Introduction to Recommender Systems

Introduction to Recommender Systems

Recommender Systems(RSs) are software tools that recommend items to users. They are used to personalize online stores for each customer, maybe with an exception to the top rated items. Recommender Systems tend to expect the most suitable items for a user, and recommends them. Recommender Systems plays an important role in websites like Amazon, YouTube, Netflix, Last.fm, IMDb, Yahoo, and TripAdvisor. The study of Recommender Systems is a relatively new research topic(1990s); the topic is being researched by ACM Recommender Systems(RecSys), institutions of higher education, and several academic journals.

The handbook is mainly divided into five sections: techniques that are used by RSs, applications and evaluations of RSs, interaction with RSs, RSs and communities, and Advanced algorithm topics.

The reasons why this technology should be used:

• Increase the number of items sold, selling more items beside the items sold without recommendations.
• Sell more diverse items, recommending items that are not easily reached without an accurate recommender system.
• Increase the user satisfaction, improving the recommender system can improved the user experience of the user with the application or the website.
• Increase the user fidelity, The longer the user interacts with the application or the website, the user model should be more accurate.
• Better understanding what the user wants
• Find Some Good Items, recommend items for the user based on the expected rating for the user on the items.
• Find all good items, recommend all the items that the user will need.
• Annotation in context, emphasize some of the items in a specific context category of items.
• Recommend a sequence, recommend a sequence of generations of an item.
• Recommend a bundle, recommend a group of items that fit altogether.
• Just browsing, during browsing recommend items that a user might be interested in.
• Find credible recommender, add features that the user can use to test the behavior of the recommender system.
• Improve the profile, This is related to the user, in case that the user can provide information about his likes or dislikes.
• Express self, help the users who are more interested in contributing to the application or the website by sharing their opinions and their beliefs rather than using the recommendations of the recommender system.
• Help others, help the users who want to evaluate items with their information.
• Influence others

Data Sources:

• Users are the users of the recommender system.
• Items in an abstract sense are the objects being recommended, the main characteristics of the items are complexity and value.
• Transactions are the log-data that keeps track of the interactions between the user and the recommender system, this data can be ratings.

Recommendation techniques:

• Content based, The system learns to recommend items similar to the previously liked items by the user, the similarity is calculated based on the attributes of the items.
• Collaborative filtering, The system recommends the items rated or used by other similar users; this is considered to be the most widely used recommender system.
• Demographic, The system recommends the items based on the demographic profile of the user, there are no proper research in the demographic recommender systems.
• Knowledge based, The system recommends items based on specific domain knowledge and how much the items match the user's needs, in other words, it recommends the useful items for the user; This system tends to show good results at the beginning of using them.
• Community based, The system recommends items to the user based on the preferences of his friends.
• Hybrid Recommender systems, The system is a based on a combination of the previously mentioned systems.