The World Wide Web provides a competitive environment for businesses, but consumers are easily distracted & if they don’t find what they want quickly, they may very well abandon ship & jump to a competitor’s Site.
One of the best ways to hold a client’s attention is to make recommendations to them that they might like while they are on your Site. However, this is easier said than done; it can get complicated. But, fret not. There are engines available which offer to make this objective possible & effective. Although many of these services are costly, closed source projects, but there are a few open source Software As A Service (SaaS) candidates, so we’ll take a look at such open source recommendation engines.
Company: ActionML
This one takes user actions, profiles, context, & item metadata into account to improve the predictive quality of its recommendations. The system uses the modern Correlated Cross-Occurrence Algorithm for recommendations, & is available on a public repository on GitHub for download. The system can be used on multiple servers to create various services, such as book, article & product recommendations.
Company: Apache
An open source machine learning server with customizable templates to quickly deploy a production-ready engine as a Web service. PredictionIO can be installed with Apache Spark, MLlib, HBase, Akka HTTP & Elasticsearch to simply scalable ML. It simplifies data management & integrates seamlessly with custom recommendation systems.
3. DLRM
Company: Facebook
The Deep Learning Recommendation Model is Facebook’s open source project, & is used in their own applications. Implemented using the open source PyTorch & Caffe2 platforms, DLRM, which is written in Python, can be used to integrate with public data sets to make predictive recommendations.
4. HapiGER
Company: HapiGER
HapiGER promises an increase in user engagement & sales via its node.js based engine. The engine uses PostgreSQL & RethinkDB event stores for persistence & scaling. The project is open source & is available for download from GitHub. The system is easy to install & the developers promise that its easy to use. HapiGER uses the the Good Enough Recommendations (GER), a scalable, simple recommendations engine.
5. RecDB
Developer: Mohamed Sarwat
An extension to the popular open source PostgreSQL database, which is available for installation on most hosting services, RecDB is easy to use since it integrates directly with common developer Stacks. It’s designed to work on UNIX operating systems & provides inbuilt queries to create predictive recommendations. Installation specifications are available on GitHub.
Image by Gerd Altmann from Pixabay
