Spam filtering, as a key problem in electronic communication, has drawn significant attention due to increasingly huge amounts of junk email on the Internet. Content-based filtering is one reliable method in combating with spammers' changing tactics. Naive Bayes (NB) is one of the earliest content-based machine learning methods both in theory and practice in combating with spammers, which is easy to implement while can achieve considerable accuracy. In this paper, the traditional online Bayesian classifier are enhanced  by two ways. First, from theory's point of view, we devise a self-adaptive mechanism to gradually weaken the assumption of independence required by original NB in the online training process, and as a result of that our NSNB is no longer ``naive''. Second, we propose other engineering ways to make the classifier more robust and accuracy. The experiment results show that our NSNB does give state-of-the-art classification performance on online spam filtering on large benchmark data sets while it is extremely fast and takes up little memory in comparison with other statistical methods.

We describe an AI-enhanced task management tool developed for a military environment, which differs from office environments in important ways: differing time scales, a focus on teams collaborating on tasks instead of an individual managing her own set of diverse tasks, and a focus on tasklists and standard operating procedures instead of individual tasks. We discuss the Task Assistant prototype, our process for adapting it from an office environment to a military one, and lessons learned about developing AI technology for a high-pressure operational environment.

Considerable progress has been made over the last 15 years on building adaptive control systems to assist pilots in flying damaged aircraft. Once a pilot has regained control of a damaged aircraft, the next problem is to determine the best site for an emergency landing.  In general, the decision depends on many factors including the actual control envelope of the aircraft, distance to the site, weather en route, characteristics of the approach path, characteristics of the runway or landing site, and emergency facilities at the site.  All of these influence the risk to the aircraft, to the passengers and crew, and to people and property on the ground.  We describe an emergency landing planner that takes these various factors into consideration and proposes possible routes and landing sites to the pilot, ordering them according to estimated risk.   We give an overview of the system architecture and input data, describe our modeling of risk, describe how we search the space of landing sites and routes, and give a preliminary performance assessment for characteristic emergency scenarios using the current research prototype.

The goal of the Pedagogical Discourse project is to develop instructional tools that will help students and instructors use discussion boards more effectively, with an emphasis on automatically assessing discussion activities and building tools for promoting student discussion participation and learning. In this paper, we present a two related participation and learning scaffolding tools that exploit natural language processing and information retrieval techniques. The PedaBot tool is designed to aid student knowledge acquisition and promote reflection about course topics by connecting related discussions from a knowledge base of past discussions to the current discussion thread. The MentorMatch tool aims at promoting student participation using student mentors, i.e., course peers with a relatively good understanding of a particular topic. The system identifies students who often provide answers on a given topic and encourages classmates to invite mentors to participate in related discussions. Both tools have been integrated into a live discussion board that is used by an undergraduate computer science course. This paper describes our approaches to applying information retrieval and natural language processing techniques in the development of the tools and presents initial results from instrumentation and survey.

Realistic rendering is the process of generating a 2D image from an abstract description of a 3D scene, aiming at achieving the quality of a photo. The quality of the generated image depends on the accuracy with which the employed render method simulates the behaviour of the light particles through the scene. According to the current practice, it is up to the user to choose optimal settings of input parameters for these methods in terms of time-efficiency, as well as image quality. This is an iterative trial and error process, even for expert users. This paper describes a novel approach based on techniques from the field of data mining and genetic computing to assist the user in the selection of render parameters. Experimental results are presented which show the benefits of this approach.

While the concept of learning by demonstration has been around for many years, recent advances in artificial intelligence technology have led to a resurgence of work in the field. We describe the development and application of learning by demonstration technology to support user creation of automated procedures for a rich collaborative planning environment that is in widespread use by the U.S. Army. User feedback and evaluation results show that the technology can be used effectively by the target user community and that it has tremendous potential for improving the speed and quality of performance for a range of critical tasks.

In this paper, an event-centric commodity trading simulation powered by the multiagent framework is presented. The purpose of this simulation platform is for training novice traders. The simulation is progressed by announcing news events that affect various aspects of the commodity supply chain. Upon receiving these events, market agents that play the roles of producers, consumers, and speculators would adjust their views on the market and act accordingly. Their actions would be based on their roles and also their private information, and collectively they shape the market dynamics. This simulation has been effectively deployed for several training sessions. We will present the underlying technologies that are employed and discuss the practical significance of such platform.

The use of statistical AI techniques in authorship recognition (or stylometry) has contributed to literary and historical breakthroughs. These successes have led to the use of these techniques in criminal investigations and prosecutions.  However, few have studied adversarial attacks and their devastating effect on the robustness of existing classification methods. This paper presents a framework for adversarial attacks including obfuscation attacks, where a subject attempts to hide their identity imitation attacks, where a subject attempts to frame another subject by imitating their writing style.  The major contribution of this research is that it demonstrates that both attacks work very well.  The obfuscation attack reduces the effectiveness of the techniques to the level of random guessing and the imitation attack succeeds with 68-91% probability depending on the stylometric technique used.  These results are made more significant by the fact that the experimental subjects were unfamiliar with stylometric techniques, without specialized knowledge in linguistics, and spent little time on the attacks. This paper also provides another significant contribution to the field in using human subjects to empirically validate the claim of high accuracy for current techniques (without attacks) by reproducing results for three representative stylometric methods.

Many city governments are under pressure to optimize the utilization of their resources to respond to fire, rescue and medical emergencies. In this paper we describe a simulation-based optimization software called SOFER that learns from a history of emergency requests to optimize the placement of resources and response policies. We describe a two-level random-restart hill climbing approach that yields policies which perform better than the current practice, satisfy the usability constraints, and are sensitive to optimization metrics and population changes. Some of the policies learned by the system give insight into response practices that would otherwise be counterintuitive.

We discuss experiences from evaluating the learning performance of a user-adaptive personal assistant agent.  We discuss the challenge of designing adequate evaluation and the tension of collecting adequate data without a fully functional, deployed system.  Reflections on negative and positive experiences point to the challenges of evaluating user-adaptive AI systems.  Lessons learned concern early consideration of evaluation and deployment, characteristics of AI technology and domains that make controlled evaluations appropriate or not, holistic experimental design, implications of "in the wild" evaluation, and the effect of AI-enabled functionality and its impact upon existing tools and work practices.