Guided Generation of Cause and Effect
This work addresses the need for better causal reasoning in NLP, though it is incremental as it builds on existing methods and resources.
The authors tackled the problem of generating cause-and-effect sentences by introducing a conditional text generation framework, which improved a state-of-the-art causal reasoning model by 3 points on the COPA challenge set.
We present a conditional text generation framework that posits sentential expressions of possible causes and effects. This framework depends on two novel resources we develop in the course of this work: a very large-scale collection of English sentences expressing causal patterns CausalBank; and a refinement over previous work on constructing large lexical causal knowledge graphs Cause Effect Graph. Further, we extend prior work in lexically-constrained decoding to support disjunctive positive constraints. Human assessment confirms that our approach gives high-quality and diverse outputs. Finally, we use CausalBank to perform continued training of an encoder supporting a recent state-of-the-art model for causal reasoning, leading to a 3-point improvement on the COPA challenge set, with no change in model architecture.