DOC: Deep Open Classification of Text Documents

Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, pages 2911–2916 Copenhagen, Denmark, September 7–11, 2017. c⃝2017 Association for Computational Linguistics DOC: Deep Open Classification of Text Documents Lei Shu, Hu Xu, Bing Liu Department of Computer Science University of Illinois at Chicago {lshu3, hxu48, liub}@uic.edu Abstract Traditional supervised learning makes the closed-world assumption that the classes appeared in the test data must have ap- peared in training. This also applies to text learning or text classification. As learning is used increasingly in dynamic open envi- ronments where some new/test documents may not belong to any of the training classes, identifying these novel documents during classification presents an important problem. This problem is called open- world classification or open classification. This paper proposes a novel deep learning based approach. It outperforms existing state-of-the-art techniques dramatically. 1 Introduction A key assumption made by classic supervised text classification (or learning) is that classes appeared in the test data must have appeared in training, called the closed-world assumption (Fei and Liu, 2016; Chen and Liu, 2016). Although this as- sumption holds in many applications, it is violated in many others, especially in dynamic or open en- vironments. For example, in social media, a classi- fier built with past topics or classes may not be ef- fective in classifying future data because new top- ics appear constantly in social media (Fei et al., 2016). This is clearly true in other domains too, e.g., self-driving cars, where new objects may ap- pear in the scene all the time. Ideally, in the text domain, the classifier should classify incoming documents to the right existing classes used in training and also detect those doc- uments that don’t belong to any of the existing classes. This problem is called open world classi- fication or open classification (Fei and Liu, 2016). Such a classifier is aware what it does and does not know. This paper proposes a novel technique to solve this problem. Problem Definition: Given the training data D = {(x1, y1), (x2, y2), . . . , (xn, yn)}, where xi is the i-th document, and yi ∈ {l1, l2, . . . , lm} = Y is xi’s class label, we want to build a model f(x) that can classify each test instance x to one of the m training or seen classes in Y or reject it to in- dicate that it does not belong to any of the m train- ing or seen classes, i.e., unseen. In other words, we want to build a (m + 1)-class classifier f(x) with the classes C = {l1, l2, . . . , lm, rejection}. There are some prior approaches for open clas- sification. One-class SVM (Sch¨olkopf et al., 2001; Tax and Duin, 2004) is the earliest approach. However, as no negative training data is used, one- class classifiers work poorly. Fei and Liu (2016) proposed a Center-Based Similarity (CBS) space learning method (Fei and Liu, 2015). This method first computes a center for …

Content truncated to 500 words