A Laravel package for systematic construction of symbolic intelligent systems

Pitòn is a Laravel package which offers machine learning utilities for solving hierarchically-arranged problems, where the available data is stored in a MySQL database, through the creation of rule-based models. Therefore, the intent of the first chapters will be to discuss what is a hierarchical structured problem, what are rule-based models, what are rules, and what algorithms are used to extract them. We will then discuss the technologies used by the package, starting from the Laravel framework, that is, the environment in which the package has been developed, and introducing the machine learning Python packages Scikit-Learn and Wittgenstein. Chapter 5 will finally present how the package works, including its structure, the structure of the database used to store the models, which we decided to separate from the application database, and how to use it at its best. Then we will present a possible application of the package to solve a real problem, that is, a social network automatic advertisement system which decides what ads to show the user based on the researches of other users with similar characteristics, also going through a possible package configuration. Last but not least, we will discuss the environment which already implements the package, that is, the University of Ferrara’s Research Center for the Study of Menopause and Osteoporosis platform, serving as a clinical decision support system and therefore allowing physicians to receive suggestions for prevention and treatment for osteoporosis, given their previous choices for similar patients.

Recommended citation: Paparella, Alberto (2021) A Laravel package for systematic construction of symbolic intelligent systems. [Laurea triennale], Università di Ferrara, Corso di Studio in Informatica [L-31].
Download Paper | Download Slides | Have a look at the code