Topic Modeling with STM: Optimizing K via a Verified Component

The Topic Extractor (STM) verified component implements a score for semantic coherence, exclusivity and other scores.
In the settings of the component an automated search can be activated for an optimal K.
If activated the component will offer an interactive view (right click > interactive view) after execution.
The computed scores are also going to be offered at its last output.

Usually it is a tradeoff between those metrics. Read more on the Verified Components descriptions.

External resources

• An Introduction to the Structural Topic Model (STM)
• stm: R Package for Structural Topic Models - Roberts, Stewart and Tingley, Journal of Statistical Software (2019)
• R Installation Guide - KNIME Docs
• Miniconda Download and Installation - Conda Docs
• Topic Assigner (STM) - KNIME Community Hub
• Topic Extractor (STM) - KNIME Community Hub
• Verified Component project - knime.com
• Verified Components project - knime.com