Using Jupyter from KNIME to embed documents

This workflow demonstrates the use of functionality defined in a Jupyter notebook from inside of KNIME. After building a topic model for a set of documents, we use a Python function from a Jupyter notebook to perform t-SNE embedding of the documents into a 2D space. As a final step we do an interactive visualization to allow exploring the results.
The execution of this workflow requires the Textprocessing and the KNIME Python Scripting extensions. The executing Python environment must be provided with scikit-learn and jupyter installations.

External resources

• t-SNE Wikipedia
• KNIME Python Integration Installation Guide
• Fun With Tags
• KNIME and Jupyter