Abstract
This chapter focuses on machine learning algorithms used in real-world applications and the possibilities of using them in libraries. Recommender systems are considered to be very valuable tools as they can be personalized to each user's preferences and needs. The main goal of the recommendation system is to aid users in finding necessary items from an overwhelming number of options. In this chapter, the authors examine two book recommender systems, explain the logic behind the used algorithms, and illustrate the way to adapt the same technologies in other libraries.
TopBackground
Machine learning refers to the development of algorithms that enable machines to mimic human intelligence. Machine learning technology has recently been used in many fields such as image recognition, biomedical applications, natural language processing, and prediction. In 1958, Rosenblatt developed the first neural network that mimicked neural cells in the human brain (Rosenblatt, 1958). In 1975 another breakthrough happened by Werbos as he developed the “multilayer perceptron (MLP),” MLP is a multilayer perceptron which is a neural network that consists of fully connected layers that can produce a set of outputs from inputs (Werbos, 1994). In 1986, Quinlan developed another machine learning technology known as decision trees (Quinlan, 1986), followed by Cortes and Vapnik’s invention of support vector machines (Cortes, 1995). Machine learning technology has grown exponentially through many inventions, such as Adaboost, random forests, and multilayer algorithms.
Arthur Samuel was the first to define machine learning as a computer’s ability to be programmed and learn on its own (Samuel, 1988). Other definitions of Machine learning are “a field of study that gives computers the ability to learn without being explicitly programmed” (Samuel, 1988). It could also be explained as “A computer program is said to learn from experience (E) concerning some class of tasks (T) and performance measure (P), if its performance at tasks in T, as measured by P, improves with experience E’ (Mitchell, 1997). Another definition by Ethem Alpaydin is “Programming computers to optimize a performance criterion using example data or experience” (Alpaydin, 2020). In all previous definitions, we can conclude that machine learning is how computers are taught to perform advanced tasks by feeding them sufficient input data to learn.
Key Terms in this Chapter
Content-Based Filtering (CBF): It is a type of filtering that uses item features to make recommendations based on users’ feedback.
Nearest Neighbor (NN) Algorithm: A machine learning algorithm that depends on proximity to classify datasets and make predictions.
Recommender System RS: It is an application of artificial intelligence and machine learning where we use an artificial intelligence algorithm to suggest products to a user based on the user’s search history and other factors.
Machine Learning (ML): The development of algorithms that enable machines to mimic human intelligence.
Feature Extraction: Refers to the process of manipulating data into a numerical feature that can be processed by the machine while preserving the information in the original dataset.
Supervised Machine Learning: This type of machine learning happens by the use of labeled datasets to train artificial intelligence algorithms to cluster datasets accurately.
Unsupervised Machine Learning: The machine uses artificial intelligence algorithms to cluster unlabeled datasets.
Reciprocal Recommender Systems (RRS): This type of Recommender System recommends items to users from producers, taking into consideration both sides’ preferences.
Collaborative Filtering (CF): It is a type of filtering that uses similarity among users and products to make recommendations.