What do you mean by autonomy?
Dr. Renata Guizzardi
Assistant Professor, IEBIS Department, University of Twente.
Autonomy is a key desired quality of many systems, for example robots, intelligent agents and autonomous cars. Without autonomy, a system is not able to decide and act on behalf of their stakeholders. However, what autonomy actually means is not clear or consensual. In this talk, we discuss an ontological view of autonomy and its implications for the elicitation and analysis of requirements aiming at the development of intelligent systems.
Renata Guizzardi is an Assistant Professor at the Industrial Engineering and Business Information Systems Department of the University of Twente, in the Netherlands. Moreover, she is a founding member of the Ontology & Conceptual Modeling Research Group (NEMO) and of the Laboratory of Supporting Technologies for Collaborative Networks (LabTAR), at UFES, Brazil, where she was based from 2009-2016. For around 30 years, she has been busy with research work on Computer-Assisted Education, Requirements Engineering, Conceptual Modeling and Ontologies, focusing on the interplay of these research areas to improve the development of information systems and organizational practices.
A Non-parametric Model for Constrained Assortment Optimization
Davide Merolla
PhD student, La Sapienza University of Rome.
This research tackles a constrained assortment optimization problem faced by retailers, where the selection of products to offer and their respective quantities must be chosen to maximize profit. Departing from traditional choice models, this study proposes a non-parametric approach to capture consumer’s familiarity with a product as well as complementarity and substitution effects between items. We formulate a mixed-integer linear programming (MILP) model that embeds these features, allowing products to deviate from historical sales but penalizing those deviations violating complementarity and substitution relations. Retailers can easily calibrate penalties and other model parameters to incorporate consumer behavior based on their specific objectives. The product hierarchy in our model is organized in a multi-level structure, forming a nested category tree. To handle this structure effectively, we devise a matheuristic approach based on decomposition: a master problem assigns capacities to macro-categories, while subproblems determine the assortment and related quantities at lower category levels by solving to optimality MILP models on sub-trees. Computational experiments conducted on different instances show the effectiveness of the proposed method, and provide insights into assortment decisions based on consumer behavior and retailer’s preferences.
Davide Merolla is an industrial PhD student at La Sapienza University of Rome, where he has been attending the PhD program in Operations Research since November 2020. He belongs to the Department of Computer, Automatic and Management Engineering (DIAG), at the Faculty of Information Engineering, Computer Science and Statistics. The main topic of his research activity is Supply Chain Optimization, conducted in collaboration with Spindox, an Italian private company working in the field of optimization and data science. He also dealt with Bilevel Programming, Portfolio Optimization and their joint applications.