Tomás E. Uribe

Active preference learning for personalized calendar scheduling assistance

We present PLIANT, a learning system that supports adaptive assistance in an open calendaring system. PLIANT learns user preferences from the feedback that naturally occurs during interactive scheduling. It contributes a novel application of active learning in a domain where the choice of candidate schedules to present to the user must balance usefulness to the learning module with immediate benefit to the user. Our experimental results provide evidence of PLIANTs ability to learn user preferences under various conditions and reveal the tradeoffs made by the different active learning selection strategies.

Integration of heterogeneous knowledge sources in the CALO query manager

We report on our effort to build a real system for integrating heterogeneous knowledge sources with different query answering and reasoning capabilities. We are conducting this work in the context of CALO (Cognitive Assistant that Learns and Organizes), a multidisciplinary project funded by DARPA to create cognitive software systems.

Deploying a personalized time management agent

We report on our ongoing practical experience in designing, implementing, and deploying PTIME, a personalized agent for time management and meeting scheduling in an open, multi-agent environment. In developing PTIME as part of a larger assistive agent called CALO, we have faced numerous challenges, including usability, multi-agent coordination, scalable constraint reasoning, robust execution, and unobtrusive learning. Our research advances basic solutions to the fundamental problems; however, integrating PTIME into a deployed system has raised other important issues for the successful adoption of new technology. As a personal assistant, PTIME must integrate easily into a users real environment, support her normal workflow, respect her authority and privacy, provide natural user interfaces, and handle the issues that arise with deploying such a system in an open environment.

Automatic analysis of firewall and network intrusion detection system configurations

Given a network that deploys multiple firewalls and network intrusion detection systems (NIDSs), ensuring that these security components are correctly configured is a challenging problem. Although models have been developed to reason independently about the effectiveness of firewalls and NIDSs, there is no common framework to analyze their interaction. This paper presents an integrated, constraint-based approach for modeling and reasoning about these configurations. Our approach considers the dependencies among the two types of components, and can reason automatically about their combined behavior. We have developed a tool for the specification and verification of networks that include multiple firewalls and NIDSs, based on this approach. This tool can also be used to automatically generate NIDS configurations that are optimal relative to a given cost function.

An Architecture for an Adaptive Intrusion-Tolerant Server

We describe a general architecture for intrusion-tolerant enterprise systems and the implementation of an intrusion-tolerant Web server as a specific instance. The architecture comprises functionally redundant COTS servers running on diverse operating systems and platforms, hardened intrusion-tolerance proxies that mediate client requests and verify the behavior of servers and other proxies, and monitoring and alert management components based on the EMERALD intrusion-detection framework. Integrity and availability are maintained by dynamically adapting the system configuration in response to intrusions or other faults. The dynamic configuration specifies the servers assigned to each client request, the agreement protocol used to validate server replies, and the resources spent on monitoring and detection. Alerts trigger increasingly strict regimes to ensure continued service, with graceful degradation of performance, even if some servers or proxies are compromised or faulty. The system returns to less stringent regimes as threats diminish. Servers and proxies can be isolated, repaired, and reinserted without interrupting service.

Conflict negotiation among personal calendar agents

We will demonstrate distributed conflict resolution in the context of personalized meeting scheduling. The demonstration will show how distributed constraint optimization can be used to facilitate interaction between cognitive agents and their users. The system is part of the CALO personal cognitive assistant that will also be explored during the demonstration.

Combining Monitors for Runtime System Verification

Abstract Runtime verification permits checking system properties that cannot be fully verified off-line. This is particularly true when the system includes complex third-party components, such as general-purpose operating systems and software libraries, and when the properties of interest include security and performance. The challenge is to find reliable ways to monitor these properties in realistic systems. In particular, it is important to have assurance that violations will be reported when they actually occur. For instance, a monitor may not detect a security violation if the violation results from a series of system events that are not in its model. We describe how combining runtime monitors for diverse features such as memory management, security-related events, performance data, and higher-level temporal properties can result in more effective runtime verification. After discussing some basic notions for combining and relating monitors, we illustrate their application in an intrusion-tolerant Web server architecture under development at SRI.