Have you ever wondered how the Spotify app on your mobile phones knows what kind of music you like or how Amazon knows what you would like to purchase? It’s all because of the recommender system that has kept on your activities and finally got to know about your choices.
A recommender system or a recommendation engine is a subclass of information filtering system tries to make predictions on user preferences and also make recommendations which should interest customers. In this technology-driven era, the use of recommendation is significantly wide.
Today, the popularity of these systems has reached such a level that there are companies that are providing open-source software as service recommender systems. One of the major benefits of using an open-source SaaS Recommender system is that you can make any modification. Also, you don’t have to put it a lot of capital to build one in-house.
In this article, we are going to have a look at five popular open-source SaaS recommender systems.
Built on top of Node.js and Redis, Raccoon is a collaborative recommendation engine and NPM module. Talking about how the engine works, it makes use of the Jaccard coefficient to know the similarity between users and k-nearest-neighbours to create recommendations.
This engine is built in such a way that any individual or business with users, a store of products/items etc. can use this open source recommendation engine. Raccoon takes care of all the recommendation and rating logic. That is not all, it can also be integrated with any database as it does not keep track of any user/item information besides a unique ID.
Written in Java, easyrec is a free and open source web application that provides personalized recommendations using RESTful Web Services and can be integrated into any web-enabled applications. How do easyrec works? Using the REST API, user actions such as viewing, buying or rating an item are sent to the easyrec and are stored in the database of the engine. Then, the analyzer periodically analyzes all recorded data and identifies patterns to generate recommendations.
LensKit is a Java-based research recommender system. Its successor, LensKit for Python — also known as LKPY, a set of Python tools for experimenting with and studying recommender systems. It also provides support for training, running, and evaluating recommender algorithms. LensKit can be used for research recommender algorithms, evaluation techniques, or user experience, and also to build the next recommender application.
Not exactly a recommender system itself, Crab is a python framework that is used to build a recommender system. It can be integrated with Python packages such as NumPy, SciPy, matplotlib etc. The main focus of the framework is to provide a way to build customised recommender system from a set of algorithms. Also, talking about algorithms, Crab provides two recommender algorithms: User-Based Filtering and Item-Based Filtering.
Apache PredictionIO is an open source Machine Learning Server built on technologies like Apache Spark, Apache HBase and Spray. The prime use of this state-of-the-art open source stack is for developers and data scientists to create predictive engines, which we also call as a recommender system for any machine learning task. Talking about installing PredictionIO, it can be installed as a full machine learning stack, bundled with Apache Spark, MLlib, HBase, Akka HTTP and Elasticsearch.
PredictionIO is fast and engines can be deployed as a web service during production. Also, it is open source that gives you the privilege to take a look at the code and know how it works.
Recommender systems or recommendation engine over the years have become an integral part of almost every online platform — be it a store, a streaming platform etc. However, when it comes to building an engine in-house, it is not an easy task. And this where SaaS recommender system comes into the play. From closed-source to open-source, SaaS recommender systems are becoming popular, as it not only save you a significant amount of money but also a lot of time.