Predict Purchasing Needs behind User Queries to Recommend Activity-Goods Pairs

  • 余 佩怡

Student thesis: Master's Thesis

Abstract

User need prediction is an important problem as it will largely improve user experience such as saving user efforts In our work we want to predict users’ purchasing needs according to queries about purchasing goods and recommend suitable activity-goods pairs For example a user wants to purchase running shoes and issues the query “running shoes” on a search engine to search goods We surmise that this user purchase running shoes for running a marathon or having a jogging habit Therefore “marathon” and “jogging” are user needs that we predict Moreover for user needs “marathon” and “jogging” we recommend activities such as “marathon competitions” and “jogging competitions” and matching goods such as “sport wristbands” and “towels” to the user We observed blog articles and found that blog articles about purchasing goods are always contain user needs Moreover titles from auction websites are contain user needs as well Therefore we proposed to predict user needs via analyzing queries blog articles and titles from auction websites In our work We classified user needs in to three aspects: style entity and action We employed Conditional Random Field (CRF) to label blog articles and titles rom auction websites and further discover user needs of the three aspects Further we find suitable activity-goods pairs and recommend them to users
Date of Award2016 Aug 22
Original languageEnglish
SupervisorWen-Hsiang Lu (Supervisor)

Cite this

Predict Purchasing Needs behind User Queries to Recommend Activity-Goods Pairs
佩怡, 余. (Author). 2016 Aug 22

Student thesis: Master's Thesis