Connecting the dots…

The idea was to put togeth­er a work­flow con­nect­ing all the cloud ser­vices we looked at dur­ing these days, basi­cal­ly: Inte­gro­mat, API, Datawrap­per, iOS noti­fi­ca­tions and Google Sheets.

The first step was decide where to get the data. Instead of using a muse­um col­lec­tion (i.e. Rijksmu­se­um), I was attract­ed by the idea of hav­ing my own data

In order to get the data, instead of using an exist­ing API from a muse­um col­lec­tion (i.e. Rijksmu­se­um) I decid­ed to cre­ate my own end point. The idea was to cre­ate a test­ing envi­ron­ment to then trans­late the con­fig­u­ra­tion to a pro­duc­tion site (i.e. the online cat­a­logue

I pre­pared the demo web­site in Dru­pal, a CMS I am used to work with, with a basic data mod­el for art­works, fake con­tent pro­duced with the Dev­el mod­ule, and the Ser­vices mod­ule to gen­er­ate the REST Server. 

The pub­lic end­point is acces­si­ble in JSON for­mat at

The sec­ond step involved extract­ing data from the web­site using the API and write it to a Google Sheet. In order to do that I cre­at­ed a “sce­nario” in Inte­gro­mat, using the fol­low­ing ser­vices: HTTP, Router, Apple iOS, Iter­a­tor and Google Sheets.

This work­flow involves get­ting the data from the API with a HTTP request, iter­a­tive­ly import data into a Google Sheet and send­ing a noti­fi­ca­tion to my phone about new data imported.

Fields writ­ten to the Google sheet are: ID, Title, Typol­o­gy, Author, Image URI, Chronol­o­gy, Cre­ation date.

This data is then processed in Google Sheet via a piv­ot table, in order to quan­ti­fy, for exam­ple, the num­ber of art­works per authors. The result­ing table is loaded into Datawrap­per to dis­play an inter­ac­tive chart, embed­d­a­ble on the web.

Some very short consideration

The most inter­est­ing part of the track for me was to famil­iar­ize with API, JSON and REST ser­vices. The next chal­lenge on this front will be for me to imple­ment a pub­lic (and maybe use­ful) API in a pro­duc­tion envi­ron­ment (can­di­dates are and

Inte­gro­mat proved to be a use­ful ser­vice to cre­ate proof-of-con­cept work­flows, but it has lim­i­ta­tion giv­en by is busi­ness mod­el (1000 requests per month for free).

The most frus­trat­ing tool for me was Datawrap­per, but I guess it is also because I spent very lit­tle time with it. Short­com­ing are that it can­not manip­u­late data, only visu­al­ize them and, most impor­tant, charts do not update even if the back­ground data change 🙁