Batch Workflow and Output Naming

Prepare the batch

Put source .set/.fdt pairs in one input directory and choose a separate output directory. Step 1 rejects mixed source directories. Use Browse Directory for a clean, reproducible file set.

Validate the text-IA configuration and label queue on one representative dataset first. Then rerun Step 1 with multiple files, load the saved Step 2 configuration, inspect the first mapped dataset, optionally import IA columns, and load the saved label queue in Step 4.

Processing behavior

EyeSort handles files one at a time. Step 2 stores temporary text-IA datasets; Step 4 applies every queued label in memory, saves one final dataset, and cleans temporary files.

Final output names are normalized to:

<base>_processed.set
<base>_processed.fdt

A batch summary uses the next free name eyesort_labeling_summary_NNN.csv and includes dataset rows plus TOTAL rows.

Validate batch consistency

Compare assignment diagnostics, matched-label counts, and missing-key lists across participants. A successful batch completion can still contain skipped or failed datasets; read MATLAB warnings and count actual output files.

Export a processing script for a reviewable record of paths, configs, conflict policy, summary behavior, and BDF generation.