VB.NET research ideas

[jamesclarke112]

I do know that this is probablly the wrong forums but... Hello all. I am just about to start writing a thesis, however I haven't decided on the topic yet. Seeing as most people in here are pretty clued up on VB.NET (and possibly SQL), do any of you know a good topic regrading the 2 that would be pushing the limits of what is already known, i.e. a good research paper topic?

Thanks

[Bulldog]

Hi James,

I'm not sure what 'level' you're looking for, so my apologies if my suggestions below are under or over the scope you're looking for or maybe a little too complicated for the time you have available. But here's a couple of topics which may have some 'meat' in them, both are kind of related.

(a) Using VB (or something) to make an inference engine. What I mean by that is, take a collection of graphics shapes for example, all scattered about, different sizes etc. An inference engine would detect whether such a pile was the same as another apparently similar pile. The output being a conclusion of similarity, including which shapes match, which dont, which almost match but are a different size, how many edges/angles match, which mathc but are a different color etc. I used a program called "Vera" a long time ago (LISP based) which did something similar. This has many uses in comparing different representations of the same thing and if you could make a VB Class for it, excellent! This topic borders on AI in fact. Comparing text is another application of the same thing. You could make some nifty graphics and text demos from this.

The nice thing about this is that if it can be done, your thesis could be "Algorithms that permit inference in VB", if you fail, the title could be "A critique of VB for the development of an inference engine"!

(b) A second topic would be using VB+SQL to make a different type of relational database. I have recently been writing an app that allows engineers to program up their test equipment (over GPIB) and subjecting a product to sequences of various electrical tests. The info gathered is collected by my app and I included support to plot graphs, write out tables for reports etc. This all works great.

But... I was asked whether I could do 'data mining' for all of the results collected over time and identify which sets of data had very 'similar' results. Dont ask me to explain what this means in practice, but as I understand it the manufacturing process 'wanders', sometimes it produces stuff perfectly, sometimes the end product is a bit faster or a bit slower, sometimes just one measured parameter ends up off target. I guess this is all to do with the mechanical, electrical or thermal drift of the machines that are used.

What the guys are asking me to do is trawl the data, compare the results for each product tested against all of the others and to pick out trends, spot 'lone wolves' etc. So I would like to be able to do something like, "If Dataset(a) IsLike Dataset(b) ToWithin 10% Then MakeSimilarGroup". I guess all kinds of statistics could be generated. Note that the task is NOT to compare a products results against a fixed specification, that's easy. The task is to compare each products results against all of the other products results and identify any apparent groups in the population. Something that might help here is a stats thing called the "Fisher Test" and another called the "Student Test" which can apparently measure equivalence in data. There are commercial packages that do mining, but making a VB class that could do this would be awesome.

Anyway, a couple of ideas.

[AIS4U]

james....:

Unless you are a computer science major, writing about a programming language would result in a pretty boring thesis, in my opinion. Also, there are 100's and probably 1,000's of book that have been written about .Net. Most such books have to do with the use of some particular aspect of the .Net framework/languages and that is about the limit of what can be written about it.

If you had the access to and the ability to, meet with several members of the development team that created the .Net framework and languages, then maybe you could "get under the hood" so to speak and bring out something new. Other than that, I really don't see what you could write about, that would actually qualify as a thesis.

Also, .Net is really too new to have much of any history (I think there have been three versions, 2002, 2003, and 2005 and maybe 2002 isn't counted).

If you had access to the source code (which of course you could never get) then you might be able to really delve into the subject. I understand that it was mainly written in the relatively new C# language, but that doesn't mean much from a thesis point of view.

I think a much more interesting subject might be to compare and contrast the development of the PC and software from the "chicken and the egg theory".

There is a lot of discussion about which comes first the more powerful PC's or the more powerful software. Some people feel that there is a conspiracy between the PC makers and the software developers.

Window's new Vista is a good example. A substantial number of the PC's out there are not capable of running Vista, at least not in their current state. So the PC makers are jumping for joy and offering a free evaluation of your PC to see if it will run Vista. Which, of course, is just a gimmick to get you into their store so they can sell you a new PC.

And then, down the road a few months/years the PC makers will come out with bigger and faster PC's so the software developers will have to write bigger and more bloated software to take advantage of the new PC capacity.

Anyway, guess I better end this before my "short comment" turns into a thesis.

Good Luck