Towards Pipeline Construction for Product Matching Task
Olga Cherednichenko, Oksana Ivashchenko, Maryna Vovk · 2022
Online shopping services are commonplace for people around the world for acquisitions. The product offers on websites are mostly given with some text descriptions and are often accompanied by specification tables. The same product can be found in many different e-shops, but information about the offers of this product can vary greatly on different electronic platforms. The task of product description categorization is not new, but the solution heavily depends on the data set of product descriptions. The pipeline is the scheme of step-by-step data processing. The aim of the paper is to construct the pipeline, for product description categorization, which should be flexible. The pipeline should allow getting access at any step during data processing and modify it by adding new functions, changing activity order, and implementing particular methods. We propose an approach to construct a flexible pipeline for product matching. The case study is given and experimental results are discussed.