Setting up Communities

To create a new community type (or topic), navigate to DATA > Communities in the Setup space:

Click the ADD COMMUNITY button or the plus-icon to create a new community type.

Configuration

Name

Name of the community type. The name is also used in the Community Type KEE configuration to refer to a community type.

Choose a facet to map to

The facet, on which the creation of communities relies and the values are mapped to communities.

Fetch facet values

Select to fetch all values of the facet chosen under Choose a facet to map to and create a community for each value.

Add manually

Comma separated values (CSV format) can be added manually to create communities.

Example Input:

name,facet_value
Netflix,Netflix Inc.
Starbucks,Starbucks Corporation

Upload file

A pre-built dictionary of communities can be uploaded in CSV or XLS format.

Content labeling (Known Entity Extraction)

If selected, creates a Known Entity Extraction (KEE) that is automatically added to all pipeline workflows. The created KEE labels newly ingested data with the communities (values of the chosen facet). You can find and edit the KEE configuration under AI STUDIO > Known Entity Extraction. The KEE configuration has the option Enable community loader set and a Community Type defined according to the name given to the created community type.

If the communities get updated, for example, a new community is added, the KEE lookup database will be synchronized and updated automatically (see here).

If you want to define communities based on a KEE see this guide (instead of creating a KEE based on communities).

Generic labeling

If selected, the KEE is configured without the default company suffix list or ngram language model but requires exact matches (min_score is set to 1.0)

Company names: optimized model to detect company names

If selected, the KEE is configured using the default company suffix list and ngram language model. The min_score for matches is set to 0.6.