Information retrieval

[ArmMiner]

Hi all,

I am writing my Master thesis and would like to know some opinions about my task (some hints maybe). So, my project description:

The product information is in excel file. There are thousands of products in the list (columns: name, productId, description, imageUrl...). The first part of the problem is to compare all products to each other (i.e. compare the descriptions). After first comparison of first two products, a mutual string has to be created, like fingerprint, so it would be compared to the third product. So, after each comparison the mutual string is updated. Also, by comparing the products the most similar ones have to be grouped. So, in result we will have grouped products. Last process is attribute extraction. This means from each group the attributes have to be extracted (color, size...).
So, I'm thinking about the ways to solve this problem. Shall I use some tools or write programme in java? Any ideas or hints?

Thanks!

Best regards
Armen

[editor]

You may want to check some tools for recommendations and associations.
Also Java is not a good language for such tasks - try to use python.

There are lots of python packages for machine learning, for example
see

http://www.kdnuggets.com/2012/11/best-python-modules-for-data-mining.html