Using topic-noise models to generate domain-specific topics across data sources.

We can see a significant trend toward higher \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\phi $$\end{document} φ -values producing significantly higher coherence scores, while low values produce lower coherence and diversity.

also in 9,775 other papers