search

dokato / healthcareonlinesources

WHAT - NHS Hackathon project
MIT License
2 stars 2 forks source link

New version of shiny app #4

Closed dokato closed 4 years ago

dokato commented 4 years ago

I cleaned a bit Shiny part of the repo and "beautified" the website. It's deployed now, but we can always come back to previous state: https://dokato.shinyapps.io/WHAT.

One critical change: healthcare experts now need to provide a password if they want their record to be saved. I know... it's not a perfect solution, but provides some sort of security layer. If you want to circulate that round your friends with expertise I'm happy to send you the password via some private channel.

BTW there's a new password for mongo too if you need it...

@KatrionaGoldmann @mbytc1 please LMK WDYT in comments or feel free to change this branch.

Known bugs:

KatrionaGoldmann commented 4 years ago

This looks great @dokato ! I particularly like the motivation/info page. I will try to have a closer look later this week but happy to merge in the mean time.

I do have a couple of questions/point for the future:

  1. Where do you envisage the password coming from? Do we need to start setting up accounts?

  2. It might also be useful to allow people to add entire domains as well as certain pages. What do you think? For example the entire NHS website can have a domain score, whereas nhs.uk/conditions/rheumatoid-arthritis/ can also have a more specific individual page score. This will mean that not every individual site on NHS, BUPA, Sun, Daily Mail etc. will need to be added for a flag to appear in the extension.

dokato commented 4 years ago

Great, I'm glad that you like it ;)

  1. So far, a provisional approach is to send an email to WHAThealthonline@gmail.com and request a password. There's a pop-up now that tells you that. I know... it's far from ideal, but keeping track of user registration would imply that we need to provide some automatic verification of the pros. We can brainstorm about it later. What I think is the most important now is to have some clear criteria of what we mean by Healthcare expert.

  2. Yes, so that's something we've been discussing with Tom and I think that this should be easily achievable in the current pipeline. So far JS code works like this: it searches in the database for the first regexp match with a record top to bottom. We could add to whatouput2json.R some heuristics, eg. sorting from longest links to shortest and that should give you score for a specific article, subpage if detected, then a root of the url.

I still have some thoughts on potential enhancements, I'll post them as separate issues soon.