Reputation: 45
I am working on a small price/product comparison website, it's a niche website related to laptops and tablets, built in php.
My problem/question is how to do the following :"matching similar products from different merchants". I mean, when the product has EAN/ISBN, a simple %LIKE% can do it. But the datafeeds I get have a lot of products missing the ean or any other unique ID. How do price comparison websites deal with this?
I'm thinking of searching for string similarity between products names, but I don't want to match : Acer iconia tab a500
and acer iconia tab a500 case as similar products. any ideas?
Thank you !
Upvotes: 3
Views: 3596
Reputation: 820
I faced a similar problem. There are different solutions.
You can find similar items with some search technology (full text search engines can be helpful) or by using some data mining methods (have a look at named entity recognition for recognizing brand, model, color..etc. and especially machine learning methods for text mining). Latter can be much more accurate if you do it well.
In both methods, then you can use some additional fuzzy logic for string comparison of the words that can be written in different ways. and general predefined rules to eliminate wrong items. for example, considering the prices can differentiate an item and its accessories although they have very similar titles.
Upvotes: 0
Reputation: 1834
To implement the comparison you have to put some tags for the products.And when a person search for a product, list the other products which have the same tag.
eg: for a laptop tags are like laptop, acer, 14", 500$(price), etc.
So when someone search laptop, list all the laptops. so that he can choose 2 of them. and make comparison.
Hope you got the concept.
Upvotes: 1