VIVO Dashboard is a Drupal-based application for analyzing publication metadata coming from VIVO, a semantic researcher profile system. VIVO Dashboard allows end users (primarily administrators) to easily generate reports in one of the three way: bar graphs, HTML lists, and spreadsheets. End users can facet publication records in a variety of ways including:
- publication type
- journal ranking
- journal name
- author name
- author type
- organizational affiliation
- first-last author affiliation
What it looks like
- Weill Cornell Medicine: https://vivodashboard.weill.cornell.edu/
How it works
All VIVO sites expose their data as RDF. VIVO Dashboard routinely retrieves batches of RDF data from VIVO and turns it into Drupal content using Drupal's "Feeds" module.
VIVO's index pages contain links to RDF files listing all individuals in a particular class. VIVO Dashboard uses these RDF lists as feeds, similar to RSS feeds, periodically checking the lists to see if publications have been added or removed. Individuals linked to publications, such as authors and journals, are imported on demand.
Note: VIVO may limit these RDF lists to 30,000 items.
The import process is quite slow. Tens of thousands of publications will typically take between a few days and a week to import, depending on the performance of the VIVO site.
VIVO Dashboard saves imported data as Drupal content.
- Publications -> nodes
- Authorships -> relations
- Authors -> nodes
- Journals -> taxonomy terms
- Departments -> taxonomy terms
By storing imported data as Drupal content it becomes possible to leverage the thousands of available Drupal modules. VIVO Dashboard uses a number of popular modules to power its faceted search interface, including:
- Search API
- Facet API
The main interface included with the application is highly customizable. These modules allow for great flexibility, and administrators can adjust the interface to suit their institution's needs.
VIVO version compatibility
VIVO Dashboard has been updated to work with the new VIVO-ISF ontology structure introduced in VIVO 1.6. The import system has been upgraded with better caching and more configurability. The Ultimate Cron module has been added as a more reliable replacement to Elysia Cron.
To install a version of VIVO Dashboard compatible with VIVO 1.5 and below, simply use the distro-1.5.make file when building the codebase, as described the "Installation" section. Note that, after building the codebase, the README for that particular version of VIVO Dashboard can be found in DRUPAL_ROOT/profiles/vivodashboard.
VIVO Dashboard is made available as a Drupal install profile and does not actually include all the required packages. Before installing, the codebase must be "built" using Drush.
Building with Drush
Install Drush by following the instructions at: https://github.com/drush-ops/drush
For VIVO 1.6+ run:
drush make https://raw.githubusercontent.com/paulalbert1/vivodashboard/master/distro.make vivodashboard
For VIVO 1.5 and below run:
drush make https://raw.githubusercontent.com/paulalbert1/vivodashboard/master/distro-1.5.make vivodashboard
If no errors were reported, you should should have a complete VIVO Dashboard codebase inside the "vivodashboard" directory.
If you have your own hosting, or want to install VIVO Dashboard in your own local development environment, review the documentation on installing Drupal here : https://drupal.org/documentation/install (Step 1 is already done).
If you have don't have a preference for hosting, Pantheon is an excellent option. Development sites are free.
If you want to get up and running quickly on your local machine, Acquia Dev Desktop is a great self-contained LAMP environment for Windows and Mac, designed for Drupal.
Installing on Pantheon
- Create a new site on Pantheon and proceed through the Drupal installation. Info you enter does not matter at this point.
- Put the Pantheon site in SFTP mode.
- Inside the "vivo-dashboard" directory you built using Drush, locate the "profiles/vivodashboard" directory.
- Connect to the Pantheon FTP server and upload your "profiles/vivodashboard" to the corresponding "profiles" directory on Pantheon. This may be inside a "code" directory.
- In the Pantheon dashboard go to "Workflow" and then "Wipe". Confirm and wipe the environment.
- Click on "Visit Development Site".
- You should see a Drupal installation page again, with an option to install VIVO Dashboard. Choose that option and proceed through the installation. If there is no such option, you probably did not upload "profiles/vivodashboard" to the correct location.
- When the installation is finished you will be taken to VIVO Dashboard's main page.
Note: If you're comfortable using Git, steps 2-4 can be done with Pantheon's Git repository for the site. You'll need to add an SSH key in the Pantheon dashboard.
Installing with Acquia Dev Desktop
- Install the Acquia Dev Desktop application
- Open the Acquia Dev Desktop Control Panel, go to Settings, then Sites, then Import.
- For the "Site path" locate the "vivo-dashboard" directory you built using Drush.
- For the database select "Create new database".
- Enter anything you want for the domain.
- After clicking "Import" you should be brought to a Drupal installation page.
- Choose the "VIVO Dashboard" option and proceed through the installation.
- When the installation is finished you will be taken to VIVO Dashboard's main page.
Starting an Import
Once installed, the first thing to do is start the import.
- Visit the /import page, then choose the "VIVO Publications" importer.
- Enter your VIVO site's URL in the form.
- Enter the URI for the class you'd like to import (e.g. http://purl.org/ontology/bibo/AcademicArticle).
- Click "Import".
- In a new browser tab visit /import/log to confirm data is being retrieved.
Note: VIVO importers other than "VIVO Publications" can be ignored. These get triggered automatically.
If you see a status message reporting "There are no new nodes" your class URI is likely incorrect.
If you see an error message, your VIVO site URL might be incorrect or VIVO could be failing to produce the RDF list for the specified class. Some VIVO sites seem to have trouble with RDF lists containing a large number of individuals. You can test this by instead trying a class that contains a smaller number of individuals (hundreds instead of thousands). See the Troubleshooting section for more information.
If the form spends a minute or two loading after clicking "Import", and you get an error without any details, it's likely that PHP hit its time limit. That's fine, the import will pick up where it left off.
Most VIVO sites contain thousands of publications, and a complete publication import may take anywhere from a few days to over a week. Because PHP has limits on the execution time for a request, Drupal must import one small chunk of data at a time via cron runs.
Review the documentation on setting up cron for Drupal: https://drupal.org/cron
For VIVO Dashboard, the optimal cron frequency is 1 minute.
The ideal way to run cron is via Drush + Ultimate Cron. The Ultimate Cron module has a special drush command "cron-run" that will be the most efficient way to keep your import running.
Example crontab entry:
*/1 * * * * drush --root=/path/to/your/drupalroot --quiet cron-run
Note: If you need to install your crontab entry on a separate server (i.e. when using Pantheon for hosting), you will need to install SSH keys and Drush aliases on that server.
Example crontab entry on a separate server:
*/1 * * * * drush @pantheon.username.vivo-dashboard.dev --quiet cron-run
If unable to use Drush, a more typical crontab entry should still work:
*/1 * * * * wget -O - -q -t 1 http://mysite.gotpantheon.com/cron.php?cron_key=rUncI0PLGq7_56SrZzNCrnerLoHZOHb7pCoFSFhBwdE
Note: The cron_key is unique for each site. You can find this URL on the Status Report (Reports -> Status report).
Cron runs are used to keep Drupal working. The two main jobs that VIVO Dashboard performs during a cron run are:
- Importing data from VIVO
- Indexing imported content for search
VIVO Dashboard comes with a module called Ultimate Cron. This module acts as a manager responsible for delegating work. Ultimate Cron knows how frequently jobs should be performed and will trigger them accordingly.
Ultimate Cron can be configured at: Configuration -> System -> Cron Settings (/admin/config/system/cron)
Resting the publication import process on VIVODashboard
Following are the procedures to reset the import on VIVODashboard.
- Login with admin privilege.
- Unlock the import by first killing/unlocking the running cron jobs on feeds: (https://vivodashboard.weill.cornell.edu/admin/config/system/cron) Default cron handler Queue: feeds_push_unsubscribe Queue: feeds_source_clear Queue: feeds_source_expire Queue: feeds_source_import
- Go to import module and unlock the import process: (https://vivodashboard.weill.cornell.edu/import/vivo_publications) Click on the UNLOCK tab to open a confirmation page. Confirm unlock by clicking on Unlock button.
SCImago journal ranking data can be imported using the Journal Rankings importer (/import/journal_rankings), which expects a TSV file. Column 1 should contain the journal name, column 2 should contain the rank, and column 3 should contain the ISSN.
Once these are imported you'll need to do a bulk update to add ranking data to previously-imported journals (/admin/structure/taxonomy/journal_rankings/update), and run a search reindex (/admin/config/search/search_api/index/authorships and /admin/config/search/search_api/index/publications). Journals added after ranking data has been imported will automatically pick up rankings based on ISSN.
Chances are you will want to adjust the existing data mappings or add new data specific to your institution. All the mapping from VIVO to Drupal happens in the Feeds importer configurations under Structure -> Feeds importers, then under Processor -> Mapping.
- VIVO Publications: /admin/structure/feeds/vivo_publications/mapping
- VIVO Authors: /admin/structure/feeds/vivo_authors/mapping
- VIVO Journals: /admin/structure/feeds/vivo_journals/mapping
It should be fairly easy to add new data properties and map them to Drupal fields. Object properties (relationships), however, are a bit more complex and mapping for these has yet to be streamlined.
Hiding rdf:type items
You will often get not-so-useful superclasses, such as "Thing" or "Agent", appearing in facets and lists. To hide those, edit the corresponding taxonomy term (/admin/structure/taxonomy/rdf_types) and click the "Hidden" option at the bottom. You will need to reindex before the change takes effect.
Note: It's a good idea to clear caches after hiding/unhiding an rdf:type term. Do this before reindexing.
Including/excluding facet items
When logged in you can edit configuration for facets by hovering over them and clicking the gear icon. Select "Configure facet filters", where you can enable either "Include" or "Exclude" filters.
The first thing to try when troubleshooting Drupal is clearing caches. You can find "Flush all caches" under the Home icon in the toolbar.
Drupal maintains a comprehensive system log at: Reports -> Recent log messages (/admin/reports/dblog)
There is aggregated view of the Feeds import log at: /import/log
By default, feed importers are configured to add lots of debug information to the log. These entries have a severity of "debug". This can be turned off per-importer in the parser plugin settings.
Problems during a VIVO import, such as failed linked data requests, are usually logged with a severity of "warning" or "error".
Failed linked data requests
If you see warnings in the Feeds log indicating that certain VIVO URIs returned no data, the requests are likely timing out. This typically happens with VIVO individuals containing very large amounts of data. You can try increasing the timeout for the importer in the parser settings (e.g. /admin/structure/feeds/vivo_publications/settings/LdImportParser).
VIVO RDF list errors workaround
If you get error messages such as "500 Internal Server Error" after submitting the Import form, the VIVO site is likely failing to produce the RDF list for the class you specified. You can verify this by visiting the index page on the VIVO site, choosing the respective class, and trying the RDF link at the top.
One way to work around this is by manually creating a text file containing all the publication URIs to be imported, with one URI per line. This can be done using VIVO's internal SPARQL endpoint. You'll then need to change the Fetcher plugin for the importer (/admin/structure/feeds/vivo_publications/fetcher) to either "File upload" or "HTTP Fetcher". Then you can specify this text file on the import form (/import/vivo_publications).
Note: VIVO's SPARQL form produces a CSV with URIs wrapped in quotes. The quotes need to be removed for VIVO Dashboard.
Cron jobs not running
If the jobs at /admin/config/system/cron are not running and you are certain that a crontab entry has been correctly configured, it may be that Ultimate Cron has gotten stuck. Unfortunately, the best way to unstick it seems to be to manually truncating the following database tables: cache_ultimate_cron, ultimate_cron_lock
Drupal's great flexibility often comes at the expense of performance. As the amount of data increases, VIVO Dashboard will become slower. There are some things you can do to give Drupal a boost.
Linked data caching (enabled by default)
The Graphite library is now used for all RDF handling during imports. Graphite will optionally cache all linked data responses to the filesystem. This saves a tremendous number of HTTP requests while importing VIVO data, so this caching is enabled by default. The cache location is inside Drupal's configured temporary directory, in a "graphite-HASH" directory. Be aware this directory can grow quite large.
VIVO Dashboard comes with a module called Entity Cache that is intended to reduce the load on the database. This module is disabled by default only because it has not had enough testing. Enabling it is perfectly safe and should lead to noticeable performance gains. However, it should be suspect if pieces of data start disappearing.
The primary module behind VIVO Dashboard's faceted search is Search API. Search API relies on a "backend" module and clever indexing to do its magic. Out of the box the database backend (search_api_db) is used for VIVO Dashboard in order to minimize hosting requirements. While this is the most convenient backend, it's not the most performant.
Using Apache Solr as the backend for Search API will yield substantially better performance. Although it hasn't been fully tested, it should simply be a matter of changing the Search API indexes to use the Solr backend already included with VIVO Dashboard (/admin/config/search/search_api).
You also, of course, must set up the Apache Solr server itself. Once you do so, you'll need to copy configuration files from the search_api_solr Drupal module to Apache Solr's "conf" directory. Instructions and be found on the search_api_solr module page: https://drupal.org/project/search_api_solr
VIVO Dashboard is set up as a Drupal install profile, with custom functionality handled by the Features module.
The custom features bundled with VIVO dashboard can be found in: DRUPAL_ROOT/profiles/vivodashboard/modules/vivodashboard
Most files in these directories are generated by features and can be ignored.
Custom code can be found in:
- The .module file for the feature (e.g. vivodashboard_publications.module)
- Module-specific includes, autoloaded by other modules (e.g. vivodashboard_import.feeds.inc)
- The plugins directory for each feature (e.g. vivodashboard_core/plugins)
Capturing local changes
As a general rule, don't touch the code inside DRUPAL_ROOT/profiles/vivodashboard. If you need to add or upgrade a contrib module, place it in sites/all.
After overriding the VIVO Dashboard defaults, when customizing things for your institution, you may want to capture your changes in code for version control. Overwriting the VIVO Dashboard features directly would make future updates difficult. A more sustainable approach extract new features modules and include them in sites/all/modules.
Sometimes it may be necessary to duplicate the entire VIVO Dashboard feature to sites/all/modules in order to make your modifications. Unfortunately Drupal 7 doesn't offer a better option for managing configuration.
Search API publications and authorships
You will find there are two Search API indexes: publications and authorships. They have the same fields indexed and the same facets enabled. The authorships index exists to work around an unfortunate limitation in Search API's Views integration.
VIVO Dashboard's publication export pages are views listing individual authorships (as opposed to publications). Normally Views would be able to accomplish this using relationships, but Search API relies on Entity API for Views integration, which has very limited support for relationships. This drupal.org issue has comments from the Entity API and Search API maintainers regarding this limitation: https://drupal.org/node/1378656
For now, maintaining two separate, almost identical, Search API indexes is the only way to accomplish both the List and Export views. Consolidated indexing code can be found in search_api_alter_callback_publications.inc