Integration of an externally running DataOntoSearch instance, providing dataset tagging and search
Extension for integrating CKAN with DataOntoSearch.
DataOntoSearch is a project which aims to make it easier to find datasets, by using a domain-specific ontology to find similar datasets. The software is run as a separate server, which other projects like CKAN can connect to.
There are two separate plugins provided with this extension. dataontosearch_tagging provides a way of associating datasets with concepts in the ontology. (Each such association is internally called a “tag”, which should not be confused with the traditional tags CKAN provide.) dataontosearch_searching provides an integrated way of searching using DataOntoSearch.
The extension adds a link you can follow when editing datasets. From there, you can change what concepts are connected to what datasets.
The extension also adds a link to the alternative search method. Following it lets you search using DataOntoSearch.
This plugin was developed for CKAN version 2.8. We have not checked what other versions it works with, but it does use features introduced in version 2.7.
To install ckanext-dataontosearch:
Ensure that the ckanext-dcat extension is installed.
Ensure that CKAN can accept multiple requests in parallel. For example, if you use gunicorn to run your application, you could use the -w flag to specify more than 1 worker: gunicorn -w 4 … (This is necessary because this extension’s request to DataOntoSearch might cause DataOntoSearch to make a request back to CKAN, so the applications would end up waiting for each other in a deadlock.) Note that the debug setting must be set to false for CKAN to work in parallel.
Activate your CKAN virtual environment, for example:
Install the ckanext-dataontosearch Python package into your virtual environment:
pip install ckanext-dataontosearch
Add dataontosearch_tagging and dataontosearch_searching to the ckan.plugins setting in your CKAN config file (by default the config file is located at /etc/ckan/default/production.ini). Both are not required, any one of them can be used alone, but that is rather uncommon. They need to be listed after the dcat plugins.
Add required settings:
# Base URL where dataset_tagger is running ckan.dataontosearch.tagger_url = https://example.com/tagger # Base URL where the search for DataOntoSearch is running ckan.dataontosearch.search_url = https://example.com/search # The DataOntoSearch Configuration to use ckan.dataontosearch.configuration = 5c7ea259c556bb42803fa17e
Restart CKAN. For example if you’ve deployed CKAN with Apache on Ubuntu:
sudo service apache2 reload
The required settings are described in the installation guide. In addition to those, you may specify the login used when connecting to DataOntoSearch:
# Username and password to use when querying and tagging datasets in # DataOntoSearch (HTTP Basic Authentication) # (optional, default: no credentials). ckanext.dataontosearch.username = aladdin ckanext.dataontosearch.password = opensesame
In addition, you can also tell the extension to use the autotagged similarity graph when searching, instead of the manual tags:
# Whether to use the autotagged graph instead of the manual one when # searching (optional, default: no). ckan.dataontosearch.use_autotag = yes
To install ckanext-dataontosearch for development, activate your CKAN virtualenv and do:
git clone https://github.com/tobinus/ckanext-dataontosearch.git cd ckanext-dataontosearch python setup.py develop pip install -r dev-requirements.txt
There are plenty of things that should be improved. Here are some of them:
Integrate concept viewing/editing with the dataset type of view, so the tabs don’t disappear once you click on “Concepts”.
Some styling improvements can be done to make it look more appealing and be easier to use.
Give feedback to the user when they save concept changes successfully.
Use progress indicator of some kind when the user submits concept changes, and stop them from submitting more than once.
Give the user an idea of how the concepts relate to one another in a hierarchy, instead of just a flat list. They should only use the most relevant, specific concepts, and not try to fit many “similar” concepts, like you would with tags or search words.
Give the user more context for each concept. There exist alternate labels that sometimes indicate what other areas that concept is covering, and some even have text that explain and show how to apply that concept. This would require changes to the dataset_tagger API in DataOntoSearch to make the information available to ckanext-dataontosearch.
Separate the two different plugins into two different Python files, per the CKAN recommendations (to avoid problems with files loading out of order).
There are also some TODO notes in the source code.
Running the Tests
To run the tests, do:
nosetests --nologcapture --with-pylons=test.ini
To run the tests and produce a coverage report, first make sure you have coverage installed in your virtualenv (pip install coverage) then run:
nosetests --nologcapture --with-pylons=test.ini --with-coverage --cover-package=ckanext.dataontosearch --cover-inclusive --cover-erase --cover-tests
Releasing a New Version of ckanext-dataontosearch
ckanext-dataontosearch is availabe on PyPI as https://pypi.python.org/pypi/ckanext-dataontosearch. To publish a new version to PyPI follow these steps:
Create a source distribution of the new version:
python setup.py sdist
Upload the source distribution to PyPI (assuming you have run pip install twine before):
twine upload dist/*
Tag the new release of the project on GitHub with the version number from the setup.py file. For example if the version number in setup.py is 0.0.2 then do:
git tag 0.0.2 git push --tags
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Hashes for ckanext-dataontosearch-0.0.6.tar.gz