networks on Sequence Modeling

Sequence to Sequence Learning with Neural Networks

Neural Machine Translation by Jointly Learning to Align and Translate

An Information Retrieval Approach to Short Text Conversation

Learning Phrase Representations using RNN Encoder–Decoder for Statistical Machine Translation

A Dataset for Research on Short-Text Conversations

Joint Language and Translation Modeling with Recurrent Neural Networks

Recurrent Continuous Translation Models

Generating Sequences With Recurrent Neural Networks

Reinforcement Learning of Question-Answering Dialogue Policies for Virtual Museum Guides

Data-Driven Response Generation in Social Media

Moses: Open Source Toolkit for Statistical Machine Translation

Partially observable Markov decision processes for spoken dialog systems

Training linear SVMs in linear time

A Systematic Comparison of Various Statistical Alignment Models

Bleu: a Method for Automatic Evaluation of Machine Translation

NJFun- A Reinforcement Learning Spoken Dialogue System

Long Short-Term Memory

Fundamental Statistics for the Behavioral Sciences

Recurrent neural network based language model

A Sino-German λ 6 cm polarization survey of the Galactic plane I . Survey strategy and results for the first survey region

Fundamental Statistics for Behavioral Sciences

Measuring nominal scale agreement among many raters.

DOI: 10.1017/S000000000000000 Printed in the United Kingdom A Survey of Statistical User Simulation Techniques for Reinforcement-Learning of Dialogue Management Strategies

Scenario Dependence : Responses can depend on a speciﬁc scenario but should not contradict the ﬁrst three criteria; From Table 2,

higher kappa value 0.448, which is larger than 0 and considered as “Moderate agreement”

the responses generated by the SMT often have the ﬂuency and grammatical errors

a) Grammar and Fluency : Responses should be natural language and free of any ﬂuency or grammatical errors

b) Logic Consistency : Responses should be logically consistent with the test post

Field of Study

Journal Information

Name

Volume

Venue Information

Name

Type

URL

Alternate Names