conciliator is a growing collection of OpenRefine reconciliation services, as well as a Java framework for creating them. A reconciliation service tries to match variant text (usually names of things) to standard IDs for the entity represented by that text.
This project supercedes refine_viaf.
Table of Contents
- Public Server
- General Features
- Data Source Features
- Running Conciliator on Your Own Computer
- Docker Image
- Configuring OpenRefine
- Creating Your Own Data Source
- Advanced Usage
- Do you use this thing??
If your needs are low and you can't or don't want to run this software yourself, you can use the public server at http://refine.codefork.com/. Visit that address for more instructions.
Out of the box support for the following data sources:
Good performance (uses threads; stable memory usage; caches results)
Super easy to run (works on Linux, Mac, Windows)
Data Source Features
Support for the following types of names provided by VIAF: Corporate Names, Geographic Names, Personal Names, Works, Expressions
"Proxy mode" to retrieve IDs used by source institutions, instead of VIAF IDs. (NOTE: hyperlinks to source record pages in OpenRefine are supported for BNE, DNB, ICCU, JPG, LC, NDL, SELIBR, SUDOC, and WKP. Links are BROKEN for BNC, BNF, DBC, and NUKAT. For all other sources, the links will take you to the VIAF page.)
Uses the ORCID v2.1 API. The detailed search results of the v1.2 API are no longer supported, so n+1 requests are made to fetch name details, which is gross, but it's the best we can do. Heavy use may cause the ORCID API to start returning rate-limiting errors.
Properties are supported as a way to do fielded searches using Solr syntax. For lists of valid field names to use in the "As Property" box, see the section titled "Fielded search for specific elements" on this page, and the list of identifier fields on the Supported Work Identifiers page.
For example, if you have a column containing Scopus EIDs, you can select the "Include?" checkbox for it and enter "eid" in the "As Property" box on the reconciliation screen.
By default, queries are keyword searches on the entire ORCID bios, which can return odd results sometimes. The "smartnames" mode (see the instructions below) splits up names and searches on the given-names and family-name fields specifically; if there are no results, it falls back to a keyword search.
Open Library has rate limits on its API, so requests are not run in a threadpool. Expect it to be slow.
Support for including additional columns (useful for specifying author(s), for example, to help narrow down searches for common book titles). If no results are found, the code tries again with only the original column.
- Any Apache Solr collection can be used as a data source. See the
sample commented-out lines in the
conciliator.propertiesfile for more details.
Running Conciliator on Your Own Computer
Install Java 1.8 or greater if you don't already have it.
Download the .jar file for the latest release. Alternatively, you can download the source code tarball or clone this repository, and build the .jar file using maven.
Run this command:
# replace VERSION with the release you downloaded java -jar conciliator-VERSION.jar
That's it! You should see some messages as the application starts up. Now you're ready to configure OpenRefine to use the service. When you're done with it, hit Ctrl-C to quit the application.
If a file named
conciliator.properties exists in the current
directory, conciliator will use the options found in it. See the
sample file in this repository.
By default, conciliator will run on port 8080, which is used in the
example URLs below. To use a different port, set the
property as follows when running the program:
java -Dserver.port=7000 -jar conciliator-VERSION.jar
A docker image created by tobinski is available here:
In OpenRefine, chose a column of names you want to reconcile, and select "Reconcile" and "Start Reconciling..." in the column pull-down menu.
Click "Add Standard Service..."
Enter a URL based on the data source you wish to use.
To reconcile against names from any VIAF source, type in:
To reconcile against a specific VIAF source, append its code to the end of the path. For example, to search only names from the Bibliothèque nationale de France, type in:
To retrieve the IDs used by source institutions, rather than VIAF IDs, use "proxy mode." For example, to search only names from the Library of Congress and retrieve their IDs, type in:
To use ORCID:
To use ORCID with "smartnames" mode when reconciliing names:
To use Open Library: (On the reconciliation screen, under the "Also use relevant details from other columns" panel, you can check the "Include?" box for columns to include in the query. Give them any name in the "As Property" box. If no results are found with these column values added to the query, the service will try again with only the original selected column.)
Follow the instructions on the dialog box to start reconciling names.
Creating Your Own Data Source
Clone this repository to get the source code. The code you create in the next steps should live under a new
com.codefork.refine.NEW_SOURCEpackage so that Spring's auto-scanning picks it up.
Create a class for your data source that extends
DataSourcefor very bare-bones functionality, or
WebServiceDataSourceif you are making requests to another web service. See the other data sources for some template code. Implement the abstract methods as required.
Create a controller that autowires your new DataSource and hooks up a unique path, e.g.
/reconcile/new_source. See VIAFController for an example.
Write a test or two if you like.
Set some default properties in
Configif your data source has any settings you want to be configurable.
Build a new .jar by running
mvn clean package. Run the .jar file as in the instructions above, and you should be able to access the service for your new data source at:
To build from the source code, install maven and type:
If you want to host this software on a server for long-term usage or
if you want to enable logging for debugging purposes, take a look at
run.sh for some helpful options.
You can change run-time options by editing the
- A few aspects of the Reconciliation Service API aren't implemented by this framework yet.
- Use dependency injection instead of singleton for the threadpool shared by all VIAF instances. Might need to rework how data sources get instantiated on-the-fly in ReconcileController.
Specification for the Reconciliation Service API:
This code drew inspiration from these other projects:
Do you use this thing??
Apparently, you do. Here's a bibliography of things that reference conciliator:
If you use conciliator, please take a few seconds to leave a comment on this page. Hearing from users really motivates me to continue improving this project.
This code is distributed under a GNU General Public License. See the file LICENSE for details.