Computer science project advice


(Theodore Gregory) #1

Hey all you lovely community folk. Bit of an odd question here but I’ll try and keep it as brief as possible.

I’ve my final year of my computer science degree fast approaching, and with that comes my dissertation to count for half of this year. I’ve slowly devised a couple of ideas, but none quite feel fleshy enough/useful enough to merit what I hope will get me a good mark :crossed_fingers: I was hoping that with everyone’s wonderful brains on this forum that someone may have something they may want made, or perhaps could help me expand/refine the ideas I’ve already got.

The main things I’m looking for in the project are:

  • Development heavy, not research based. Very keen to seize this opportunity to strengthen my dev skills. Only idea 1 is like this so it’s currently my favourite, despite being a tad pointless and perhaps not beefy enough for a top grade?
  • Freetrade uses a lot of cool technologies (https://stackshare.io/freetrade), and it’s these that I want to focus on using, especially to up my android know how.

The following are my current ideas:

  1. Ordering the pages of an unordered report/research paper based on interpretation and context of opening and ending sentences, where there are no page numbers. Extendable to an app with scanning pages in and performing ocr. (Idea from a scene in The Post)
  2. Work-mode notification manager. Delaying notifications and applying ml to determine those that are important/that you care about. Presenting unimportant in batches.
  3. Modular system to piece together, and perhaps destroy, your online footprint. Or instead of modular, it would learn how forum sites look and work etc and do some machine learning to be able to delete your data on any site.
  4. Flip side: System where you give everything you know about someone and it attempts to track them down and accumulate their entire online profile.

Anyhow, apologies for the long post, but thought it’d be really useful to get any ideas from you wonderful people. Really looking to get my teeth into something chunky that I’ll be proud to have made.

Thanks all and happy tradin’ :rocket:

Edit:
@Viktor gave me an idea relating to number 2, whereby I could attempt to devise an algorithm to prioritise his Slack notifications, altering him only the most important, which could be nifty


#2

Viktor’s idea is good and in my opinion getting a grounding in ML is the way to go these days.

Another ML idea might be to detect bots and fake accounts on social networks or detect scams on Twitter, look at replies to legit crypto posts.


(Alex Sherwood) #3

Yeah idea 2 is my favourite too. Pretty straightforward on the face of it for consumer apps but lots of interesting exceptions to manage.

Slack’s cool because they’ve already got the foundation - what channels does the user consider important + highlight words & mentions - but they’re not using that stuff to filter notifications. And that’s why I can’t switch on notifications for Slack on my Apple Watch :cry: I really like what Google’s been doing in this area so far, particularly with Inbox.

Anyway, that’s enough of my rambling, good luck with your project! :slight_smile:


(Theodore Gregory) #4

@saf I think you’re right re the importance of getting a good grounding of ML, certainly would be a useful skills for the future. Like the idea of fake account detection too :ok_hand:

@alexs it’s certainly hugely frustrating when you’re being bombarded with non-useful notifications, and think you’re right perhaps it does have some potential for growth.

Thanks both of you, think I may call summer travels to an early half to try and get my head into the ML world ahead of term. Realised earlier this year I’d been foolish to have not chosen those modules. Nothing like a big project to make up for lost time! :metal: