Generate Taxonomies Automatically

The TaxonomyGenerate action allows you to generate a hierarchical taxonomy from one or more clusters or query results.

The taxonomy generator adapts the Bayesian and information theoretic methods to concept selection. It applies Bayesian algorithms to identify statistical relationships between concepts and sets of concepts (at the document and document set level). It then filters them to form the hierarchical structure of the final taxonomy.

You can write the taxonomy to disk as a directory structure, or import the taxonomy into the category hierarchy.

NOTE:

Before you create a taxonomy, ensure that your IDOL Content component index does not contain duplicate documents or text that is repeated in multiple documents (for example, document headers). Ensure that these are stripped out at the import stage to gain optimal results.

You can set up a schedule that runs the TaxonomyGenerate action at regular intervals.

Generate a Taxonomy from Clusters

Use the TaxonomyGenerate action with the SourceJobName and Cluster parameters to generate a taxonomy from one or more clusters. For example:

action=TaxonomyGenerate&SourceJobName=Taxonomy1&Cluster=0,1

In this example, the IDOL Category component generates a taxonomy from the Taxonomy1 cluster.

Generate a Taxonomy from Query Results

Use the TaxonomyGenerate action with the DREQuery parameter to generate a taxonomy from a query. For example:

action=TaxonomyGenerate&DREQuery=new+tax+cuts

In this example, the IDOL Category component generates a taxonomy from the results that it returns from its data index for the query new tax cuts.

Schedule Taxonomy Generation

You can set up a schedule to run the TaxonomyGenerate action at regular intervals.


_HP_HTML5_bannerTitle.htm