Trainable Referring Expression Generation using Overspecification Preferences
This work addresses the data efficiency issue in referring expression generation for natural language processing applications, but it is incremental as it builds on existing methods by grouping speakers.
The paper tackled the problem of requiring extensive individual training data for speaker-dependent referring expression generation models by grouping speakers based on overspecification preferences, resulting in improved performance over personalized methods as shown in intrinsic evaluations.
Referring expression generation (REG) models that use speaker-dependent information require a considerable amount of training data produced by every individual speaker, or may otherwise perform poorly. In this work we present a simple REG experiment that allows the use of larger training data sets by grouping speakers according to their overspecification preferences. Intrinsic evaluation shows that this method generally outperforms the personalised method found in previous work.