Philippe J. Giabbanelli

Combining fuzzy cognitive maps with agent-based modeling: Frameworks and pitfalls of a powerful hybrid modeling approach to understand human-environment interactions

Abstract Agent-based modeling (ABM) is an established technique to capture human-environment interactions in socio-ecological systems. As a micro-model, it explicitly represents each agent, such that heterogeneous decision-making processes (e.g. based on the beliefs and experiences of stakeholders) can anticipate the socio-environmental consequences of aggregated individual behaviors. In contrast to ABM, Fuzzy Cognitive Mapping takes a macro-level view of the world that represents causal connections between concepts rather than individual entities. Researchers have expressed interest in reconciling the two, i.e. taking a hybrid approach and drawing of the strengths of each to more accurately model socio-ecological interactions. The intuition is to take FCMs, which can be quickly developed using participatory modeling tools and use them to create a virtual population of agents with sophisticated decision-making processes. In this paper, we detail two ways in which this combination can be done, and highlight the key questions that modelers need to be mindful of.

Effects of success v failure cases on learner-learner interaction

Abstract Studies have found that students struggle to challenge their peers and engage in co-construction of knowledge when in asynchronous problem-based learning (PBL) contexts. In other settings, case libraries have been shown to support problem solving competencies, such as argumentation and problem representation. However, research has yet to study how the design and types of cases impact learner-learner interaction. To accommodate that gap, this study used content analysis and sequential analysis to ascertain how learner interaction differed when participants had access to success- and failure-based case libraries. Results found the failure-based condition had higher overall number of postings and differed in terms of the number of elicitations and planning (meta) interactions. Finally, results of the sequential analysis indicated participants in the success-based condition were more likely to begin planning their final assignment earlier, while the failure condition was more likely to continue engaged in collaborative problem-solving with their peers. Given these differences, the findings suggest failure-based cases may serve as a catalyst for learner-learner interaction when compared with success-based cases. Implications for practice, case-based reasoning, and failure-driven memory theory are discussed.

Teaching Computational Modeling in the Data Science Era

Abstract Integrating data and models is an important and still challenging goal in science. Computational modeling has been taught for decades and regularly revised, for example in the 2000s where it became more inclusive of data mining. As we are now in the ‘data science’ era, we have the occasion (and often the incentive) to teach in an integrative manner computational modeling and data science. In this paper, we reviewed the content of courses and programs on computational modeling and/or data science. From this review and our teaching experience, we formed a set of design principles for an integrative course. We independently implemented these principles in two public research universities, in Canada and the US, for a course targeting graduate students and upper-division undergraduates. We discuss and contrast these implementations, and suggest ways in which the teaching of computational science can continue to be revised going forward.

Combining association rule mining and network analysis for pharmacosurveillance

Retailers routinely use association mining to investigate trends in the use of their products. In the medical world, association mining is mostly used to identify associations between symptoms and diseases, or between drugs and adverse events. In comparison, there is a relative paucity of work that focuses on relationships between drugs exclusively. In this work, we use the Medical expenditure panel survey to examine relationships between drugs in the United States. In addition to examining the rules generated by association mining, we introduce the notion of a target drug network and demonstrate via different drugs that it can offer additional medical insight. For example, we were able to find drugs that are commonly taken together despite containing the same active compound. Future work can expand on the concept of target drug network, for example, by annotating the networks with the compounds and intended uses of each drug, to yield additional insight for pharmacosurveillance as well as pharmaceutical companies.

Modeling information spread in polarized communities: Transitioning from legacy media to a Facebook world

Rumors have played an important role in social life for centuries, with early examples including their use to steer Roman politics. Todays world includes entire industries focused on digital misinformation, whose rumors can spread quickly via social networks such as Facebook not only because of their structure (e.g., clustering) but also because individuals can place an excessively high trust in information originating from their friends. Relaying information from our friends and ignoring or being unaware of other opinions leads to polarized groups, such as liberals or conservatives in a political context. While numerous models of rumor spreads have been proposed, their focus was more often on the conditions to stop/verify one rumor than in accounting for a polarized context. In this paper, we develop a new model of rumor spread with two different susceptibility rates, which can be used to investigate cases in which the population can be sub-divided with respect to one rumor (e.g. based on political opinions or socio-economic factors such as educational attainment). We describe the dynamics of the model using differential equations, and present numerical results regarding the model behavior with respect to key parameters such as the rate with which rumors are forgotten. While our work took into account network features (e.g., average degree), it is of particular interest for future work to examine the interplay between the network structure and the distribution of susceptibility rates.

A Novel Visualization Environment to Support Modelers in Analyzing Data Generated by Cellular Automata

In the ‘big data’ era the attention is often on deriving models from vast amounts of routinely collected data, for example to lear about human behaviors. However, models themselves can produce a large amount of data which has to be analyzed. In this paper, we focus on visually exploring data produced by a type of discrete simulation models known as ‘cellular automaton’ (CA). In particular, we visualize two-dimensional CA with square cells, which can intuitively be thought of as a grid of colored cells. This type of CA is usually visualized using a slider to display the whole grid at each time of the simulation, but this can make it challenging to see patterns over the whole simulations because of change blindness. Consequently, our new visualization framework uses a temporal clock glyph to show the successive states of each cell on the same display. This approach is illustrated for three classical models using CA: an epidemic (a human health model), sandpiles (a self-organized dynamical system), and fire spread (a geographical model). Several improvements to the framework are discussed, in part based on feedback collected from trained modelers.