TkTC: A framework for top-k text classification of multimedia computing in wireless networks

Kai Wang, Yuqi Liu, Bin Cao, Jing Fan

Abstract: The development of wireless networks and multimedia computing generates a large number of texts, which makes text classification become an important task. However, due to the complex semantic information of the text, traditional text classification that aims to predict the most possible label among all labels may not predict the correct label for the text. Thus, we hope the classifier could “recommend” k labels for each text, as long as the ground truth label is included, we call this top-k text classification. Recent work of top-k classification focuses on constructing a top-k classification loss function without considering the position of the ground truth label. In this paper, we propose $$\hbox{T}k\hbox{TC}$$ T k TC : a framework for top-k text classification, where a novel loss function that considers the position of ground truth label and the number of prediction simultaneously. Finally, we conduct extensive experiments on different datasets, and the experimental results show that our method outperforms state-of-the-art baselines in most cases.