This java project is used to extract structured metadata from the
products collected from different brand websites.
Structured metadata for is simply a set of
key value pairs that describe any given product,
for example - (SLEEVE=SLEEVELESS), (LENGTH=MINI), (MATERIAL=COTTON). This is critical for creating a good recommendation
engine.
Such metadata can be found as properties on the product page or inside the product description
as unstructured text. Some visual items like sleeve and length can also be detected by
analyzing the product image.
Key value pair example directory:
link
product dataset:
link