Localized user-driven topic discovery via boosted ensemble of nonnegative matrix factorization

Keyphrases

• User-centric 100%
• Non-negative Matrix Factorization 100%
• Topic Discovery 100%
• Topic Modeling 40%
• Ensemble Model 40%
• Residual Matrix 40%
• Document-oriented 20%
• Supervised Learning 20%
• Factor Matrix 20%
• Redundant Information 20%
• User Interaction 20%
• Weighting Scheme 20%
• Ensemble Methods 20%
• Keyword-based 20%
• Modeling Context 20%
• Local Weighting 20%
• Givens Matrix 20%
• Low-rank Factor 20%
• Gradient Boosting Model 20%
• Document Corpus 20%

Computer Science

• nonnegative matrix factorization 100%
• Topic Modeling 40%
• Supervised Learning 20%
• User Interaction 20%
• Ensemble Method 20%
• Gradient Boosting 20%