References & Citations
Computer Science > Computation and Language
Title: TAN-NTM: Topic Attention Networks for Neural Topic Modeling
(Submitted on 2 Dec 2020 (this version), latest version 9 Jul 2021 (v2))
Abstract: Topic models have been widely used to learn representations from text and gain insight into document corpora. To perform topic discovery, existing neural models use document bag-of-words (BoW) representation as input followed by variational inference and learn topic-word distribution through reconstructing BoW. Such methods have mainly focused on analysing the effect of enforcing suitable priors on document distribution. However, little importance has been given to encoding improved document features for capturing document semantics better. In this work, we propose a novel framework: TAN-NTM which models document as a sequence of tokens instead of BoW at the input layer and processes it through an LSTM whose output is used to perform variational inference followed by BoW decoding. We apply attention on LSTM outputs to empower the model to attend on relevant words which convey topic related cues. We hypothesise that attention can be performed effectively if done in a topic guided manner and establish this empirically through ablations. We factor in topic-word distribution to perform topic aware attention achieving state-of-the-art results with ~9-15 percentage improvement over score of existing SOTA topic models in NPMI coherence metric on four benchmark datasets - 20NewsGroup, Yelp, AGNews, DBpedia. TAN-NTM also obtains better document classification accuracy owing to learning improved document-topic features. We qualitatively discuss that attention mechanism enables unsupervised discovery of keywords. Motivated by this, we further show that our proposed framework achieves state-of-the-art performance on topic aware supervised generation of keyphrases on StackExchange and Weibo datasets.
Submission history
From: Milan Aggarwal [view email][v1] Wed, 2 Dec 2020 20:58:04 GMT (772kb,D)
[v2] Fri, 9 Jul 2021 10:32:01 GMT (4401kb,D)
Link back to: arXiv, form interface, contact.