2015 Cheriton Research Symposium

10:00 a.m. - 3:00 p.m.

DC 1302 - Mark Giesbrecht - Welcome and Opening Remarks

10:30 a.m. - 11:30 a.m.

DC 1302 - Jimmy Lin - Dogma and Heresy in Information Retrieval

The not uncommon refrain that search is a "solved problem" arises mostly from the false equivalence between search (i.e., finding information) in general and desktop web search. Indeed, there is so much more to search than just web pages: email search, geo search, electronic discovery, question answering, real-time search, just to name a few real-world applications. In this talk, I will begin with an overview of dogma — "how (web) search is done today" — and then present a number of heretical viewpoints that challenge dogmatic assumptions. Some of these concern the construction and organization of inverted index structures, architectures for distributed search, and interfaces for complex exploratory search tasks. I will tie many of these ideas to broader computing trends and speculate on future directions in information retrieval and "big data" research.

12:00 p.m. -
1:00 p.m.

DC 1302 - Anna Lubiw - Flattening and Unfolding Convex Polyhedra

The problem of flattening or unfolding a polyhedron has applications in manufacturing 3D shapes out of metal, cardboard or plastic.

I will talk about work with a recent graduate student, Stephen Kiazyk, on a new method to cut the surface of a convex polyhedron and unfold it into a simple non-overlapping polygon in the plane.

I will also talk about work with a former student, Eric Demaine, plus others, on a natural method to continuously flatten the surface of any convex polyhedron to the plane (e.g. imagine flattening a paper bag) without making any cuts in the surface. Of necessity the surface must be flexible.

DC 1302 - Jesse Hoey - Socio-Cultural and Affective Intelligence for Cognitive Assistive Technologies 

Over the past decade, my team has developed probabilistic, decision-theoretic, and vision-based technologies that can assist persons with cognitive disabilities with a range of activities using monitoring and cueing (prompting). Although these technologies are reasonably effective, the interventions (e.g. audio-visual cues or prompts) are rendered (mostly) independently of users and context. In this talk, I will discuss how this is a significant barrier to adoption, and how this barrier can be lifted by automated reasoning about the personal (socio-cultural and affective) identities of users, and about the affective and emotional delivery of interventions. I will introduce a socio-cultural reasoning engine called "Bayesact" that can be used to provide this level of affective reasoning. Bayesact is based on 25 years of research in sociology and cognitive psychology. Bayesact can learn the affective identity of a user during an interaction, and can tailor prompts/cues/treatments/interventions to specific individuals in a way that ensures smoother and more effective uptake and response. I will give an introduction to this reasoning engine, and will discuss how affective reasoning is being used to create truly adaptive treatment and therapeutic technologies.