Mitigating Temporal-Drift: A Simple Approach to Keep NER Models Crisp

Understand

Performance of neural models for named entity recognition degrades over time, becoming stale.

• This degradation is due to temporal drift, the change in our target variables' statistical properties over time.
• This issue is especially problematic for social media data, where topics change rapidly.
• In order to mitigate the problem, data annotation and retraining of models is common.