Frame Shift Prediction
This work addresses the challenge of automatically creating multilingual FrameNets through annotation projection, which is incremental as it builds on existing methods for frame semantics.
The paper tackles the problem of predicting frame shifts in translation, which involves linguistic material evoking different frames across languages, and demonstrates that graph attention networks with auxiliary training can learn cross-linguistic frame-to-frame correspondence to predict these shifts.
Frame shift is a cross-linguistic phenomenon in translation which results in corresponding pairs of linguistic material evoking different frames. The ability to predict frame shifts enables automatic creation of multilingual FrameNets through annotation projection. Here, we propose the Frame Shift Prediction task and demonstrate that graph attention networks, combined with auxiliary training, can learn cross-linguistic frame-to-frame correspondence and predict frame shifts.