Phase of Information Gathering:
This gathers pertinent user data, such as attributes, habits, or the substance of the resources a user visits, in order to create an user account or model for prediction tasks. A well-built user profile/model is a prerequisite for an accurate recommendation agent. To make reasonable recommendations straight away, the system has to know as much as it can about the individual. Recommender systems rely on a variety of inputs, including the most practical, high specific feedback, which risk severe user input about their interest in the item, or implicit feedback, which is derived by implicitly inferring user preferences by observing user behaviour. The combination of implicit and explicit feedback can also result in hybrid feedback. A user profile on an e-learning platform is a selection of private data connected to a particular user. Cognitive aptitudes, intellectual prowess, learning preferences, interests, and system interaction are all included in this data. The information required to create a picture of the user is often retrieved from the user profile. A user profile thus describes a straightforward user model. Any recommendation system’s capacity to accurately reflect users’ current interests is crucial to its effectiveness. For any prediction technique to produce useful and accurate recommendations, reliable models are essential.
- Explicit feedback: In order to build and enhance his model, the system typically asks the user through all the system display to offer ratings for products. The quantity of user-provided ratings determines how accurate the recommendation is. This method’s only drawback is that consumers must put forth some effort and aren’t always willing to provide sufficient data. Since it does not involve deriving preferences from actions, explicit feedback is still perceived as providing more reliable data. It also offers transparency into the recommender system, which leads to a marginally higher perceptions suggestion quality and more self belief in the recommendations.
- Implicit feedback: The system automatically deduces the user’s preferences by keeping track of the user’s various behaviours, including previous purchases, navigational history, and hours invested on specific web pages, links that the user followed, email content, and button clicks, among other things. By inferring the user’s preferences from their interaction with the system, implicit feedback lessens the burden on the user. However, the method is less precise and does not involve any human effort. Additionally, it has been suggested that implicit choice data may actually be more objective since there is no self-image or image maintenance bias caused by users answering in a socially acceptable manner.
- Hybrid feedback: In a hybrid system, the advantages of implicit and explicit input can be merged to reduce their shortcomings and produce the highest-performing system possible. This can be done by using implicit data as both a control on explicit evaluation or by limiting the user’s ability to provide explicit input to situations in which he explicitly expresses interest.
Phase of Learning: The feedback obtained during the information gathering phase is filtered using a training algorithm to identify and capitalise on the user’s qualities.
Phase of recommendation: It suggests or makes predictions about the kinds of products the user would like. This can be done directly using the dataset gathered during the information gathering phase, which may be memory-based or model-based, or indirectly using the system’s observations of the user’s actions.