Generating Chinese Classical Poems with RNN Encoder-Decoder
This work addresses automated poetry generation for Chinese literature, but it is incremental as it builds on existing sequence-to-sequence methods.
The authors tackled generating Chinese classical poems by using an RNN encoder-decoder to create quatrains from a topic word, learning semantics, structure, and patterns without templates. They reported that their system outperformed other competitive systems, with attention capturing word associations and inverted training improving performance.
We take the generation of Chinese classical poem lines as a sequence-to-sequence learning problem, and build a novel system based on the RNN Encoder-Decoder structure to generate quatrains (Jueju in Chinese), with a topic word as input. Our system can jointly learn semantic meaning within a single line, semantic relevance among lines in a poem, and the use of structural, rhythmical and tonal patterns, without utilizing any constraint templates. Experimental results show that our system outperforms other competitive systems. We also find that the attention mechanism can capture the word associations in Chinese classical poetry and inverting target lines in training can improve performance.