My Blog: Imaginary Friends and Colleagues

23 December 2019

The demonstration website for our fake university needs lot of staff to teach the subjects and manage the business. How else will we attract all those potential student applicants?

For my list of names, I went straight to the British Surnames website and clicked onto their random names generator. I love this website, because, and I quote:

The British Surnames random name generator doesn't just pick names at random, it weights the selection according to the actual probability of finding that name in the population at large. So, for example, it's more likely to give you "John Smith" than "Edmund Sidebottom" even though the latter is a possible result. This makes it more realistic than many random name generators which simply pick randomly from a list and thus tend to overemphasise uncommon names.

It's a good list to start with... A 50/50 gender split, and a nice mix of BAME names in there as well.

From this list I programmatically generated each person's username by concatenating forenames and surnames. I also created an email address in the format [forename].[surname]

Using the list of subjects I created earlier, I then allocated most of the names to teach a subject, with some topics having more than one lecturer, senior lecturer, and/or a subject leader... and allocated job titles accordingly. The remaining names will be allocated Professional Services (administration) job titles, once the Groups and Teams are created.

From the names, I also populated their honorific titles and imagine appropriate post-nominal letters.


Adding photographs was a much trickier proposition. I couldn't just Google random images to use here... that could get me into a lot of trouble! I also couldn't afford to use a stock image library, due to the high costs involved. Instead, what I needed was a source of good-quality profile photos that weren't subject to copyright or where the subject wasn't identifiable.

After much searching, I eventually came across a website called that fit the bill. It uses a Generative Adversarial Network (GAN) computer program to mash together the components of multiple faces sourced from the internet. Using the Flickr-Faces-HQ Dataset (Creative Commons BY-NC-SA 4.0 licensed), each resulting composite face is totally unique.

It's not without its flaws. It doesn't like hats, earrings, or anything where the source images had something covering the face (hands, microphones, etc). This meant reloading the generator multiple times in order to get a single believable photo out of it.

Eventually I succeeded in collecting 100+ images of sufficient quality, which I then converted to grey-scale, watermarked, and joined with my name dataset:

Jamine AstorRussell WiltshireTimothy MartinDarren HelliwellCarl ChungChristina BoothJulia PoonAsa BaudinMolly RichardsGlen JeffriesMark JacksonAmit PatelGina WilkesPatricia TylerSia KohliJohn McMullanNicole SmithPatricia IngramRobert HandleyBarbara BarkerMarcia AllenMaxine HillJane MorganAnne NeildCraig McDavidMandy PaulEileen EarlsMark ReynoldsonGeorgina WattsPaulette ConnorPeter MorrisSarabjeet SethiAndrew HawkinJane HarrisNatalie CromieErdine KingDavid GoochAndrew GreenDavid PoppletonJosephine GibsonGordon LockhartJohn WallAnita DunnAnthony YoungJohn PetrieJonathan QuinnBrian CookTina FoxWilliam DoodySonya LanfearJoanne WinsonDavid GreenJanet BartonColin BloomAndrew EdmondsDean SheehanGeorgina GardnerNicholas PritchardEmma NunnCaroline AventMichelle WellsMark HinksMohammed AshrafAudrey HillDavid ReidNiels PetersenAmanda CharnleyElizabeth ToppsStuart HendersonIan BrackenburyJulie BennettIan MathiasKitta McBrideTimothy SmithJune CrookNorman MacIverHelen BullivantMichael MilliganAmanda ColledgeMohson RasoolSamantha BennettRichard OlssonHayley ClarkRobert HardySylvie MundellDarren HeadleyMark AspinwallHettie TimmsPetrina OliverAlan WhittakerDavid DaviesSusan MacKenzieDeborah ThurstonGeoffrey AshworthValerie SoperSimon ArmitageDavid SessionsSamantha FletcherVictoria Mulligan

Transferring the Data

With this compiled list of people I now have the values to use in the following datatypes:

comments powered by Disqus