schema

A Python implementation of SCHEMA - An Algorithm for Automated Product Taxonomy Mapping in E-commerce.

Based on the SCHEMA algorithm proposed by Aanen, et al.

Ref: http://disi.unitn.it/~p2p/RelatedWork/Matching/Aanen_eswc_2012.pdf

Author: David Ng [email protected], [email protected]

Required modules

• nltk with wordnet download (see http://www.nltk.org/data.html)
• Levenshtein
• pyxdameraulevenshtein

Example usage

import schema

# create source and candidate paths
source_path     =  schema.Path().add_node(..)
candidate_paths = [schema.Path().add_node(..), ..]

# generate key paths and match
keypathgen = schema.KeyPathGenerator(source_path, candidate_paths)
source_key_path                  = keypathgen.source_key_path()
matched_key_paths, matched_paths = keypathgen.matched_candidate_key_paths()

# rank and print results
ranker = schema.KeyPathRanker()
for i,candidate_key_path in enumerate(matched_key_paths):
    rank = ranker.rank(source_key_path, candidate_key_path)
    print rank, matched_paths[i]