Identifying Semantically Difficult Samples to Improve Text Classification
This work addresses the challenge of handling ambiguous or confusing samples in text classification, which is incremental as it builds on existing methods for analyzing semantic embeddings.
The paper tackles the problem of improving text classification by identifying semantically difficult samples, defined as non-obvious cases in semantic embedding space, and proposes a penalty function to measure sample difficulty, resulting in consistent improvements of up to 9% across 13 standard datasets.
In this paper, we investigate the effect of addressing difficult samples from a given text dataset on the downstream text classification task. We define difficult samples as being non-obvious cases for text classification by analysing them in the semantic embedding space; specifically - (i) semantically similar samples that belong to different classes and (ii) semantically dissimilar samples that belong to the same class. We propose a penalty function to measure the overall difficulty score of every sample in the dataset. We conduct exhaustive experiments on 13 standard datasets to show a consistent improvement of up to 9% and discuss qualitative results to show effectiveness of our approach in identifying difficult samples for a text classification model.