Progressive Class Semantic Matching for Semi-supervised Text Classification

Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 291 - 302 July 10-15, 2022 ©2022 Association for Computational Linguistics Seed-Guided Topic Discovery with Out-of-Vocabulary Seeds Yu Zhang1, Yu Meng1, Xuan Wang1, Sheng Wang2, Jiawei Han1 1University of Illinois at Urbana-Champaign, IL, USA 2University of Washington, Seattle, WA, USA {yuz9,yumeng5,xwang174,hanj}@illinois.edu swang@cs.washington.edu Abstract Discovering latent topics from text corpora has been studied for decades. Many existing topic models adopt a fully unsupervised setting, and their discovered topics may not cater to users’ particular interests due to their inability of leveraging user guidance. Although there exist seed-guided topic discovery approaches that leverage user-provided seeds to discover topic- representative terms, they are less concerned with two factors: (1) the existence of out-of- vocabulary seeds and (2) the power of pre- trained language models (PLMs). In this paper, we generalize the task of seed-guided topic discovery to allow out-of-vocabulary seeds. We propose a novel framework, named SEE- TOPIC, wherein the general knowledge of PLMs and the local semantics learned from the input corpus can mutually benefit each other. Experiments on three real datasets from differ- ent domains demonstrate the effectiveness of SEETOPIC in terms of topic coherence, accu- racy, and diversity.1 1 Introduction Automatically discovering informative and coher- ent topics from massive text corpora is central to text analysis through helping users efficiently di- gest a large collection of documents (Griffiths and Steyvers, 2004) and advancing downstream appli- cations such as summarization (Wang et al., 2009, 2022), classification (Chen et al., 2015; Meng et al., 2020b), and generation (Liu et al., 2021). Unsupervised topic models have been the main- stream approach to topic discovery since the pro- posal of pLSA (Hofmann, 1999) and LDA (Blei et al., 2003). Despite their encouraging perfor- mance in finding informative latent topics, these topics may not reflect user preferences well, mainly due to their unsupervised nature. For example, given a collection of product reviews, a user may be specifically interested in product categories 1The code and datasets are available at https://github.com/yuzhimanhua/SeeTopic. Table 1: Three datasets (Cohan et al., 2020; McAuley and Leskovec, 2013; Zhang et al., 2017) from different domains and their topic categories (i.e., seeds). Red: Seeds never seen in the corpus (i.e., out-of-vocabulary). In all three datasets, a large proportion of seeds are out- of-vocabulary. Dataset Category Names (Seeds) SciDocs (Scientific Papers) cardiovascular diseases chronic kidney disease chronic respiratory diseases diabetes mellitus digestive diseases hiv/aids hepatitis a/b/c/e mental disorders musculoskeletal disorders neoplasms (cancer) neurological disorders Amazon (Product Reviews) apps for android books cds and vinyl clothing, shoes and jewelry electronics health and personal care home and kitchen movies and tv sports and outdoors video games Twitter (Social Media Posts) food shop and service travel and transport college and university nightlife spot residence outdoors and recreation arts and entertainment professional and other places (e.g., “books”, “electronics”), but unsupervised topic models may generate topics containing dif- ferent sentiments (e.g., “good”, “bad”). To con- sider users’ interests and …

Content truncated to 500 words