When: Wednesday, September 26, 12:00 p.m.

Where: 3305 Newell-Simon Hall

Daniel Tunkelang, Chief Scientist, Endeca

Special Seminar

Abstract:
In 2005, Gary Marchionini delivered a lecture entitled "Toward Human-Computer Information Retrieval," in which he put forth a vision of "systems that involve people and machines continuously learning and changing together." One of Marchionini's observations was that "Systems should increase user responsibility as well as control; that is, information systems require human intellectual effort, and good effort is rewarded."

Modern information access systems illustrate the power of HCIR: they sidestep the usual IR tradeoff of precision vs. recall by progressively soliciting more information from the user. Fundamentally, these approaches focus not so much on ranking query results as on digesting them, since the latter enables an effective and efficient communication channel between the system and the user.

At Endeca, we have embraced the vision of HCIR. In this talk, we will discuss novel techniques we have developed in support of this vision. We will also discuss the problem of evaluating HCIR systems, since a robust evaluation framework is key to making progress in their development.

As Chief Scientist at Endeca, Daniel Tunkelan develops innovative approaches to address the problems of information access. Endeca's software helps over 400 organizations integrate, discover, and navigate enterprise data to solve business problems associated with information overload. Endeca's customers include a wide range of enterprises, such as Bank of America, Boeing, Costco, Defense Intelligence Agency, General Motors, Home Depot, Hyatt, IBM, Nike, Time Warner, and Walmart. Daniel received his PhD in Computer Science from CMU and completed his undergraduate studies in Computer Science and Mathematics at MIT.

<< Back