Opening up activity data in Devon
Dave Barter, CEO, NautoGuide Ltd
As avid consumers of open data, we were very excited to discover the OpenActive standard and understand the potential for discovering and publishing activities to enhance wellbeing. Our company is located in Brixham, Devon and we’ve been working on an initiative to create a hyper-local events and activities portal.
It didn’t take long to create a discovery and consumption engine that allowed us to find OpenActive feeds and download the categorised data. A few more nights of candle burning and we had a portal ready to deliver it to our town. We’ve called it Locaria, don’t ask why as we’ve forgotten. We pressed the button, waited a few hours and “hey presto” a database full of locations, events and other opportunities for increasing the heartbeat. This was connected to a map and we zoomed in on Brixham.
Oh dear! Only a handful of events and sessions anywhere near us, all of them a long way out of town and requiring a traverse of the dreaded “Windy Corner”, not a nice place to be regardless of your mode of transportation. This was frustrating as we work in a shared office building and our noticeboards are a mosaic of yoga classes, pilates posters and other notices attempting to steer us away from the canteen. Why aren’t they in OpenActive? If they were more people would know about them. Could we come up with something to help? The candles were burnt for a few (many) more hours. Some deft python and a lot of integration later we had our activity scanner up and running.
Simply take a photo of a poster, leaflet (or even a screenshot) and then leave it to our various algorithms to process.
First, we extract the text for the image and do a quick positional analysis. Then we look for patterns that match our required data, specifically; a title, description, dates, email addresses, URLs and any QR codes on the paper. Once this is done we create a candidate list and present it back to the user.
Locaria has done a reasonable job of finding things. The description looks close, the postcode is spot on (so we can locate it), the email address is right and, for those with an observant eye, the date is actually correct. As there were two dates mentioned in the text we have a candidate list, one of them is incorrect as it’s today’s date. So we will stick with Monday 10th. The description needs a slight edit and then the event can be submitted. Locaria confirms the changes the user has made.
Now we have some additional training data. The user removed the email address from the description as it had been captured elsewhere. The system can learn over time from these edits and thus improve our description classifier. All that needs to happen now is moderation and the addition of any missing data (we need an activity classification). Following this, the event can be published to OpenActive via an RDPE feed. We’ve adapted this process to work with URLs as well and thus significantly reduced the time and effort needed to get an activity into the standard. We’ll be piloting this in Brixham with the aim of creating a suite of tools to support rapid ingestion and publishing of event data.
Dave Barter is the founder of Nautoguide Ltd. Nautoguide is in the process of creating an entirely open-sourced platform called Locaria for the consumption and distribution of open data for wellbeing. Contact dave@nautoguide.com for more information.
