Julian Jarrett

Collaborative Infrastructure for On-Demand Crowdsourced Tasks

Increasing popularity in the use of crowd sourcing has led to many tasks that can be fulfilled by the wisdom of human actors. When natural disasters or criminal activities occur, then sometimes crowd sourced tasks must be generated in real-time and must be fulfilled in an on-demand fashion. Effective use of crowd sourcing techniques requires an array of services that fulfil many dimensions of the overall problem such as resource selection and allocation, solution selection, and compensation. Any architecture that can provide these services in real-time, on demand requires a dynamic configurable infrastructure. This paper describes an adaptive framework for on-demand crowd sourced tasks supported by a design pattern-inspired architecture.

Mobile Services for Enhancing Human Crowdsourcing with Computing Elements

Crowdsourcing enables one to leverage the power of the crowd. Normally, it involves utilizing humans for tasks that machines have difficulty performing. We propose a system, delivered as a mobile service, which dynamically adapts to the application domain and selects a combination of human and machine crowdsourcing components. Our work is towards the design of elastic systems that adaptively optimizes the use of human and automated software resources in order to maximize overall performance. We propose a performance model that predicts both human and machine outcomes for a certain task and then optimizes task assignment accordingly. Our experimentation shows that our proposed system significantly enhances the outcome precision of a crowdsourced task.

Interoperability and Scalability for Worker-Job Matching across Crowdsourcing Platforms

Crowdsourcing labor market platforms consist of a variety of jobs spanning multiple problem domains and their respective specialized or diverse worker pools. Each platform currently operates independently and isolated from the potential benefits of sharing job and worker pool data across platforms. Previous work introduces infrastructure that optimizes the sharing of both job and worker data collectively, called the open push-pull model. In this paper, to support automated recommendation of workers, we introduce an interoperability standard and computational method that facilitates the aggregation of job data while supporting scalability in response to increasing volumes of data. (i.e. workers and jobs continuously entering the system).

Evaluation of User Engagement on Social Media to Leverage Active and Passive Communication

Individuals in society differ ideologically both online and offline. As the nature of discussions and communication evolve, so do the dynamics within collective groups. User participation on issues such as political discourse affect the opinions of collective groups prior to, during, and after the occurrence of significant events. Changes in engagement can be influenced by choice in words during these discussions. This results in naturally insulating effects that prevent a more comprehensive discussion, and a further challenge exists when opposing, less vocal voices, have a disproportional impact in less conspicuous ways. This paper introduces a communicative model to understand event stimuli triggering user participation of both active and passive actors. This approach contributes to developing more engaging on-line discussion as the nature of the communication evolves.

Regression and Mental Models for Decision Making on Robotic Biped Goalkeepers

We investigate the decision-making and behavior of robotic biped goalkeepers, applied to the RoboCup 3D Soccer Simulation League. We introduce two approaches to the goalkeepers behavior: first a heuristics-based approach that uses linear regression and Kalman filters for improved perception, and another based on mental models which uses nonlinear regression for ball trajectory filtering. Our experiments consist of 30,000 kick-and-save tests, using 100 random angle and distance kicks from six distance categories and four angle categories repeated 30 times. Our benchmark results show that both proposed approaches bring significant improvements for the goalkeepers save success rates

Self-Generating a Labor Force for Crowdsourcing: Is Worker Confidence a Predictor of Quality?

>

WKSP) Sentiment Analysis of Twitter Samples That Differentiates Impact of User Participation Levels

200i?ź% and validate the applicability of the novel mental model based decision-making process.

Adaptive Computing (and Agents) for Enhanced Collaboration

Crowd sourcing is a paradigm where activities are outsourced to human actors (i.e. The crowd) with the aim of discovering and evaluating solutions. This paradigm can also be extended to develop a collective intelligence of large-scale crowd communities that when combined with traditional computing resources can derive solutions that neither humans nor machines can solve alone. Such hybrid systems, or elastic systems, could involve large numbers of people with varying expertise, skills, interests, and incentives and varied computing resources. Elastic frameworks have been proposed to improve the performance of these systems to make them more efficient, robust, and scalable. To meet these requirements, we investigate a novel approach that provides decision support and risk assessment in an elastic framework. In our approach, we infer a probabilistic framework of a hybrid system and use probabilistic odds as a quantitative measure of the capability of human and computing resources to execute a task. As new evidence becomes available, we propagate updated odds throughout our framework to update our prior belief and risks for computing elements. In this approach, the elastic framework can exploit this information in such a way that self-learning is coupled with the ability to extract actionable insights that optimize judgment under uncertainty.