Bluetooth dosimeter - COVID 19 topical!
Contact tracing is a contentious issue and fraught with technical problems. However, a simple “bluetooth dosimeter” which records unique bluetooth devices you have seen and when (after discarding your own, housemates and neighbours!) is an index of how much social contact you may have had. If we presented this to people and showed them the times of day etc they were in busy places, perhaps even augmented with their location data, this might form the basis for behaviour change to reduce exposure, e.g. avoiding busy shopping or travel times. As with the COVID tracing on iPhone and Android, if attempting to do this bluetooth tracking in an app on a phone it would need to run all the time, or be builtin by Apple and/or Google. Instead we could use a simple arduino (like) device which could be always on in your pocket and simply download the data once a day (when plugged in to recharge) to a platform for visualization. We can supply the hardware if anyone is interested in building the visualization, possibly linking to location data and perhaps even running a few trials.
Back to the future 1. - Can we fix the email reading experience
Develop a set of text processing tools that would allow email threads simply to be presented more akin to social media conversations, eliding all the indented repeated nonsense, legal disclaimers, email headers and other fluff that generally makes reading it more tedious than it needs to be. This project could focus on really doing a knock out job on the text processing or include more UI experience around the overall interface. Personally, I don’t get why any of the major companies have not done this yet!
Back to the future 2. - Personalised RSS reader mobile app
RSS for those who have forgotten it, or never heard of it, provides a feed of updates to websites - for example news sites use RSS feeds allowing you to monitor new news stories and download short abstracts. So subscribing to a set of RRS feeds provides you with a personally curated news feed you can read through a news reader aggregator app - I used feedly. Contrast this with having a social media company decide which news sources it might promote based on possibly unknown profiling, and foreign powers paying to promote certain stories. However, if you subscribe to lots of RSS feeds you are overwhlemed by irrelevant news. So the project is to build an RSS news reader app for mobile (several excellent projects last year did webservices) that allows the creation and then refinement, by monitoring reading habits, of a personal news profile to prioritize news stories. This would then provide personalised news feeds without invading privacy. Enchancements could include reading your own social media output to augment the profile, sharing the profile across multiple of your devices along with which articles are read/unread.
Edge based analytics
Projects like Databox provide a specific plaform for processing your personal data and only sharing limited amounts of derived statistical data to feed large scale data analytics. This idea of federated learning or edge based learning is gaining popularity because it is a) to a greater extent privacy preserving and b) consumes a lot less bandwidth. However, the approach does not need a specialised Databox, but can simply be an app on your phone. The project is to build some interesting distributed edge based analytics system on either the phone or a PC/Mac, whichever is your favourite development environment. As an example, we might wish to do social media sentiment analysis by generating the profile of how a individual user feels around certain topics (e.g. “how annoyed are you by the government’s COVID communications?”) by having an app process the user’s private social media and share the resulting profile to generate the “big picture”.