A convolutional route to abbreviation disambiguation in clinical text

It is found that for some common abbreviations, sense distributions mismatch between the test and auto generated training data, and mitigating the mismatch significantly improved the model accuracy, and neural networks can simplify development of a practical abbreviation disambiguation system.

Abstract

The neural network models work well in disambiguating abbreviations in clinical narratives, and they are robust across datasets. This avoids feature-engineering for each dataset. Coupled with an enhanced auto-training data generation, neural networks can simplify development of a practical abbreviation disambiguation system.