Categorizing products?

deep

Hello,

This is a hard one... i have been thinking for days and now I need help 🙂

I need a script that can categorize new products automatically just like pricegrabber.com does. When you search for a product on pricegrabber.com, lets say we search for "sneakers" we get a list of sneakers, all different brands. Clicking on one will list all the shops selling the sneakers you klicked on.

Now, to the problem 🙂 :

In many cases shops dont use the Exact same name on the Exact same product. I noticed this while browsing through different categories... In some cases shops use different names on the same product... and this gives me a problem.

Different names would by default make my script think that these are who compleatly different products and not list them as one splitting up the two shops that sell these products and making it hard to compare prices 🙁

Any idea how to solve this? If you know of any good scripts, examples, demos anything you can think of I would strongly appreciate it!

Thanks.

axsdeny

I would think that your best bet is to set up a categorization scheme for the client. When a new product is entered, they can choose what category(ies) the product belongs to. If that isn't an option, then you still need to create arrays of possible products that fall in a category. If you are going to do that, you may as well create another table that stores those possible values and assign them a keyword.

Not sure if this makes sense, but normalizing your keywords, and assigning all item types to a keyword (or more) would make the query much easier and cleaner than doing pure text comparison processing.

deep

Hello,

Thanks for the reply. I do use normalization to make out in which category a product should go into, I even let the user write the category in the productfile that he submits to make sure the product is listen in the right category.

Problem comes when I want to group the products. Example:

Product 1: Blade Trinity
Product 2: Blade Trinity (2-disc) Extended version

Product 1 and 2 are exactelly the same product but from 2 different shops using different names and I want to group the products... In this case I dont know how to gruop them and be sure that its the same product... look at this example:

Product 3: Warcraft 3
Product 4: Warcraft 3 Expansion

In this case the products almost have the same name But are two compleatly different products... see my dilema? 🙁

Whats bugging me is that others seam to have solved this somehow... pricegrabber.com is one example. They maybe have people that manually make sure the products are grouped right... The best way would to do this automatically, it saves time and money...

Any ideas??

Thanks!

_RL_Ian

You have to find some sort of common identifier to determine if the product is the same or not. Is the name the only thing you have to work with? Because you could say, if the name is different but the SKU number (product number) is the same, then we know that the products are the same even though the names of them are different.

deep

Yea, I did think about using a product id number but there is a problem there too... in my case the merchant when listing his products needs to know each unique product number (on my site) and add it to the productfile (that contains all the products he wants to show on my site) for each product.

That would be a massive job if the merchant has more then 100 files listed in his productfile... :bemused:

Still not sure how others have solved this problem... Sure I could parse the name and normalize it but that wouldnt give a 100% grouping... 🙁

Alternativly build a controlpanel where the merchant can see if a product can't be matched and then let him manually choose the group...

ideas?

_RL_Ian

deep wrote:
Yea, I did think about using a product id number but there is a problem there too... in my case the merchant when listing his products needs to know each unique product number (on my site) and add it to the productfile (that contains all the products he wants to show on my site) for each product.

That would be a massive job if the merchant has more then 100 files listed in his productfile... :bemused:

Still not sure how others have solved this problem... Sure I could parse the name and normalize it but that wouldnt give a 100% grouping... 🙁

Alternativly build a controlpanel where the merchant can see if a product can't be matched and then let him manually choose the group...

ideas?

Sounds like a decent scenerio. Maybe when they enter the product, you could do check to see if any "similar products exist". If so, you could list all the similar products. Then ask the user, "Did you mean to select one of these already existing products?" If they choose one, then you're set, else their new product will get entered.