Neural Text Generation from Structured Data with Application to the Biography Domain (2016-03-24T00:00:00.000000Z)

TL;DR

A neural model for concept-to-text generation that scales to large, rich domains and significantly out-performs a classical Kneser-Ney language model adapted to this task by nearly 15 BLEU is introduced.

Abstract

This paper introduces a neural model for concept-to-text generation that scales to large, rich domains. We experiment with a new dataset of biographies from Wikipedia that is an order of magnitude larger than existing resources with over 700k samples. The dataset is also vastly more diverse with a 400k vocabulary, compared to a few hundred words for Weathergov or Robocup. Our model builds upon recent work on conditional neural language model for text generation. To deal with the large vocabulary, we extend these models to mix a fixed vocabulary with copy actions that transfer sample-specific words from the input database to the generated output sentence. Our neural model significantly out-performs a classical Kneser-Ney language model adapted to this task by nearly 15 BLEU.

Authors

Michael Auli

14 papers

David Grangier

6 papers

R. Lebret

2 papers

TL;DR

Abstract

Authors

References43 items

Attention with Intention for a Neural Network Conversation Model

What to talk about and how? Selective Generation using LSTMs with Coarse-to-Fine Alignment

Semantically Conditioned LSTM-based Natural Language Generation for Spoken Dialogue Systems

Neural Responding Machine for Short-Text Conversation

Show, Attend and Tell: Neural Image Caption Generation with Visual Attention

Deep visual-semantic alignments for generating image descriptions

From captions to visual concepts and back

Show and tell: A neural image caption generator

Unifying Visual-Semantic Embeddings with Multimodal Neural Language Models

Addressing the Rare Word Problem in Neural Machine Translation

Neural Machine Translation by Jointly Learning to Align and Translate

Numerical Computations with GPUs

Surface Realisation from Knowledge-Bases

The Stanford CoreNLP Natural Language Processing Toolkit

Fast and Robust Neural Network Joint Models for Statistical Machine Translation

Word Embeddings through Hellinger PCA

A Global Model for Concept-to-Text Generation

Scalable Modified Kneser-Ney Language Model Estimation

Introduction to "This is Watson"

Controlling User Perceptions of Linguistic Style: Trainable Generation of Personality Traits

A Probabilistic Forest-to-String Model for Language Generation from Typed Lambda Calculus Expressions

A Simple Domain-Independent Probabilistic Approach to Generation

Generative Alignment and Semantic Parsing for Learning from Ambiguous Supervision

Learning Semantic Correspondences with Less Supervision

Generating Approximate Geographic Descriptions

Automatic generation of weather forecast texts using comprehensive probabilistic generation-space models

Freebase: a collaboratively created graph database for structuring human knowledge

Generation by Inverting a Semantic Parser that Uses Statistical Machine Translation

Generating Multilingual Descriptions from Linguistically Annotated OWL Ontologies: the NaturalOWL System

Stochastic Language Generation Using WIDL-Expressions and its Application in Machine Translation and Summarization

Generation of Biomedical Arguments for Lay Readers

Aggregation via Set Partitioning for Natural Language Generation

Collective Content Selection for Concept-to-Text Generation

Choosing words in computer-generated weather forecasts

A Neural Probabilistic Language Model

Using Natural Language Generation for Navigational Assistance

Trainable approaches to surface natural language generation and their application to conversational dialog systems

Stochastic Language Generation for Spoken Dialogue Systems

Generation that Exploits Corpus-Based Statistical Knowledge

Building Natural-Language Generation Systems

Efficient BackProp

Comparing Automatic and Human Evaluation of NLG Systems

Content Planner Construction via Evolutionary Algorithms and a Corpus-based Fitness Function

Field of Study

Venue Information

Name

Type

URL

Alternate Names