Yes Berlin which I am very excited about, even more so as I am attending the AWS summit there. This will be an opportunity to get inspiration for Smap 2.0, the second generation of Smap software. I will be updating this post regularly up until the end of the Summit to record some of my thoughts on where the Smap architecture is headed.
Smap 2.0
- A new system that will work with the existing Mobile Data Collection Servers and Phones
- Runs in cloud
- React.js front ends
- GraphQL services where possible
- Serverless back ends
- Single sign on across mobile devices and servers
- Will be used to deliver notifications and process management capability to the existing servers
- Manage notifications to and from mobile phones. For example notifying a phone that a new version of a survey is available
- Plus much more
AWS Summit
None of the following comments represent design decisions for Smap 2.0 they just take the concepts from a topic and construct some potential scenarios.
Summit Demo: MXNet on the edge. How to build a deep learning model that works on edge devices
A Smap server could be considered an edge device. The data collected could feed a machine learning application to provide real time insight into data and quality assurance.
The data processed by the machine learning could include the time taken to answer each question in order to identify poorly complete surveys. The collection of question timing data is currently being added to ODKCollect and fieldTask in a joint Smap / ODK project.
The demo at the AWS summit is going to use a Raspberry Pi as the edge device to recognise art objects. A mobile phone running fieldTask could also be an edge collection device interacting with Smap 2.0. This is how the phone will get notifications to refresh forms etc.
MXNet is a neural network application from Apache which can be trained in order to recognise patterns.
Summit Demo: Utilize Powerful Visual Search and Discovery Capabilities with Amazon Rekognition
Rekognition is one of several AI services provided by AWS
- Rekognition: Image recognition
- Polly. Text to speech
- Lex. Speech and text recognition and language understanding
It looks like there are quite a few sessions on AI at the summit and adding AI capabilities to Smap is something we are considering so this could be useful.