I'm planning on building a feature for an e-commerce platform I developed in Java to display related products in much the same way Amazon does. There are a few different metrics for relating products that I want to explore.
- Purchase history (purchased at the same time)
- Related by family/type (similar product classifications)
- Intentionally related (boosting results; "Buy this!")
While I would probably be able to develop my own datamining library, it wouldn't be very portable and I dare say it wouldn't be very good either.
There are several packages out there for doing this sort of thing but I don't feel like I am in a position to evaluate which package or solution would work best for me. Any input anecdotal or from personal experience would be greatly appreciated.
Note: I've tagged this as Neural networking because of a python talk I was at where a neural-like-network was used for datamining, I'm not convinced a neural network is the best choice for this job.