Recommender systems were originally defined as systems in which people provide recommendations as inputs, which the system then aggregates and directs to the appropriate recipients. The term now has a broader connotation, describing any system that outputs individualized recommendations or has the effect of guiding the user in a personalized way to interesting or useful objects within a wide space of possible options. Indeed, recommendation systems are defined as software agents that elicit the interests and preferences of users. They are considered as a specific form of information filtering to predict the interests or preferences that users would have attributed to an object or an element not considered before, which saves a lot of time.

Recommender systems can be seen as a response given to users having difficulty in making a decision within the framework of the use of a traditional information retrieval system. The user and the items are two basic elements in any recommender system. The user corresponds to the person using the system, while the item corresponds to what the system recommends to the user.  In other terms, a recommendation system aims to provide a user with relevant objects (items) according to his preferences. It is an information filtering system with the aim of predicting the positive appreciation of a user for an object (film, book, service, question, etc.) or a social element (person, community, etc.). It consists first in collecting information from a user, then in building a user model from this data and finally in deducing a list of recommendations for this user.

The rise of the Web and its popularity have notably contributed to the establishment of such systems as in the field of e-commerce, Cloud Computing, Collaborative Question Answering systems but several other fields are also concerned by recommendation systems as in Business Intelligence context.

Several factors are taken into account in order to categorize recommendation systems such as knowledge of user’s profiles and preferences and knowledge of the items to be recommended. From these factors, several types of recommendations are produced. The most commonly used types are content filtering and collaborative filtering. Several methods in Automatic Learning and Data Mining fields are used in recommendation systems.

Overall, the challenge of implementing recommendation systems requires a multidisciplinary approach, involving several technique and depending on the target domain as Artificial Intelligence, Learning techniques ,  Information Retrieval techniques, Data mining, etc. It requires continuous research, collaboration, and adaptation to stay ahead in several domains.