Modelling Requirements for Content Recommendation Systems

TL;DR

This paper explores modeling requirements for content recommendation systems in social networks, emphasizing user role dynamics and proposing a combined approach using i-star and Petri Nets to better capture system behaviors.

Contribution

It identifies the limitations of i-star for dynamic modeling and introduces a hybrid modeling approach with Petri Nets to improve requirement representation.

Findings

i-star alone cannot model user role dynamics effectively

Petri Nets capture system dynamics and concurrency

Hybrid approach enhances requirement analysis for OSNs

Abstract

This paper addresses the modelling of requirements for a content Recommendation System (RS) for Online Social Networks (OSNs). On OSNs, a user switches roles constantly between content generator and content receiver. The goals and softgoals are different when the user is generating a post, as opposed as replying to a post. In other words, the user is generating instances of different entities, depending on the role she has: a generator generates instances of a "post", while the receiver generates instances of a "reply". Therefore, we believe that when addressing Requirements Engineering (RE) for RS, it is necessary to distinguish these roles clearly. We aim to model an essential dynamic on OSN, namely that when a user creates (posts) content, other users can ignore that content, or themselves start generating new content in reply, or react to the initial posting. This dynamic is key to designing OSNs, because it influences how active users are, and how attractive the OSN is for existing, and to new users. We apply a well-known Goal Oriented RE (GORE) technique, namely i-star, and show that this language fails to capture this dynamic, and thus cannot be used alone to model the problem domain. Hence, in order to represent this dynamic, its relationships to other OSNs' requirements, and to capture all relevant information, we suggest using another modelling language, namely Petri Nets, on top of i-star for the modelling of the problem domain. We use Petri Nets because it is a tool that is used to simulate the dynamic and concurrent activities of a system and can be used by both practitioners and theoreticians.