digital footprints

Introducing recent research and development in urban cognitive analytics to sense population dynamics and people-place interactions


During the past month I have had the opportunity to present some of my research. First, at the conference ‘Smart Cities and Planning: New Urban Agenda, New Urban Analytics’ sponsored by the MacArthur Foundation and held at the Senate House in London on 29th – 30th November 2017. And then to an audience of cognitive and data scientists at an event organised by Brainpool Ltd on 1st December 2017.

The first talk focused on the role of ‘reality’ data in estimating the active population at a place of interest. I use the phrase ‘reality data’ rather than ‘big data’ because, often, the data I’m working with is not at all big. Quite the opposite. It is often sparse, noisy and suspected of containing demographic bias. It is the environmental and behavioural data traces emitted during digitised interactions, whether a mobile phone connecting to a wifi network or its owner sharing a social media update online. Part of my research is evaluating whether such data can be relied upon to better understand people-place interactions and inform urban decisions. And one aspect of that is to estimate the active population at a place of interest. A lot of urban decisions need to consider the size of the population of an area. To find that number out, the most reliable source is administrative data – the census. But the census provides a residential population count. If you wanted to know the population at risk of exposure to an incident, knowing where they live is only useful if the incident happens at a time of day when most people are at home. More recently, an ‘ambient’ population count has been introduced which redistributes the residential population according to land-use patterns, estimating where people work, study and play to produce an average workday population count. But that is still a generalisation. I explored using mobile data readings as a weighting to reveal the dynamic nature of an active population. The talk has been featured on The Bartlett web site at University College London, where I am conducting my research: Sensing Digitally Gets Us Closer to Reality. The slides are embedded below:

The second talk focused on whether or not we can digitally sense situations and contexts as they emerge in near real-time. If you have ever watched Derren Brown performing his tricks of the mind, he has often demonstrated how an environment can be set-up so that people make a choice they think is random but in fact is based on behavioural nudges they have been exposed to throughout the day. For example, he asked a person to think up a tune in their head and promptly guessed what it was. Whilst they thought the choice was just something that popped into their head, it turned out that various members of the production team had hummed/played/whistled the tune throughout the day. One was stood next to him in the lunchtime queue at a sandwich shop. Another was playing the tune busking on the street as he walked by. Another turned up to perform a repair within the office he worked at. And yes, you guessed it, was humming the tune the whole time.

We pick up cues from our surroundings and they affect the decisions we make. I’m curious to find out if we can digitally sense these cues to gain a better understanding of people-place interactions. What makes a place popular or not. What sort of conditions will attract or repel people from an area? We intuitively sense a ‘vibe’ when we enter room. Is it possible for that vibe to be detected digitally? To explore these questions, I have been experimenting with various cognitive algorithms including natural language processing, sentiment and emotion analysis, network analysis and topic modelling. The slides from the Brainpool talk are embedded below and include some of my initial findings and a brief explanation of the model.

I also have the model running online running at the moment, to test whether or not social media can be used to sense emerging situations. You can view it at http://joiningdots.net/spatialmedia. It is focused on central London and is very much a prototype work-in-progress but gives an idea of what I am experimenting with. It extracts contexts from the content of tweets, either geotagged within the spatial range or referencing one of a set of landmarks being monitored. If you’re interested and would like to find out more about my research or have some feedback, please get in touch!

And finally, that Derren Brown demonstration of mind reading…

References


digital footprintsFeatured image: mobile digital footprints on three consecutive Sundays in May 2017 around the Queen Elizabeth Olympic Park, London. Background image courtesy of OpenStreetMap contributors.

Category:
Behaviour, Blog, Data Science, Presentations
Tags:
, , ,

Join the conversation! 2 Comments

  1. This is very interesting research. I recently attended a “Digital Marketing” conference to see what kind of services are available for “sentiment analysis” around an event. I was amazed how at the amount of information that can be available in very near time analysis.

  2. Thanks Dan. Is any info from that conference available online? I’d be interested to know which sentiment algorithms they used and to look into their approaches to near real-time analysis.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: