Frictionless Information Systems (FIS) are a class of software systems that automatically bring users information and services related to what they are doing. These systems bridge the gap between the information a user wants and the repositories that contain the information - making the user more effective and informed. By mining information about who you are, where you are, what you are doing, or what you are interacting with, these systems can seamlessly deliver information that can help people as they move around the world. From systems that provide relevant, on-point information and services while you watch TV to intelligent search assistant that suggest relevant resources while you read and write documents on your desktop, the Frictionless Information System fundamentally changes the way people interact with information. A user's context is incredibly powerful. At the FIS's core is context. Effectively capturing and analyzing user's context allows the FIS to provide highly relevant information and resources to users engaged in a variety of tasks. This dissertation describes a framework for maximizing the relevance of results provided by FISs by exploiting a user's context to ensure the user is delighted and informed by the quality and breadth of results presented. The techniques described are designed to work in real-time and scale to include a wide range of common user contexts. I will explore two FISs, Watson and Beyond Broadcast, that utilize these techniques to perform proactive, contextual information retrieval in two different domains. To better understand the effectiveness of this framework, I will present a laboratory evaluation that explores the relevance of results produced by our systems. The goal of this work is to provide a real-time, scalable, and effective framework for integrating a user's context into automatic information retrieval.

Last modified

• 06/25/2018

Creator

• Sanjay Chand Sood

DOI

• https://doi.org/10.21985/N2PH7J

Subject

• Computer Science

Keyword

• relevance assessment
• query formation
• agents
• information retrieval
• result processing

Date created

• 2007-05-11

Resource type

• Dissertation

Rights statement

• In Copyright