Can Distributional Semantic Models identify substitutes and complements in shopping cart data?Tools Gaur, Divya (2018) Can Distributional Semantic Models identify substitutes and complements in shopping cart data? [Dissertation (University of Nottingham only)]
AbstractSubstitutive and Complementary relationships are natural attributes behind products and crucial to identify in order to understand a product’s functional value, in turn consumer needs. Recent research claimed that distributional semantic models can effectively identify and retrieve substitutes and complements for any product in the shopping cart data, but falls short in providing any empirical evaluation. The aim of this dissertation was to test the validity of this hypothesis. Different types of DSMs were implemented and evaluated on shopping cart data for identifying substitutes. The results revealed unexpectedly poor performance of all the DSMs in retrieving appropriate substitutes and provided strong evidence of model’s incapability to capture these relations. Subjected to further evaluation, this study indicated a need for more intricate models to identify complex relationships of substitutes and complements.
Actions (Archive Staff Only)
|
Tools
Tools
![[img]](/style/images/fileicons/application_pdf.png)
