Story Ending Generation with Incremental Encoding and Commonsense Knowledge

Story ending generation is a strong indication of story comprehension. This task requires not only to understand the context clues which plays the most important role in planning the plot, but also to handle implicit knowledge to make a reasonable, coherent story. In this paper, we devise a novel model for story ending generation. The model adopts an incremental encoding scheme with multi-source attention to deal with context clues spanning in the story context. In addition, the model is empowered with commonsense knowledge through multi-source attention to produce reasonable story endings. Experiments show that our model can generate more reasonable story endings than state-of-the-art baselines.