[SPARK-9888] Update LDA User Guide

• Type: Documentation
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 1.5.0
• Component/s: MLlib
• Target Version/s: 1.5.0
• Assignee: Feynman Liang
• Reporter: Feynman Liang
• Shepherd: Joseph K. Bradley
• Created: 12/Aug/15 20:16
• Updated: 26/Aug/15 00:39
• Resolved: 26/Aug/15 00:39

Description

LDA has received numerous updates in 1.5, including:

• OnlineLDAOptimizer:
• Asymmetric document-topic priors
• Document-topic hyperparameter optimization
• LocalLDAModel
• predict
• logPerplexity / logLikelihood
• DistributedLDAModel:
• topDocumentsPerTopic
• topTopicsPerDoc
• Save/load

It is important to note that OnlineLDAOptimizer=>LocalLDAModel and EMLDAOptimizer=>DistributedLDAModel now support different features. The user guide should document these differences.

Issue Links

• is related to: SPARK-5989 Model import/export for LDAModel
• is related to: SPARK-5572 LDA improvement listing
• is required by: SPARK-9668 ML 1.5 QA: Docs: Check for new APIs
• links to: [Github] Pull Request #8254 (feynmanliang)