Simon Breslav

Coupling stochastic occupant models to building performance simulation using the discrete event system specification formalism

When applying occupant models to building performance simulation (BPS), it is common practice to use a discrete-time approach requiring fixed time steps. Consequently, a simulated occupants decisions do not increase in frequency in response to rapid changes in environmental conditions. Furthermore, as illustrated in this study through the analysis of a discrete-time EnergyPlus simulation, changing the time step between simulation runs may have a dramatic effect on BPS predictions. It is therefore necessary to adhere to a prescribed time step, which may complicate the synchronization of events when models of different domains are coupled. The main contribution of this study is an investigation of the viability of employing the discrete event system specification (DEVS) formalism to represent occupant behaviour without fixed and prescribed time steps. Results indicate that using an adaptive time advancement scheme, the DEVS formalism permits realistic patterns of decision-making while facilitating the coupling of stochastic occupant models with thermal and heating, ventilation and air-conditioning models.

Mimic: visual analytics of online micro-interactions

We present Mimic, an input capture and visual analytics system that records online user behavior to facilitate the discovery of micro-interactions that may affect problem understanding and decision making. As aggregate statistics and visualizations can mask important behaviors, Mimic can help interaction designers to improve the usability of their designs by going beyond aggregates to examine many individual user sessions in detail. To test Mimic, we replicate a recent crowd-sourcing experiment to better understand why participants consistently perform poorly in answering a canonical conditional probability question called the Mammography Problem. To analyze the micro-interactions, the Mimic web application is used to play back user sessions collected through remote logging of client-side events. We use Mimic to demonstrate the value of using advanced visual interfaces to interactively study interaction data. In the Mammography Problem, issues like user confusion, low confidence, and divided-attention were found based on participants changing their answers, doing repeated scrolling, and overestimating a base rate. Mimic shows how helpful detailed observational data can be and how important the careful design of micro-interactions is in helping users to successfully understand a problem, find a solution, and achieve their goals.

Benefits of visualization in the Mammography Problem

Trying to make a decision between two outcomes, when there is some level of uncertainty, is inherently difficult because it involves probabilistic reasoning. Previous studies have shown that most people do not correctly apply Bayesian inference to solve probabilistic problems for decision making under uncertainty. In an effort to improve decision making with Bayesian problems, previous work has studied supplementing the textual description of problems with visualizations, such as graphs and charts. However, results have been varied and generally indicate that visualization is not an effective technique. As these studies were performed over many years with a variety of goals and experimental conditions, we sought to re-evaluate the use of visualization as an aid in solving Bayesian problems. Many of these studies used the classic Mammography Problem with visualizations portraying the problem structure, the quantities involved, or the nested-set relations of the populations involved. We selected three representative visualizations from this work and developed two hybrid visualizations, combining structure types and frequency with structure. We also included a text-only baseline condition and a text-legend condition where all nested-set problem values were given to eliminate the need for participants to estimate or calculate values. Seven hundred participants evaluated these seven conditions on the classic Mammography Problem in a crowdsourcing system, where micro-interaction data was collected from the participants. Our analysis of the user input and of the results indicates that participants made use of the visualizations but that the visualizations did not help participants to perform more accurately. Overall, static visualizations do not seem to aid a majority of people in solving the Mammography Problem. Graphical abstractDisplay Omitted HighlightsWe evaluate the use of visualization as an aid in solving Bayesian problems.We propose comparability criteria for consistency in visualization design.We created a design space including existing and novel visualizations.700 participants evaluated 7 different visualizations of the Mammography Problem.Participants made use of the visualizations but they did not improve performance.

PhenoBlocks: Phenotype Comparison Visualizations

The differential diagnosis of hereditary disorders is a challenging task for clinicians due to the heterogeneity of phenotypes that can be observed in patients. Existing clinical tools are often text-based and do not emphasize consistency, completeness, or granularity of phenotype reporting. This can impede clinical diagnosis and limit their utility to genetics researchers. Herein, we present PhenoBlocks, a novel visual analytics tool that supports the comparison of phenotypes between patients, or between a patient and the hallmark features of a disorder. An informal evaluation of PhenoBlocks with expert clinicians suggested that the visualization effectively guides the process of differential diagnosis and could reinforce the importance of complete, granular phenotypic reporting.

Annotation Graphs: A Graph-Based Visualization for Meta-Analysis of Data Based on User-Authored Annotations

User-authored annotations of data can support analysts in the activity of hypothesis generation and sensemaking, where it is not only critical to document key observations, but also to communicate insights between analysts. We present annotation graphs, a dynamic graph visualization that enables meta-analysis of data based on user-authored annotations. The annotation graph topology encodes annotation semantics, which describe the content of and relations between data selections, comments, and tags. We present a mixed-initiative approach to graph layout that integrates an analysts manual manipulations with an automatic method based on similarity inferred from the annotation semantics. Various visual graph layout styles reveal different perspectives on the annotation semantics. Annotation graphs are implemented within C8, a system that supports authoring annotations during exploratory analysis of a dataset. We apply principles of Exploratory Sequential Data Analysis (ESDA) in designing C8, and further link these to an existing task typology in the visualization literature. We develop and evaluate the system through an iterative user-centered design process with three experts, situated in the domain of analyzing HCI experiment data. The results suggest that annotation graphs are effective as a method of visually extending user-authored annotations to data meta-analysis for discovery and organization of ideas.

Designing DEVS visual interfaces for end-user programmers

Although the Discrete Event System specification (DEVS) has over recent decades provided systems engineers with a scalable approach to modeling and simulation, the formalism has seen little uptake in many other disciplines where it could be equally useful. Our observations of end-user programmers confronted with DEVS theory or software suggest that learning barriers are largely responsible for this lack of utilization. To address these barriers, we apply ideas from human–computer interaction to the design of visual interfaces intended to promote their users’ effective knowledge of essential DEVS concepts. The first step is to propose a set of names that make these concepts easier to learn. We then design and provide rationale for visual interfaces for interacting with various elements of DEVS models and simulation runs. Both the names and interface designs are evaluated using the Cognitive Dimensions of Notations framework, which emphasizes trade-offs between 14 aspects of information artifacts. As a whole, this work illustrates a generally applicable design process for the development of interactive formalism-based simulation environments that are learnable and usable to those who are not experts in simulation formalisms.

Practical aspects of the DesignDEVS simulation environment

DesignDEVS is a simulation development environment based on the Discrete Event System Specification (DEVS) formalism. This paper provides an in-depth overview of the software while focusing on the practical considerations influencing its design. Practitioners who stand to benefit from systems engineering will approach formalism-based simulation tools with little knowledge of the underlying theory. It is therefore important that theoretical principles, such as the separation of model and simulator, be emphasized by the user interface. Other practical aspects of DesignDEVS include the simplicity of atomic model code, a focus on coupling for collaboration purposes, the enforcement of essential modeling constraints, and a reliance on best practices in cases where strict enforcement might inconvenience users. In DesignDEVS, an issue we refer to as the Insidious Pointer Problem is aggressively tackled through run-time error handling. By contrast, the separation of output values from state transitions is left as a best practice for the sake of user convenience. The design decisions explained in this paper are relevant to developers of other formalism-based tools seeking widespread adoption of scalable modeling and simulation practices.

Can we design an industry classification system that reflects industry architecture

ABSTRACT Interdependencies amongst firms with complementary capabilities lead to the emergence of stable patterns of interfirm relationships observed in global value chains and ecosystems. But current standard industry classification systems group industries into higher order aggregates based on similarity criteria, ignoring the complementarities that induce interdependence. We show how systems theory can be used to design an industry classification system that captures the interindustry interdependencies manifested by buy-sell transactions between firms. Our arguments are three. First, that we can improve upon currently available industry classification systems by clearly identifying criteria for grouping industries into higher order aggregates such as sectors. Second, that a top level grouping based on demand will divide the economy into sectors in a manner that is consistent with global value chains and other configurations of interfirm networks. And third, that roles within demand-based sectors are the redundant feature of interindustry relations that allow us to describe the economy simply. We support our arguments with visualizations of over 53,000 of the largest interfirm transactions in the U.S. economy between 1976 and 2010.

Interactive Instruction in Bayesian Inference

An instructional approach is presented to improve human performance in solving Bayesian inference problems. Starting from the original text of the classic Mammography Problem, the textual expression is modified and visualizations are added according to Mayer’s principles of instruction. These principles concern coherence, personalization, signaling, segmenting, multimedia, spatial contiguity, and pretraining. Principles of self-explanation and interactivity are also applied. Four experiments on the Mammography Problem showed that these principles help participants answer the questions at significantly improved rates. Nonetheless, in novel interactivity conditions, performance was lowered suggesting that more interaction can add more difficulty for participants. Overall, a leap forward in accuracy was found, with more than twice the participant accuracy of previous work. This indicates that an instructional approach to improving human performance in Bayesian inference is a promising direction.