machine learning - Identify a matching algorithm -


i new nlp/ml/pattern matching or recognition. wondering best way match different items based on title, description, etc. eg:

if there 3 items:

item 1: title: belkin bluetooth headset usb - abd13432 item 1: description: bluetooth device following specs:  75 w power, 3.5 mm jack, etc item 1: model no: abd13432 item 1: upc code: 000000022221 item 1: product image: <img1>  item 2: title: belkin headset:  item 2: description: device works on rf, , has 2.5 mm jack 25 w power  item 2: model no: 13432 item 2: upc code: 000022022221 item 2: product image: <img1>  item 3: title: belkin headset wireless - abd 13432 item 3: description: world's best headphone item 3: model no: abd-13432  item 3: upc code: 000000022221 item 3: product image: <img1> 

item 1 , item 3 same , item 2 different. upc code great indicator if same item issue seller can input upc code wants. image matching not indicator since seller can input image wants to.

in particular case, model no , upc contributes more weights other features.

when items have different model no or upc, consider semantic similarity short sentences feature learning algorithm.

you might want have this paper. case in product/e-commerce domain, might want build own domain corpus other use general wordnet.


Comments

Popular posts from this blog

c++ - Creating new partition disk winapi -

Android Prevent Bluetooth Pairing Dialog -

VBA function to include CDATA -