Accurate Use of Label Dependency in Multi-Label Text Classification Through the Lens of Causality
This addresses bias issues in multi-label classification for text analysis applications, representing an incremental improvement over existing dependency-based methods.
The paper tackles the problem of prediction bias in multi-label text classification caused by models misusing label dependency correlations as shortcuts rather than properly fusing text information. The proposed CounterFactual Text Classifier (CFTC) eliminates this bias using causal inference, significantly outperforming baselines on three datasets.
Multi-Label Text Classification (MLTC) aims to assign the most relevant labels to each given text. Existing methods demonstrate that label dependency can help to improve the model's performance. However, the introduction of label dependency may cause the model to suffer from unwanted prediction bias. In this study, we attribute the bias to the model's misuse of label dependency, i.e., the model tends to utilize the correlation shortcut in label dependency rather than fusing text information and label dependency for prediction. Motivated by causal inference, we propose a CounterFactual Text Classifier (CFTC) to eliminate the correlation bias, and make causality-based predictions. Specifically, our CFTC first adopts the predict-then-modify backbone to extract precise label information embedded in label dependency, then blocks the correlation shortcut through the counterfactual de-bias technique with the help of the human causal graph. Experimental results on three datasets demonstrate that our CFTC significantly outperforms the baselines and effectively eliminates the correlation bias in datasets.