Predicting E-commerce Item Sales With Web Environment Temporal Background

In this paper, we study the effect of Web environment temporal background in pre-dicting e-commerce item sales, especially  those in temporary sales. Temporary sales nowadaysare a popular strategy for quickly clearing inventories. For traditional  recommender systems,predicting the sales of an item is done based on its past purchase records. For temporarysales items, however, such records are not available. In order to make recommendation forsuch items, contextual information, such as product descriptions, is usually used. We investi-gate whether temporal background in the Web environment can be additional useful contextualinformation in recommender systems. It is assumed that items consistent with the temporalbackground would have higher demands. We propose a method for representing the temporalbackground using word embeddings of e-commerce activities and social media data, and eval-uate their effect on sales prediction. Through empirical analysis with real-world data, we foundthat temporal background does have positive effects for sales prediction. The findings in thispaper can be conveniently incorporated into future recommender system designs.