As large-scale online classes become more prevalent there is great interest in finding ways to model students at scale in these classes in order to predict outcomes. Student models, if successful, would help determine strong predictors of student success, which would highlight potential causal factors for such success, allowing schools to focus on refinements and interventions that positively impact their student outcomes. In this research, TutorGen has partnered with Western Governors University (WGU), a large online university, and gathered data at scale in order to build exploratory models to predict student outcomes. This paper presents our results so far in successfully identifying students who will pass (or even take) the final exam. We have examined the order in which students take courses, as well as the timing of starting and completing work; our initial analysis reveals that these are strong predictors of course outcomes.

Computer simulations of complex food-webs are an important tool for deepening our understanding of these systems. Yet most computer models assume, rather than generate, key system-level patterns, or use mathematical modeling approaches that make it difficult to account fully for non-linear dynamics. In this article we present a computer simulation model that addresses these concerns by focusing on assumptions of agent attributes rather than agent outcomes. Our model utilizes the techniques of Complex Adaptive Systems and Agent-Based Modeling so that system-level patterns of a general ecosystem emerge from the interactions of thousands of individual simulated agents. This methodology has been validated in previous work by using this general simulation model to replicate fundamental properties of an ecosystem, including: (1) the predator-prey oscillations found in Lotka-Volterra; (2) the “stepped pattern” of biomass accrual from resource enrichment; (3) the Paradox of Enrichment; and (4) Gause’s Law. In this work we explore further the fundamental properties of this generative model in the context of the Red Queen Hypothesis, also referred to as the “arms race” between antagonistic species, e.g. predators and prey. We find that improvements in the competitive landscape for a single entity in a predator species does not generally confer a benefit on the predator species as a whole, and may even be detrimental to the predator population. This non-intuitive result is shown through two methods of adjusting the predators effectiveness in consuming prey. We further explore this idea by explicitly accounting for individual entity's energy requirements, and also allowing evolutionary adaptation for an effectiveness / energy trade-off. 1.0 Overview The literature on marine and terrestrial ecosystems is long and varied, encompassing both theoretical models (e.g.: Grimm, 1999; DeAngelis & Mooij, 2005) and empirical surveys (e.g.: Christensen et al., 2003; Frank et al., 2005). Some significant differences between model results and real-world surveys have persisted for years, and it has been difficult identifying fundamental principles relative to the many complicating factors that can be found in existent ecosystems. For example, in the early 1980s Oksanen et al. examined multiple trophic levels in a predator-prey system using mathematical models, in order to determine whether species population (bio-mass) is fundamentally controlled by resources – as was the conventional wisdom at the time – or dominated by predation (Oksanen et al., 1981). In describing this work, Power states that these models produce “a stepped pattern of biomass accrual” (Power, 1992); Brett and Goldman further characterize the Oksanen et al. results, saying that “In food webs with an odd number of trophic levels, increases in primary production should lead to increased biomass for odd-numbered trophic levels and no change in biomass for even-numbered trophic levels. Conversely, in food webs with an even number of trophic levels, increases in primary production should lead to increased biomass for even-numbered trophic levels and no change in biomass for odd-numbered trophic levels” (Brett & Goldman,

The amount of data available to build simulation models of schools is immense, but using these data effectively is difficult. Traditional methods of computer modeling of educational systems often either lack transparency in their implementation, are complex, and often do not natively simulate non-linear systems. In response, we advocate a Complex Adaptive Systems approach towards modeling and data mining. By simulating agent-level attributes rather than system-level attributes, the modeling is inherently transparent, easily adjustable, and facilitates analysis of the system due to the analogous nature of the simulated agents to real-world entities. We explore the design a CAS model of schools using multiple levels of data from varied data streams.

We present a new ITS system called SCALE (Student Centered Adaptive Learning Engine), which is focused on improving learning outcomes by using data collected from existing and emerging educational technology systems combined with machine learning techniques to automatically generate adaptive capabilities. This allows for the creation of intelligent tutoring systems in a less costly fashion in terms of time and effort. SCALE uses data logs collected from an existing educational technology system to create the initial adaptivity and then improves over time as additional data is added or with the help of human input. This paper describes two main adaptive capabilities of problem selection and hint generation.

Computer simulations of complex food-webs are important tools for deepening our understanding of these systems. Yet most computer models assume, rather than generate, key system-level patterns, or use mathematical modeling approaches that make it difficult to fully account for nonlinear dynamics. In this paper, we present a computer simulation model that addresses these concerns by focusing on assumptions of agent attributes rather than agent outcomes. Our model utilizes the techniques of complex adaptive systems and agent-based modeling so that system level patterns of a marine ecosystem emerge from the interactions of thousands of individual computer agents. This methodology is validated by using this general simulation model to replicate fundamental properties of a marine ecosystem, including: (i) the predator–prey oscillations found in Lotka–Volterra; (ii) the stepped pattern of biomass accrual from resource enrichment; (iii) the Paradox of Enrichment; and (iv) Gause's Law.

An article in Science magazine (Bhattacharjee, 2007) discussed how the U.S. military is interested in enlisting the help of multidisciplinary scientific experts to better understand “how local populations behave in a war zone.” The article mentioned the “Human Social Culture Behavior Modeling” (HSBC) program at U.S. Department of Defense and indicated, through a few anecdotal examples, the types of prior research emanating from multidisciplinary fields that may be considered the state of the art.

The Association for the Advancement of Artificial Intelligence was pleased to present the 2011 Fall Symposium Series, held Friday through Sunday, November 4–6, at the Westin Arlington Gateway in Arlington, Virginia. The titles of the seven symposia are as follows: (1) Advances in Cognitive Systems; (2) Building Representations of Common Ground with Intelligent Agents; (3) Complex Adaptive Systems: Energy, Information, and Intelligence; (4) Multiagent Coordination under Uncertainty; (5) Open Government Knowledge: AI Opportunities and Challenges; (6) Question Generation; and (7) Robot‐Human Teamwork in Dynamic Adverse Environments. The highlights of each symposium are presented in this report.

The Association for the Advancement of Artificial Intelligence was pleased to present the 2010 Fall Symposium Series, held Thursday through Saturday, November 11–13, at the Westin Arlington Gateway in Arlington, Virginia. The titles of the eight symposia are as follows: (1) Cognitive and Metacognitive Educational Systems; (2) Commonsense Knowledge; (3) Complex Adaptive Systems: Resilience, Robustness, and Evolvability; (4) Computational Models of Narrative; (5) Dialog with Robots; (6) Manifold Learning and Its Applications; (7) Proactive Assistant Agents; and (8) Quantum Informatics for Cognitive, Social, and Semantic Processes. The highlights of each symposium are presented in this report.

We present the results of a pilot study created to explore a subset of complex data, utilizing an agent-based model simulation tool. These results center on data taken from hospital admission records, tracking patient attributes and how they relate to patient outcomes. The focus of this work is to highlight three design principles: 1) using an iterative process between the modeling of a system and the grounding of that simulation with real-world data; 2) a focus on agent primitives, emphasizing bottom-up emergence of effects, rather than top-down control; and 3) integration of various "theories" of patient care and hospital effectiveness as a method for experimentation with Complex Adaptive System-based data mining.

A Complex Adaptive System is a collection of autonomous, heterogeneous agents, whose behavior is defined with a limited number of rules. A Game Theory is a mathematical construct that assumes a small number of rational players who have a limited number of actions or strategies available to them. The CAS method has the potential to alleviate some of the shortcomings of GT. On the other hand, CAS researchers are always looking for a realistic way to define interactions among agents. GT offers an attractive option for defining the rules of such interactions in a way that is both potentially consistent with observed real-world behavior and subject to mathematical interpretation. This article reports on the results of an effort to build a CAS system that utilizes GT for determining the actions of individual agents. © 2009 Wiley Periodicals, Inc. Complexity, 16,24–42, 2010