Mapping product taxonomies in e-commerce

In this paper we propose SCHEMA, an algorithm that automatically maps heterogeneous product taxonomiesin the domain of e-commerce. SCHEMA employs a custom word sense disambiguation technique,based on the Lesk algorithm, in combination with the semantic lexicon WordNet. For findingcandidate target categories and determining the path-similarity we propose a semantic category matchingalgorithm that takes into account the disambiguation process of a category. The mapping quality score iscalculated using the Damerau-Levenshtein distance and a node-dissimilarity penalty. The performance ofSCHEMA was tested on three real-life datasets and compared to PROMPT and the algorithm proposedby Park & Kim. The comparison shows that SCHEMA improves considerably recall and F1-score, whilemaintaining similar precision.