Processors
2. Processors — How to Configure Computation
Each interface contains a list of processors. A processor is a named unit of work with a registered type and its own parameters.
Each processor type has its own configuration parameters, documented on dedicated pages:
- Features Processors — image features and embedding processors
- Metrics Processors — completeness, representativeness, and diversity processors
- Gap Processors — domain gap processors
2.1 Common Processor Structure
<interface>:
processors:
- name: <processor_name> # unique name within this interface
type: <processor_type> # registered type (e.g., completeness, image_features)
columns: # which columns to process
input: ["col_a", "col_b"]
# ... processor-specific parameters ...
Every processor has:
name— human-readable identifier; appears in output filenames and result columnstype— selects the processor implementation from the plugin registrycolumns— which data columns to operate on (see below)- Processor-specific parameters — documented per processor below
2.2 Columns
The columns key tells a processor where to find the data it needs:
columns:
input: ["col_1", "col_2"] # columns to process (wildcards supported: "feature_*")
exclude: ["meta_*"] # columns to exclude from processing (wildcards supported)
rename: # rename computed output columns
- from: <metric_name>
to: <renamed_metric>
prefix: <prefix> # prefix added to computed column names
suffix: <suffix> # suffix added to computed column names
Computed column naming: <prefix><renamed_metric><suffix>