Project Ideas

phunks

Hello,

I am a final year CS student who is really struggling to come up with an idea for my final year project. A lot of my lecturers are encouraging to develop on Android but I have already made an app in the past and would like to try something new.

I have been hoping to develop a full stack web application using some sort of javascript front end that will tie into a server/database. My thinking is this will prepare me for industry a lot more than android development.

My thoughts so far would be to pull some data down such as website hits, prices, twitter sentiment do some clever machine learning with it and present it in a nice shiny front end however I cant for the life of me think of something.

If anyone has any ideas or can point me in the right direction would be a huge help.

Thank you

Sheeps

Why don't you tie the two together? Come up with a simple idea for a web app (think simple ecommerce or social network) that you can deploy on an application server and expose the back end and service layer through a web service api for the app. Very simple application architecture.

Browser   -> web app   \
                         -------> service layer ----------> dao layer --------> Database
App     -> web service  /

Evil Phil

How about a final year project suggestion tool? No, seriously.

You could make it work in a very defined area of computer science to keep scope narrow, and have people enter loose terms that would then search abstracts of peer reviewed journals looking for relevant articles. Aggregate these into search results and suggest a project or projects that could use those same results as supporting references. Web API with a nifty little android or HTML5 interface on the front end and Robert is your mother's sibling of a male persuasion.

Jim2007

Well I don't know if it would be an interesting CS project, but...

If you could come up with a tool or methodology to reliably anonymise financial databases. I get asked if there is such a tool two or three times a week and the answer is always the same: No, there are a few candidates but they usually die when used in the real world. It is not as easy as it sounds - you need to anonymise, all names, addresses, account numbers etc, but you have to do it in such a way that you preserve the relationships defined in the database. You also have to have a way to allow the owner to reverse the process or map the anonymised data to the real thing should they need to.

It is as dull as ditch water, but would be of interest to employers in the financial sector.

kyote00

something about the internet of things.

arduino + node.js + cloud db

phunks

Sheeps wrote: »

Why don't you tie the two together? Come up with a simple idea for a web app (think simple ecommerce or social network) that you can deploy on an application server and expose the back end and service layer through a web service api for the app. Very simple application architecture.

Browser   -> web app   \
                         -------> service layer ----------> dao layer --------> Database
App     -> web service  /

Yeah this type of architecture sounds perfect. Its something I have never actually implemented in college and would be beneficial when looking for a job.

phunks

Thanks everyone for the replies helping getting the creative juices flowing.

Jim2007 wrote: »

If you could come up with a tool or methodology to reliably anonymise financial databases. I get asked if there is such a tool two or three times a week and the answer is always the same: No, there are a few candidates but they usually die when used in the real world. It is not as easy as it sounds - you need to anonymise, all names, addresses, account numbers etc, but you have to do it in such a way that you preserve the relationships defined in the database. You also have to have a way to allow the owner to reverse the process or map the anonymised data to the real thing should they need to.

Would that not be pushing into the realms of cryptography a bit more with keys to encrypt and decode it? Maybe I am looking at this the wrong way but other than a simple cypher this encryption would not be of huge interest to me.

Evil Phil wrote: »

How about a final year project suggestion tool? No, seriously.

This is actually a great idea covering IR in a module I am doing so it is certainly something I could expand upon.

Evil Phil

Cool. What I was thinking was pick a subset of 5 - 10 journals and build a prototype using them.

fergalr

Jim2007 wrote: »

Well I don't know if it would be an interesting CS project, but...

If you could come up with a tool or methodology to reliably anonymise financial databases. I get asked if there is such a tool two or three times a week and the answer is always the same: No, there are a few candidates but they usually die when used in the real world. It is not as easy as it sounds - you need to anonymise, all names, addresses, account numbers etc, but you have to do it in such a way that you preserve the relationships defined in the database. You also have to have a way to allow the owner to reverse the process or map the anonymised data to the real thing should they need to.

It is as dull as ditch water, but would be of interest to employers in the financial sector.

There's two ways to go about that problem:


1) The easy way: Transform the dataset by mapping each account number to a new number, then distribute the dataset but without the mapping (which is kept secret). This way all the relationships are preserved for analysis, and only the owner can reverse the mapping*.

(* As long as everyone promises not to use the relationships present, in a combination with other real world data, to deduce the mapping.)


2) The impossible way: This other way, which preserves the relationships, but stops people attacking the anonymity, is not possible.

magooly

phunks wrote: »

Hello,

I am a final year CS student who is really struggling to come up with an idea for my final year project. A lot of my lecturers are encouraging to develop on Android but I have already made an app in the past and would like to try something new.

I have been hoping to develop a full stack web application using some sort of javascript front end that will tie into a server/database. My thinking is this will prepare me for industry a lot more than android development.

My thoughts so far would be to pull some data down such as website hits, prices, twitter sentiment do some clever machine learning with it and present it in a nice shiny front end however I cant for the life of me think of something.

If anyone has any ideas or can point me in the right direction would be a huge help.

Thank you

Try something like www.eirwig.com using twitters streaming api and google maps api