In this paper we propose and investigate a novel end-to-end method for
automatically generating short email responses, called Smart Reply. It
generates semantically diverse suggestions that can be used as complete email
responses with just one tap on mobile. The system is currently used in Inbox by
Gmail and is responsible for assisting with 10% of all mobile responses. It is
designed to work at very high throughput and process hundreds of millions of
messages daily. The system exploits state-of-the-art, large-scale deep
learning.
We describe the architecture of the system as well as the challenges that we
faced while building it, like response diversity and scalability. We also
introduce a new method for semantic clustering of user-generated content that
requires only a modest amount of explicitly labeled data.
Description
Smart Reply: Automated Response Suggestion for Email
%0 Journal Article
%1 kannan2016smart
%A Kannan, Anjuli
%A Kurach, Karol
%A Ravi, Sujith
%A Kaufmann, Tobias
%A Tomkins, Andrew
%A Miklos, Balint
%A Corrado, Greg
%A Lukacs, Laszlo
%A Ganea, Marina
%A Young, Peter
%A Ramavajjala, Vivek
%D 2016
%K NL lstm
%T Smart Reply: Automated Response Suggestion for Email
%U http://arxiv.org/abs/1606.04870
%X In this paper we propose and investigate a novel end-to-end method for
automatically generating short email responses, called Smart Reply. It
generates semantically diverse suggestions that can be used as complete email
responses with just one tap on mobile. The system is currently used in Inbox by
Gmail and is responsible for assisting with 10% of all mobile responses. It is
designed to work at very high throughput and process hundreds of millions of
messages daily. The system exploits state-of-the-art, large-scale deep
learning.
We describe the architecture of the system as well as the challenges that we
faced while building it, like response diversity and scalability. We also
introduce a new method for semantic clustering of user-generated content that
requires only a modest amount of explicitly labeled data.
@article{kannan2016smart,
abstract = {In this paper we propose and investigate a novel end-to-end method for
automatically generating short email responses, called Smart Reply. It
generates semantically diverse suggestions that can be used as complete email
responses with just one tap on mobile. The system is currently used in Inbox by
Gmail and is responsible for assisting with 10% of all mobile responses. It is
designed to work at very high throughput and process hundreds of millions of
messages daily. The system exploits state-of-the-art, large-scale deep
learning.
We describe the architecture of the system as well as the challenges that we
faced while building it, like response diversity and scalability. We also
introduce a new method for semantic clustering of user-generated content that
requires only a modest amount of explicitly labeled data.},
added-at = {2016-08-22T21:43:36.000+0200},
author = {Kannan, Anjuli and Kurach, Karol and Ravi, Sujith and Kaufmann, Tobias and Tomkins, Andrew and Miklos, Balint and Corrado, Greg and Lukacs, Laszlo and Ganea, Marina and Young, Peter and Ramavajjala, Vivek},
biburl = {https://www.bibsonomy.org/bibtex/2eea3d3f491f2537b10b122dc57ec6ae1/demoncoder38},
description = {Smart Reply: Automated Response Suggestion for Email},
interhash = {204bf6e34ea20e631d3f4434451a0b2f},
intrahash = {eea3d3f491f2537b10b122dc57ec6ae1},
keywords = {NL lstm},
note = {cite arxiv:1606.04870Comment: Accepted to KDD 2016},
timestamp = {2016-08-22T21:43:36.000+0200},
title = {Smart Reply: Automated Response Suggestion for Email},
url = {http://arxiv.org/abs/1606.04870},
year = 2016
}