A data analytics specialist at the US Department of Homeland Security has put out feelers for research collaboration with New Zealand institutions.
There might even be some US government funding available, he says, in an approach that will be enthusiastically received in at least one quarter -- the Human Interface Technology (HIT) Lab at Canterbury University.
Jim Thomas visited New Zealand last month and presented a summary of some of his work, concerning detection of unusual patterns in data that may indicate plans for terrorism or other crime.
He has recently obtained clearance from the US government to enter jointly funded ventures with Canadian research teams and sees no reason why similar permission could not be obtained for New Zealand, he says.
Thomas's research and development is concerned with finding new ways of interacting with large amounts of data, particularly unstructured data -- enabling zettabytes (10 to the power 21 bytes) of data to be economically represented in ways that make patterns such as clusters of similarity and outliers readily appreciable by the human eye.
He gave examples of spotting similarities in criminals' modus operandi and discerning unusual patterns of keywords in financial transactions, to focus investigation in areas where it is more likely to be productive.
While there is a lot of new mathematics and science involved, he says the interfaces must hide the complexity, so law-enforcement officers untutored in the scientific basis of the tools can use them in productive ways. Interfaces should resemble the tools and data formats users are already used to in their work, says Thomas. "Law-enforcement officers like to work with lists" for example.
The tools are not simply concerned with one person's analysis of sets of data, but with the mechanism of collaboration and discourse among a team of investigators suggesting and testing different ideas in a process of discovery.
This aspect includes attention to the physical form of the display hardware.
"We find tables particularly beneficial," Thomas says; data representations are displayed on a flat horizontal surface so several people can examine and manipulate the display concurrently while talking with one another face-to-face, with eye contact, facial expression and body language as another stream of data.
Trying out new visual encoding of information and new styles of interaction often produces a purely pragmatic sense of success: "There are times when we feel 'this works kinda good', but then we have to build the science behind it," he says.
Thomas made an attempt at establishing a Kiwi cooperative venture two years ago, through Mark Billinghurst at HITlab, but nothing came of the effort.
Local funding was the problem, according to Billinghurst, raising the prospect of a successful venture this time, if US funding proves possible.
"We hosted Jim Thomas in New Zealand a couple of years ago and we were and are very keen to collaborate with him," says Billinghurst, by email.
"The challenge at the time was that there seemed be no funding available from government to support such a collaboration. We approached several funding agencies (FRST, NZTE, MED, and so on) but there did not seem to be any funding instruments to provide some research collaboration."
The HIT Lab NZ does not have enough of its own resources to seriously collaborate without external research funding support, he says.
"His field of research is very relevant to what we are interested in doing and we would love to collaborate with him more closely. We believe we have a lot of technology that could be used to do some interesting research in Visual Analytics.
"Hopefully there will be some funding opportunities that will allow this to happen in the future," Billinghurst says.