The professor who helps people find things

Helping people find things is how one of the country's leading computer scientists describes what he does for a living. He must do this very well given that it helped him win a Royal Irish Academy Gold Medal.

"I build information systems that help people to find what they are looking for," says Prof Alan Smeaton, professor of computing at Dublin City University. "It wasn't always planned that way but this is what I realise that I do."

He doesn’t do it on his own. Smeaton is a founding director of the Insight Centre for Data Analytics, a centre that involves the work of 400 researchers based in a number of institutions, including DCU.

Insight has 52 companies involved in its work, which focuses on what is commonly known as Big Data. It also happens to be the largest Science Foundation Ireland research centre in terms of money invested: the State and industry stumped up a five-year budget of €85 million.

Areas under research

He has been helping people find things for a long time. Twenty-five years ago he was working on forming connections between documents in hypertext and how he could use topology to find things more quickly.

The range of areas under research has telescoped, however, and it is not really just about data: it is more about what you can get from it.

Subjects under study include memory retrieval in dementia patients, neuroscience, computer learning, journalism, media, health and personal sensing. All can be enhanced by making more efficient use of data sources, but the key is deriving value from information stores.

“What makes Insight different is there is such a wide range of application areas. That is our biggest challenge, given we could be pulled in every direction,” Prof Smeaton says.

Some of the work is intriguing, for example research into memory retrieval systems that could help Alzheimer’s and dementia patients. It involves neuroscience and how the brain works, but translates back to everyday experiences.

The work involves accumulating the person’s memories and activities and day-to-day things, he says. Using wearable cameras and building up a digital footprint, the information about the person and their environment accumulates.

“These are people’s own personal memories from today or yesterday or last week.”

If a person struggles to hold on to what they should be doing next or where they were or should be, they can tap into the data stores to help them re-establish a context. “It is useful for those who don’t have good memory: Alzheimer’s and dementia patients and after brain injuries. It helps them with short-term memory.”

The work and approach is based on reminiscence therapy, whereby a single memory can trigger a cascade of memories.

He likens it to how he reacts to the smell of Christmas cake. It releases a string of memories involving his mother making the cake and how it tasted.

“It acts as a kind of personal reminiscence, a Proustian moment if you have something embedded in the past,” he says. French author Marcel Proust wrote about this “involuntary moment” when memories unexpectedly arise that immediately release others.

“We have this huge capacity to remember things, but a poor ability to navigate our memories, for example ‘Where did I put my keys?’ We are trying to understand why people remember some things and not others. If we had some insight into this then it might help with memory difficulties. ”

Automatic captioning

Another area under study is teaching a computer to describe the contents of pictures and videos, something that could help support automatic captioning.

This would offer a major benefit given the volume of images and videos that are fed into social media or captured by systems. The object would be for the computer to view the image and be able to put together a short caption on what is visible.

They are making some progress: systems are now able to spot a traffic cone with about 60 per cent accuracy, recognise a road with 75 per cent accuracy and identify a truck with 80 per cent accuracy.

“If you combine things you get an ensemble effect that gives you interdependencies, and interdependencies can make good descriptors for digital content,” he says.

If the system got good enough it could be used to discard unsuitable material, for example badly focused pictures or poorly shot videos. “If you analyse the content and describe what is going on, you then have a technology to help reduce volumes and only save the best material.”

INFORMATION OVERLOAD: BIG DATA AND WHO OWNS IT
You might not know it but as you drive along the M50 you are contributing important data used to control automated traffic-management systems.

As you progress down the road, your mobile phone switches from base station to base station, and this fixes your location. This also tells the system how quickly – or slowly – your phone is linking to stations along the route.

The phones carried by other drivers around you are doing the same thing, allowing the system to detect when congestion slows things down. When things really slow, it causes the overhead electronic signs to switch on their warning that delays are likely due to heavy traffic.

The question is, who owns this information: you or the National Roads Authority? Information is captured from us all the time, such as when we are using a computer, driving or walking, and where we go and how long we remain there.

We surrender the data but who has ownership of it? Data harvesting has raced ahead of controls to regulate it. “We are not in a good place with the ownership of personal data,” says Prof Alan Smeaton.

He expresses his delight about receiving the 2015 Royal Irish Academy Gold Medal for science.

“It was a super occasion for me and my family. There was a great buzz and I was very happy.”

Prof Philip R Lane, governor of the Central Bank, also received a Gold Medal, for his research into international trade and the world economy.