Justifying convolutional neural network with argumentation for explainability

Saung Hnin Pwint Oo, Nguyen Duy Hung, Thanaruk Theeramunkong

Abstract: Convolutional neural network (CNN) has emerged as one of the most accurate methods for sentiment analysis, but it is largely uninterpretable, while case-based reasoning (CBR) is less accurate but offers interpretable outputs in the form of arguments from analogy. This paper presents an approach to combine these two methods, CNN for accuracy and CBR for explainability, using an assumption-based argumentation (ABA) framework. Our approach focuses on justifying CNN outputs using analogous sentences from CBR, while ensuring that the combined process is argumentative and hence self-explainable. To demonstrate the proposal, we construct a CNN model M1 and a CBR model M2 for sentiment analysis using different subsets of a dataset of which the remaining part is used for testing and comparing these input models with combined models. For an input sentence, if M1 and M2 predict the same sentiment, then the analogous sentence, which M2 finds, is used to explain the sentiment. If they give conflicting sentiments, a hybrid model M3 determines which one should be followed using a system of strict rules that takes into account how assertive M1 and M2 are. Another hybrid model M4, which is implemented by an ABA framework, improves on M3 by considering the probability distribution of the set of all labels from M1, and the second (or third) most similar sentences from M2. M3 and M4 preserve the accuracy of the CNN model (specifically, 88.32% and 88.28% in comparison with 87.59% accuracy of the CNN). They justify 69.95% and 74.53% of CNN outputs, respectively.

External IDs:dblp:journals/informaticaSI/OoHT23