Today’s highlight was going to Engineering.com’s office to learn in person how they have their current technologies fitting together. We know that we will eventually need to link in the data we gather to their Elasticsearch service so that it improves the quality of recommendations made to their users. We learned that they use AWS’s API Gateway to create serverless APIs that their DNN CMS connects to for recommendations.
That API provides a layer of abstraction. The DNN system doesn’t know how it gets recommendations, it just knows how to query it for them. This is good because it means we can build either a separate API linked to by their API or simply augment their API to add in our data. Or, we will simply add our collected data to their Elasticsearch service. Right now, every 15 minutes, they scan for changes in their articles etc and move that data into Elasticsearch so that it’s usable for searching. We may create a tool to periodically inject data about the users, and help them create Elasticsearch queries that use that new data. Either way, everything seems nicely-decoupled and open. It shouldn’t be hard to add in our features.
Progress on the React app “Visual Analysis Tool” (VAT) we’re creating is going well too. I’m moving in the queries that my team mate developed so that we can see how they look in it. Because of the pattern I developed for storing the queries, it’s very fast and easy to add a new query in. The React app gets the info from the server serving the app, so there’s no changes to be made to the client side React app code when a new query is added. We’ll be having our client module code start running on a small part of their site on Tuesday, so that we can begin collecting useful data to play with. That’s when the VAT will start to shine.