Neural Metaphor Detection in Context
This work addresses the challenge of metaphor detection in natural language processing, which is important for applications like text understanding and generation, but it is incremental as it builds on standard neural models with improved context handling.
The authors tackled the problem of detecting metaphorical word use in context by developing end-to-end neural models, specifically BiLSTMs that operate on complete sentences, which achieved a new state-of-the-art on existing verb metaphor detection benchmarks and showed strong performance on joint prediction tasks.
We present end-to-end neural models for detecting metaphorical word use in context. We show that relatively standard BiLSTM models which operate on complete sentences work well in this setting, in comparison to previous work that used more restricted forms of linguistic context. These models establish a new state-of-the-art on existing verb metaphor detection benchmarks, and show strong performance on jointly predicting the metaphoricity of all words in a running text.