Like a Good Nearest Neighbor: Practical Content Moderation and Text Classification

Anonymous

Abstract: Text classification systems have impressive capabilities but are infeasible to deploy and use reliably due to their dependence on prompting and billion-parameter language models. SetFit (Tunstall et al., 2022) is a recent, practical approach that fine-tunes a Sentence Transformer under a contrastive learning paradigm and achieves similar results to more unwieldy systems. Inexpensive text classification is important for addressing the problem of domain drift in all classification tasks and especially in detecting harmful content, which plagues social media platforms. Here, we propose Like a Good Nearest Neighbor (LaGoNN), a modification to SetFit that introduces no learnable parameters but alters input text with information from its nearest neighbor, for example, the label and text, in the training data, making novel data appear similar to an instance on which the model was optimized. LaGoNN is effective at flagging undesirable content and text classification and improves SetFit's performance. To demonstrate LaGoNN's value, we conduct a thorough study of text classification systems in both the context of content moderation under four label distributions and in a more general classification setting.

Paper Type: long

Research Area: NLP Applications

Languages Studied: English