Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Bayesian Text Classification and Summarization via A Class-Specified Topic Model
Feifei Wang, Junni L. Zhang, Yichao Li, Ke Deng, Jun S. Liu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use CSTM instead of two-stage LDA+RF | topic-model classifier · 20 Newsgroups test, supervised | classification accuracy: 71.80% -> 85.65% | Improved |
| use CSTM instead of DiscLDA | topic-model classifier · 20 Newsgroups test, supervised | classification accuracy: 81.38% -> 85.65% | Improved |
| use CSTM instead of LDA per class | topic-model classifier · 20 Newsgroups test, supervised | classification accuracy: 33.27% -> 85.65% | Improved |
| use CSTM instead of two-stage LDA+RF | topic-model classifier · 20 Newsgroups test, labeled fraction 10% | classification accuracy: 64.78% -> 68.95% | Improved |
| use CSTM instead of two-stage LDA+RF | topic-model classifier · 20 Newsgroups unlabeled training docs, labeled fraction 10% | classification accuracy: 67.60% -> 64.81% | Hurt |
| use phrases (bigrams, trigrams) instead of words | modified CCS · 20 Newsgroups test, labeled fraction 10% | classification accuracy: 16.74% -> 37.75% | Improved |
| add class-specific topics to LDA (CSTM) with fewer topics | LDA · 20 Newsgroups test | log marginal likelihood: -2.103 × 107 -> -6.803 × 106 | Improved |
| use bigrams or trigrams instead of words as analysis units (ablation) | LDA and CSTM · 20 Newsgroups | classification accuracy | Hurt |
| vary prior alpha (0.1, 0.5, 1) and beta (ablation) | CSTM · 20 Newsgroups test | classification accuracy | No clear effect |
| increase labeled fraction phi | CSTM · 20 Newsgroups test | classification accuracy: 68.95% -> 85.01% | Improved |
Closest papers by meaning
- Diagnosing and Improving Topic Models by Analyzing Posterior Variability (AAAI 2018 · similarity 0.86)
- Online Bayesian Moment Matching for Topic Modeling with Unknown Number of Topics (NeurIPS 2016 · similarity 0.85)
- Knowledge-Aware Bayesian Deep Topic Model (NeurIPS 2022 · similarity 0.85)
- A Statistical Approach for Optimal Topic Model Identification (JMLR 2022 · similarity 0.85)
- Dirichlet belief networks for topic structure learning (NeurIPS 2018 · similarity 0.84)
- Jointly Modeling Topics and Intents with Global Order Structure (AAAI 2016 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
