Discover Qualified Sellers Providing Suitable Goods for Popular Products

  • 周 洵

Student thesis: Master's Thesis


Due to online ecommerce has been booming for a decade As mentioned above shopping on the Internet has become main shopping type for modern people and one of the indispensable part in our life And the concept of “ showrooming ” also rise in recent year Showrooming means consumer examine products in physical store and then make their purchase online because the cheaper price Unfortunately the online product which looks similar may not the same product and the statistics of transactional dispute tells us the main dispute type is still about product quality and product appearance In the situation that we can’t touch the physical product the seller’s quality is an important factor which we should consider More specifically we aim at helping consumers identify which sellers can satisfy consumers’ requirement and have the better quality at the time According to our observation of query logs and product data the search result of e-commerce website still use keyword match technique The observation also showed us that the information in the product description product question and answers between buyer and seller can provide more detail for each consumers to check whether a seller can satisfy consumers’ requirement Therefore we utilize the information in the product to help consumers finding the products which seller can satisfy their requirements The sellers who can satisfy consumers will be regard as candidate sellers And we employ a classifier and five feature to identify whether a candidate seller is a qualified seller The experiment result inspire that the identification rate for need and focus will be lower in the product category with individual requirement At the same time we find the consumer sensitive feature can make better the performance of finding qualified for the query with individual requirement when we use different feature combination Otherwise we classify other categories with simple requirement by all features and it will have the better result to find more qualified seller
Date of Award2016 Aug 24
Original languageEnglish
SupervisorWen-Hsiang Lu (Supervisor)

Cite this