Research map / Probabilistic and scientific modeling / Variational and generative sampling
TopicEq: A Joint Topic and Mathematical Equation Model for Scientific Texts
Michihiro Yasunaga, John D. Lafferty
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 |
|---|---|---|---|
| jointly train the context topic model with an equation LSTM | TopicEq topic model · ContextEq-400K test, 50 topics | topic coherence (NPMI): .085 -> .097; +0.01 | Improved |
| jointly train the context topic model with an equation LSTM | TopicEq topic model · ContextEq-400K test, 100 topics | topic coherence (NPMI): .084 -> .094 | Improved |
| add equations as bags of tokens instead of omitting equation input (ablation) | TopicEq topic model · ContextEq-400K test, 50 topics | topic coherence (NPMI): .085 -> .087; +0.002 | Improved |
| replace equation LSTM with equation bag-of-tokens model (ablation) | TopicEq topic model · ContextEq-400K test, 50 topics | topic coherence (NPMI): .097 -> .087 | Hurt |
| shuffle equation token order (ablation) | TopicEq topic model · ContextEq-400K test, 50 topics | topic coherence (NPMI): .097 -> .086 | Hurt |
| condition equation LSTM on topic information | Equation generation · ContextEq-400K test, 100 topics | equation perplexity: 5.81 -> 5.34 | Improved |
| condition equation LSTM on topic information | Equation generation · ContextEq-400K test, 100 topics | syntax error rate: 15.3 -> 11.7 | Improved |
| embed the topic vector inside the LSTM cell rather than adding it at the output | Topic-dependent equation LSTM · ContextEq-400K test, 100 topics | equation perplexity: 5.41 -> 5.34 | Improved |
| embed the topic vector inside the LSTM cell rather than adding it at the output | Topic-dependent equation LSTM · ContextEq-400K test, 100 topics | syntax error rate: 12.5 -> 11.7 | Improved |
| make symbol-to-word alignment depend on topic | Math-symbol phrase prediction · ContextEq-400K test, 50 topics | phrase prediction perplexity: 602 -> 406 | Improved |
| make symbol-to-word alignment depend on topic | Math-symbol phrase prediction · ContextEq-400K test, 100 topics | phrase prediction perplexity: 602 -> 387 | Improved |
| add joint topic-aware alignment learning | TopicEq topic model · ContextEq-400K test, 50 topics | topic coherence (NPMI): .085 -> .088 | Improved |
Closest papers by meaning
- OpenWebMath: An Open Dataset of High-Quality Mathematical Web Text (ICLR 2024 · similarity 0.84)
- LLM-SRBench: A New Benchmark for Scientific Equation Discovery with Large Language Models (ICML 2025 · similarity 0.83)
- A Compact Model for Mathematics Problem Representations Distilled from BERT (AAAI 2025 · similarity 0.83)
- Topic-VQ-VAE: Leveraging Latent Codebooks for Flexible Topic-Guided Document Generation (AAAI 2024 · similarity 0.83)
- Exploring Equation as a Better Intermediate Meaning Representation for Numerical Reasoning of Large Language Models (AAAI 2024 · similarity 0.83)
- On the Affinity, Rationality, and Diversity of Hierarchical Topic Modeling (AAAI 2024 · similarity 0.83)
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.
