Affiliation augmentation for augment_pipeline. The pipeline relies on a set of predefined affiliation strings and affiliation data to assign canonical affiliations to affiliation metadata; it uses the results of both human- curated affiliation data and those derived from machine learning.

There are two queues:

• augment-affiliation: take a record's affiliation and (try to) augment it

• output-record: send the results of augmentation to master-pipeline via msg

The pickle file contains two dicts: affil_dict, the dictionary of curated affiliation strings and their IDs; and canon_dict, the dictionary of canonical IDs, abbreviations, and parent-child relationships.

To generate the pickle file from the command line, run the pipeline with:

run.py -mp

assuming the file paths and names in config.py are the locations for all of these files.

To send a single record through the augmentation pipeline to master, run the pipeline with:

run.py -d

This assumes that config.PICKLE_FILE already exists. If the pipeline is successful, the record in master_pipeline.records for 2002ApJ...576..963T should have been augmented with the affiliation A00928 (Yale University Astronomy Department) for all three authors.

To augment a set of records in file FILEPATH, run the pipeline with:

run.py -f FILEPATH

Matthew Templeton, ADS