Proceedings of the Workshop on Methods for Optimizing and Evaluating Neural Language Generation

NeuralGen is the first workshop on Methods for Optimizing and Evaluating Neural Language Generation, being held at NAACL 2019 in Minneapolis, Minnesota. The goal of this workshop is to discuss new frontiers for language generation that address some of the recurring problems in existing techniques (eg. bland, repetitive language). More specifically, this workshop is aimed at sharing novel modeling techniques that go beyond maximum likelihood training, new techniques for robust evaluation and interpretation of model output, and strategies for generalization of generation systems.We are pleased to have received 42 submissions, covering a wide range of topics related to modeling, evaluation and analysis of novel generation systems. 17 of the submissions have been accepted into the final program (approximately 40% acceptance rate). The workshop schedule includes 11 archival papers and 17 poster presentations. We are also thankful to have seven invited speakers: Kyunghyun Cho, He He, Graham Neubig, Yejin Choi, Alexander Rush, Tatsunori Hashimoto, and Hal Daumé III. The workshop also includes a panel discussion from the speakers and spotlight talks for a selection of accepted papers.