I know that there is a separate thread for project ideas. But, there is also something called department pride. :-P So, I thought it would be good if I post my project ideas here. Most of the projects would be programming which might require a considerable amount of research. There would be a few projects which are basic programming but might require some understanding of rudimentary concepts. Please feel free to email me in case you have any doubts. But I am sure to not answer any questions that Google can give out. So, please make sure that you have thoroughly checked the internet for answers before pinging me.

So, here we go...

1] Mapreduce is a programming paradigm which was invented at Google. Hadoop is a Java based open source framework for map reduce used by many projects, notably Yahoo. CUDA is a framework for using GPUs for processing. CUDA is specially suited for parallel tasks and also supports Mapreduce. Would it be possible for clusters running Hadoop to offload some of their tasks to the GPU if they are available?

Any task that is assigned to a Hadoop client must utilize both the CPU and GPU (using CUDA if available). This would be an interesting possibility to explore.

2] Most websites support the resume function for downloads today. Say that you are on a slow connection but you are downloading a large file using HTTP (and not P2P). Would it be possible for you to team up with friends who want to share their bandwidth? The file can be downloaded in different parts on different computer and assembled together later. Challenges include proper co-ordination, error checking and reassembly.

If done, this would be a very useful project too. I had given this idea to Bala of CS 2009 some time back. So, please get in touch with him if you want to explore this idea further.

That's all for now. Keep visiting this thread. I will post ideas here from time to time.

I am just leaving a problem statement here. Let us see whether anybody could come up with solutions to the problem. I have two solutions to the problem, which I will post later. I will leave the problem statement here for you to think on it.

Many a time on the internet, we try and find the pros and cons of a particular topic/issue/product. It is not very easy. The idea is to provide a simplified solution to this. Below is a dilemma that was faced recently…

“I was reading about the Semantic Web. I hit the article about Metacrap. After reading the article, I wanted to locate other articles which supported or spoke against the points which are listed in the article. Other articles could disagree on a high-level premise or could disagree with a particular point. There was no way in which I could do this. I used Google to fetch all articles which referenced this article. But again, there was no way in which I could find whether an article spoke for or against Metacrap. I was specifically looking for an article which disagreed.” - W

This was my post. I did not notice that alfredd was logged in at my system.

Regards,
W.

Note: You might want to read the last two paragraphs at least twice.

If you are just into Web Programming, this is a "fun" project to do. It is called D-Index.

Functional Specifications
It is like a sensitivity index for dowry, specially to be used for people in the IT industry. The user, who is already married needs to enter the following details (everything at the time of marriage):

> Year of Marriage
> Role (from a predefined set)
> Years in the IT field
> State (from a predefined set)
> Religion
> Caste
> Sub Caste
> Sub Sub Caste (optional)
> Dowry Received

No other data should be captured. Complete anonymity must be maintained.

A person who is planning to get married must be able to pull up data based on different parameters to see the range of dowry he can get. The web application must also be able to show the trends over the years (for different parameters) using charts.

Technical Specifications
You can use this project as a platform to learn some web technology. Feel free to use anything.

However, LAMP would be preferred. You can replace MySQL with PostGRE if needed. Any Debian based distro would be preferred. And for the charts, you can use the Google Chart API if you do not want to build your own charting library (this can actually be done as a separate project). Other open source charting libraries are also available.

The entire application must be installable using an installation script. The installation script must automatically download unavailable packages, configure the packages and deploy the application automatically.

Important Stuff!!
DOWRY IS EVIL!! I am a strictly anti-dowry person. You can find my views on dowry here.

Sick and tired of successive dowry related issues, including one death, at my work place, the supposed IT community, I came up with this satirical project to poke humour at the practice and the people. Anyway, it would be a simple learning opportunity.