Unsupervised Latent Dirichlet Allocation for supervised question classification
作者：
Highlights：
• We introduce a new classification algorithm based on LDA topic models.
• The model benefits from a topic-categories mixture model in addition to the original LDA.
• The proposed model can be used for classifying text to categories/domains.
• A set of 2800 Persian questions and a set of 1000 German questions from CQA websites have been annotated for our experiments.
• Our model was compared to the state-of-the-art classification algorithms.
摘要
•We introduce a new classification algorithm based on LDA topic models.•The model benefits from a topic-categories mixture model in addition to the original LDA.•The proposed model can be used for classifying text to categories/domains.•A set of 2800 Persian questions and a set of 1000 German questions from CQA websites have been annotated for our experiments.•Our model was compared to the state-of-the-art classification algorithms.
论文关键词：Community-based QA,Question classification,LDA
论文评审过程：Received 24 August 2015, Revised 21 August 2017, Accepted 4 January 2018, Available online 4 February 2018, Version of Record 4 February 2018.
论文官网地址：https://doi.org/10.1016/j.ipm.2018.01.001
