The following is a web service that interacts with the OpenRefine Reconciliation Service API to reconcile names from the Library of Congress Name Authority File (LCNAF) and subjects from the Library of Congress Subject Headings (LCSH).
This service attempts to fetch names and subjects from the Library of Congress using the following methods sequentially:
- suggest - Query the Library of Congress using their suggest API, which returns a JSON list of the best suggestion(s).
- didyoumean - Query the Library of Congress using their didyoumean API, which returns an XML list of close matches to your query
- web scraping - As a last resort, the service will perform web scraping on the first page of results for a search query of the name or subject.
The reconciliation score, which indicates how good the match is, is determined using the Python difflib library.
- Ensure Python 3 is installed. This program was developed with Python 3.4.3.
- Download this repository locally (
git cloneor.zip) - Navigate to your local copy of the program in the command line interface
- Install the program requirements by typing
python -m pip install -r requirements.txt - Start the program by typing
python LoCreconcile.py(or run in IDLE or another IDE)
- Click the arrow in the title column of the column of names and/or subjects you wish to reconcile.
- Click
Reconcile > Start reconciling... - Click the
Add Standard Service...button in the bottom left of the reconciliation menu - Under
Enter the service's URL, enter the URLhttp://127.0.0.1:5000/reconcile/LoC - Note that after the service is added once per the previous steps, you will simply be able to select "LC Reconciliation Service" from the reconciliation menu in the future.
- In the following menu,
Namesreconciles from LCNAF,Subjectsreconciles from LCSH, andLoCreconciles from both. - Having
Auto-match candidates with high confidenceselected will automatically reconcile perfect matches - If you do not quantify
Maximum number of candidates to return, the program will attempt to return up to 3 candidate matches for each name/subject. - Click
Start Reconciling
The results of reconciliation will be links to URIs of the best matching names and subjects the service could find.
Example: http://id.loc.gov/authorities/names/n85243950
One of the best ways to expedite the reconciliation process is to start by exploring names which were near-perfect matches, having
reconciliation scores of .80+ first, using the best candidate's score facet, continuing to decrement the score range until the matches no longer seem
correct. Consult the OpenRefine wiki pages on reconciliation
and the Reconciliation Service API for more information. You can also search the web for guides,
such as this one.
Other reconciliation services:
- Christina Harlow's lc-reconcile
- codeforkjeff's conciliator (for LC identifiers via VIAF, plus much more)
- A list of services from the OpenRefine Wiki.