Switching Contexts: Transportability Measures for NLP
This work addresses the need for better performance estimation in NLP when deploying models to new domains, but it is incremental as it builds on existing statistics without introducing a major breakthrough.
The paper tackles the problem of estimating NLP model performance in new contexts by proposing transportability measures based on domain similarity, demonstrating their application in Named Entity Recognition and Natural Language Inference tasks.
This paper explores the topic of transportability, as a sub-area of generalisability. By proposing the utilisation of metrics based on well-established statistics, we are able to estimate the change in performance of NLP models in new contexts. Defining a new measure for transportability may allow for better estimation of NLP system performance in new domains, and is crucial when assessing the performance of NLP systems in new tasks and domains. Through several instances of increasing complexity, we demonstrate how lightweight domain similarity measures can be used as estimators for the transportability in NLP applications. The proposed transportability measures are evaluated in the context of Named Entity Recognition and Natural Language Inference tasks.