During the past few years, with the explosion of information being served as online media content, eCommerce, and streaming platforms, finding and accessing desired information in an optimal way has become crucial for many companies and research centers. With multiple platform players constantly competing for users’ attention, the actionable information provided by recommendation systems are one of Internet’s most popular features. They are also pragmatically useful in attracting users to a company’s web offerings.

Recommendation systems (also referred as recommendation engines) are critical in some industries, for example in Video on Demand (VoD) or eCommerce platforms, where they can generate a significant amount of income and help companies differentiate from competitors. Further, recommendation systems have the potential to change the way websites and applications communicate with users —they are essentially a solution that offers relevant and effective information in the wrapper of a personalized user experience. This is vital in the digital era, where users may be overwhelmed by the flood of information they receive from digital media and other sources. Therefore, providing the pertinent information they may need at each specific moment is key to standing above the crowd.

The primary goals of recommendation systems are to encourage demand as well as to actively attract users. Since one of the main aspects of recommendation systems is proactivity, they will suggest information to the user that may be of interest without the user having to look for it explicitly and in accordance with his or her predilections, which may be unintentionally limiting.  These recommendations are based on users’ past activity, attributes, contextual information and in similar users based on their buying histories, demographic data, and other attributes.

Most Recommendation systems can be classified in three different groups: Content-based, Collaborative Filtering and Hybrids.

  • Content-based systems These are built on the knowledge the system has about the items already selected or considered (looked at by the user), in order to provide suggestions about other items similar to those the user examined. This type of recommendation system focuses on the items themselves, relying on their characteristics and recommending other items with similar attributes or metainformation. Content-based recommendations are very powerful since minimum product information is needed for them to operate.
  • Collaborative filtering systems Recommendation systems based on collaborative filtering are based on the detection of users that are similar to the active user (in order to recommend items already rated by those similar users but not by the active user). In other words, they answer the query, “Show me information that other people similar to me have been interested in.”  This approach assumes that users with similar profiles will tend to perform comparable actions on a set of same (or nearly identical) items. These systems use the information gathered from past user interactions with the system (either implicitly or explicitly, for example, visit or view details) so as to suggest items which will presumably be of interest for the user. The main advantage of this approach is that it does not require any information about users or items in advance, making it appropriate in many situations.
  • Hybrid recommendation systems are essentially combinations of the previous two approaches. For example, they can use collaborative filtering and content-based information together to suggest a broader range of items accurately. The complexity of these hybrid systems tends to make them more expensive for the content producer to implement.

We observe that, in an attempt to enrich the user experience of site visitors, more and more companies (and individual business units) across many different business areas — often in the R&D domain where collaboration between users, along with rapid access to relevant information — are starting to implement recommendation strategies. These companies recognize the potential of providing good suggestions and are making the investment necessary to bring that benefit to customers.


Author: Antonio Perez

Antonio Perez is a Technical Architect and Lead at Copyright Clearance Center, with interest in everything related to Technology and Software, particularly Web, Search, NLP and Machine Learning. Fan of all the new technologies, he is always eager to learn and grow both personally and professionally. He has participated and delivered projects for important organizations and customers around the world. Graduating with a degree in Computer Engineering, Perez also holds a Master Degree in Engineering and Software Technology from the University of Seville and he is committer and collaborator in the Apache Software Foundation. Apart from technology, he also likes playing sports, watching his favorite TV series and playing video games.
Don't Miss a Post

Subscribe to the award-winning
Velocity of Content blog