The Metanome project is a joint project between the Hasso-Plattner-Institut (HPI) and the Qatar Computing Research Institute (QCRI). Metanome provides a fresh view on data profiling by developing and integrating efficient algorithms into a common tool, expanding on the functionality of data profiling, and addressing performance and scalability issues for Big Data. A vision of the project appears in SIGMOD Record: "Data Profiling Revisited". Please find the the BibTex, EndNote, and ACM Ref citations for referencing the Metanome project in scientific works on the Metanome project website.
The Metanome tool is supplied under Apache License. You can use and extend the tool to develop your own profiling algorithms. The profiling algorithms, which we provide on our Algorithm releases page and the metanome-algorithms repository, have Apache copyright as well.
The Metanome platform itself is a backend service that communicates over an HTTP REST API with a user facing web frontend. This frontend is provided in a seperate Metanome Frontend repository. Building the Metanome tool, therefore, requires you to pull the frontend as a submodule (see below).
Building Metanome Locally
Metanome is a java maven project. So in order to build the sources, the following development tools are needed:
- Java JDK 1.8 or later
- Maven 3.1.0
Make sure that all three are on your system's PATH variable when running the build.
Pull Metanome Frontend Submodule
Before executing the build you have to clone the Metanome Frontend into the project.
git submodule init git submodule update
Metanome can be build by executing:
mvn clean install (or for a parallel build
mvn -T 1C clean install)
If the frontend build fails due to missing or incompatible Angular packages, it often helps to re-run the build.
When the built has finished, Metanome can be packaged together with a Tomcat webserver, some test data, and some test algorithms. To speedup builds this package is not created in the default maven profile. The deployment package can be created by executing the build with the deployment-local profile:
mvn verify -P deployment-local
or by executing package on the deployment project directly:
mvn -f deployment/pom.xml package
Note that if metanome has not been installed before creating the package (via mvn clean install), dependencies will be retrieved online, which can result in a deprecated package!
To start the Metanome frontend you then have to execute the following steps in the deployment folder:
- Go into the unzipped folder and start the run script, either
- Open a browser at http://localhost:8080/
All Metanome releases can be found on the Metanome releases page.
Developing a profiling algorithm for Metanome
If you want to build your own profiling algorithm for the Metanome tool, the best way to get started is our Skeleton Project. It contains an algorithm frame and a test runner project, with which you can run and test your code (without a running Metanome tool instance). For more details, check out the contained README.txt file.
Since many profiling algorithms use similar techniques for the discovery of dependendencies, its worth checking out the following resources as well:
- metanome-algorithms including many implementations of novel and popular profiling algorithms for various types of metadata.
- Metanome-Data-Structures including, for instance, position list indexes (PLIs), which many algorithms use for candidate validation; see also pli-benchmarks
The Metanome tool, information for algorithm developers and contributors to the project can be found in the github wiki.
Deploy Metanome Remote
It is possible to deploy Metanome using PaaS providers like (Amazon Beanstalk, Heroku or Google App Engine). We provide additional configs and documentation how to deploy Metanome on these in the github wiki.
The Metanome modules are continuously deployed to sonatype and can be used by adding the repository:
<repositories> <repository> <id>snapshots-repo</id> <url>https://oss.sonatype.org/content/repositories/snapshots</url> </repository> </repositories>
Git Commit Hooks
The project is using license-maintainer as Pre-Commit Git Hook to keep the license information in all Java, XML and Python files up to date. To use it you have to execute the
./add_hooks.sh shell script which is creating an pre-commit hook symlink to the license-maintainer script.
The project follows the google-styleguide please make sure that all contributions adhere to the correct format. Formatting settings for common ides can be found at: http://code.google.com/p/google-styleguide/
All files should contain the apache copyright header. The header can be found in the