Automated Sequence Labeling for Causality Extraction in Medical Domain

The main factor hindering the progress of causality extraction research is the lack of a unified causal sequence labeling technique. One of the most effective methods for solving sequence labeling problems is Conditional Random Field-based neural models. Conditional Random Field are typically used for structured labeling or sequence prediction tasks, aiming to predict a structured output based on a set of input features. This research proposes a method for automated causality sequence labeling using a Conditional Random Field that utilizes the Part of Speech features. We also propose a new label to improve the results. We evaluate our method in the medical domain with Indonesian language datasets and compare it with traditional machine learning algorithm and rule-based methods to assess the performance. Experimental results show that our proposed method achieves the highest F1 score of 0.80 for causality sequence labeling when compared to traditional methods.