Using GAN to Generate Sport News from Live Game Stats

One goal in artificial intelligence field is to create well-formed and human-like natural language text given data input and a specific goal. Some data-to-text solutions have been proposed and successfully used in real applied domains. Our work focuses on a new domain, Automatic Sport News Generating, which aims to produce sport news immediately after each match is over so that both time and labor can be saved on writing the news articles. We propose to use Generative Adversarial Networks (GAN) architecture for generating sport news based on game stats. Our model can automatically determine what is worth reporting and generate various appropriate descriptions about the game. We apply our approach to generate NBA (National Basketball Association) game news. Especially, This paper focuses on reporting the summary of game result and performance of players. Our model achieves good results on both tasks. To our best knowledge, this is the first work based on GAN to generate sports news using game statistics.

[1]  Ting Liu,et al.  Aspect Level Sentiment Classification with Deep Memory Network , 2016, EMNLP.

[2]  Rob Fergus,et al.  Deep Generative Image Models using a Laplacian Pyramid of Adversarial Networks , 2015, NIPS.

[3]  Michael White,et al.  EXEMPLARS: A Practical, Extensible Framework For Dynamic Text Generation , 1998, INLG.

[4]  Ben Poole,et al.  Categorical Reparameterization with Gumbel-Softmax , 2016, ICLR.

[5]  Yoshua Bengio,et al.  Generative Adversarial Nets , 2014, NIPS.

[6]  J. Coch,et al.  Quality tests for a mail generation system , 1995 .

[7]  Sebastian Nowozin,et al.  f-GAN: Training Generative Neural Samplers using Variational Divergence Minimization , 2016, NIPS.

[8]  Jeffrey Dean,et al.  Efficient Estimation of Word Representations in Vector Space , 2013, ICLR.

[9]  Albert Gatt,et al.  Automatic generation of textual summaries from neonatal intensive care data , 2009 .

[10]  Yann LeCun,et al.  Deep multi-scale video prediction beyond mean square error , 2015, ICLR.

[11]  Xiaojun Wan,et al.  Towards Constructing Sports News from Live Text Commentary , 2016, ACL.

[12]  José Coch.  System Demonstration Interactive Generation And Knowledge Administration In Multimeteo , 1998, INLG.

[13]  John Glover,et al.  Modeling documents with Generative Adversarial Networks , 2016, ArXiv.

[14]  Pieter Abbeel,et al.  InfoGAN: Interpretable Representation Learning by Information Maximizing Generative Adversarial Nets , 2016, NIPS.

[15]  Christian Ledig,et al.  Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network , 2016, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[16]  Ronald J. Williams,et al.  Simple Statistical Gradient-Following Algorithms for Connectionist Reinforcement Learning , 2004, Machine Learning.

[17]  Richard I. Kittredge,et al.  Using natural-language processing to produce weather forecasts , 1994, IEEE Expert.

[18]  Soumith Chintala,et al.  Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks , 2015, ICLR.

[19]  Lantao Yu,et al.  SeqGAN: Sequence Generative Adversarial Nets with Policy Gradient , 2016, AAAI.

[20]  Wojciech Zaremba,et al.  Improved Techniques for Training GANs , 2016, NIPS.