LensKit is an implementation of collaborative filtering algorithms and a set of tools for benchmarking them. This readme is about working with the LensKit source code. For more information about LensKit and its documentation, visit the web site or wiki. You can also find information on the wiki about how to use LensKit without downloading the source code. If this is your first time working with LensKit we recommend checking out the Getting Started guide.
LensKit is made available under the GNU Lesser General Public License (LGPL), version 2.1 or later.
LensKit is built and deployed via Maven. To install it, check out
this repository and run mvn install
; it is then available to other
projects by depending directly on it as above (consult pom.xml
for
the version to use; all LensKit modules share the same version). The
source code can also be checked out and used in most Java IDEs.
NetBeans and IntelliJ both include Maven support, LensKit should
import fine with no special tricks.
LensKit is comprised of several modules. The top-level lenskit
module serves as a container to build them and provide common settings
and dependencies. The other modules are as follows:
lenskit-api
-- the common, public recommender API exposed by LensKit, independent of its actual implementations.lenskit-test
-- infrastructure and helper code for testing.lenskit-parent
-- infrastructure module that is the parent of most modules other thanlenskit-test
. This is mostly to provide a common build and test environment for the rest of LensKit. If you create a new module, make this module its parent.lenskit-data-structures
-- common data structures used by LensKit. These are split from-core
so the API can depend on them.lenskit-core
-- the core support code and configuration facilities for the rest of LensKit. It is the entry point for most of what you want to do with LensKit, providing support for configuring and building recommenders.lenskit-knn
-- k-NN recommenders (user-user and item-item collaborative filtering).lenskit-svd
-- the FunkSVD recommender (and eventually real SVD recommenders).lenskit-slopeone
-- Slope-One recommenders.lenskit-eval
-- the evaluation framework and APIs, along with a command line evaluation runner.lenskit-eval-maven-plugin
-- a Maven plugin for running LensKit algorithm evaluations and experiments.lenskit-package
-- a metapackage for preparing binary distributions, including scripts for running the evaluator.lenskit-archetype-fancy-analysis
andlenskit-archetype-simple-analysis
-- archetypes for creating user projects using LensKit.
We welcome contribution to LensKit. If you are looking for something to work on, we recommend perusing the open tickets on our Trac or asking on the mailing list.
We prefer to receive code submissions as GitHub pull requests. To do this:
- Fork the LensKit repository (
grouplens/lenskit
) on GitHub - Push your changes to your fork
- Submit a pull request via the GitHub web interface
When submitting a pull request via GitHub, you warrant that you either own the code or have appropriate authority to submit it, and license your changes under LensKit's copyright license (LGPLv2.1+).
These archetypes are to build simple
(lenskit-archetype-simple-analysis
) or more sophisticated
(lenskit-archetype-fancy-analysis
) projects for doing analysis of
recommender algorithms. Detailed information about using these
archetypes are in the src/main/resources/archetype-resources/Readme.md
file, which are installed in the top-level directory when a user uses
this archetype to generate a project.
- Be careful editing the pom in src/main/resources/archetype-resources,
because its variables are substituted
at two different times. Variables like
${project.version} are substituted at the time the archetype is run by a user to create a project. Variables with backslashes in front of them like $ {version} are left as variables by the archetype, so they can be substituted at project build time.