Explainable Agreement through Simulation for Tasks with Subjective Labels
This addresses the issue of overestimating classifier performance in subjective tasks like information retrieval for researchers, but it is incremental as it builds on existing agreement analysis methods.
The paper tackles the problem of evaluating classifiers on subjective labeling tasks by proposing a user simulation method to estimate the maximum achievable scores given noisy agreement data, concluding that a commonly-used controversy detection dataset is exhausted and requires more data to distinguish between techniques.
The field of information retrieval often works with limited and noisy data in an attempt to classify documents into subjective categories, e.g., relevance, sentiment and controversy. We typically quantify a notion of agreement to understand the difficulty of the labeling task, but when we present final results, we do so using measures that are unaware of agreement or the inherent subjectivity of the task. We propose using user simulation to understand the effect size of this noisy agreement data. By simulating truth and predictions, we can understand the maximum scores a dataset can support: for if a classifier is doing better than a reasonable model of a human, we cannot conclude that it is actually better, but that it may be learning noise present in the dataset. We present a brief case study on controversy detection that concludes that a commonly-used dataset has been exhausted: in order to advance the state-of-the-art, more data must be gathered at the current level of label agreement in order to distinguish between techniques with confidence.