We Know What You Want to Buy: A Demographic-based
System for Product Recommendation On Microblogs
Wayne Xin Zhao
1
, Yanwei Guo
2
, Yulan He
3
, Han Jiang
2
, Yuexin Wu
2
and Xiaoming Li
2
1
School of Information,Renmin University of China, China
2
School of Electronic Engineering and Computer Science,Peking University, China
3
School of Engineering & Applied Science, Aston University, UK
batmanfly@gmail.com, pkuguoyw@gmail.com, y.he@cantab.net,
jianghan08@gmail.com, wuyuexin@gmail.com, lxm@pku.edu.cn
ABSTRACT
Product recommender systems are often deployed by e-commerce
websites to improve user experience and increase sales. How-
ever, recommendation is limited by the product informa-
tion hosted in those e-commerce sites and is only triggered
when users are performing e-commerce activities. In this
paper, we develop a novel product recommender system
called METIS, a MErchanT Intelligence recommender Sys-
tem, which detects users’ purchase intents from their mi-
croblogs in near real-time and makes product recommen-
dation based on matching the users’ demographic informa-
tion extracted from their public profiles with product demo-
graphics learned from microblogs and online reviews. METIS
distinguishes itself from traditional product recommender
systems in the following aspects: 1) METIS was develope-
d based on a microblogging service platform. As such, it
is not limited by the information available in any specific
e-commerce website. In addition, METIS is able to track
users’ purchase intents in near real-time and make recom-
mendations accordingly. 2) In METIS, product recommen-
dation is framed as a learning to rank problem. Users’ char-
acteristics extracted from their public profiles in microblogs
and products’ demographics learned from both online prod-
uct reviews and microblogs are fed into learning to rank al-
gorithms for product recommendation. We have evaluated
our system in a large dataset crawled from Sina Weib o. The
experimental results have verified the feasibility and effec-
tiveness of our system. We have also made a demo version
of our system publicly available and have implemented a live
system which allows registered users to receive recommen-
dations in real time.
Categories and Subject Descriptors
I.2.7 [Artificial Intelligence]: Natural Language Process-
ing—Text analysis; H.3.1 [Information Storage and Re-
trieval]: text mining
Permission to make digital or hard copies of all or part of this work for personal or
classroom use is granted without fee provided that copies are not made or distributed
for profit or commercial advantage and that copies bear this notice and the full cita-
tion on the first page. Copyrights for components of this work owned by others than
ACM must be honored. Abstracting with credit is permitted. To copy otherwise, or re-
publish, to post on servers or to redistribute to lists, requires prior specific permission
and/or a fee. Request permissions from permissions@acm.org.
KDD’14, August 24–27, 2014, New York, NY, USA.
Copyright 2014 ACM 978-1-4503-2956-9/14/08 ...$15.00.
http://dx.doi.org/10.1145/2623330.2623351.
Keywords
E-commerce; product recommender; product demographic;
microblog
1. INTRODUCTION
Recent years have witnessed a great success of e-commerce
websites such as Amazon and eBay as they transcend geo-
physical barriers and allow individuals or business to for-
m transactions anywhere and anytime. A technique widely
adopted by e-commerce companies is to exploit product rec-
ommender systems to improve user experience and increase
sales. Research has found that product recommendation
largely influences consumers’ purchase decisions and is like-
ly to boost sales [23, 26, 14].
Generally speaking, there are two major challenges in the
design of product recommender systems. Firstly, it is dif-
ficult to find out users’ demographic information, which is
essential for making right recommendations to right person-
s. Existing recommender systems relying on collaborative
filtering explore techniques for matching users with similar
interests and make recommendations on this basis. Never-
theless, it is well-known that collaborative filtering suffer-
s from the “cold-start” problem when a recommender sys-
tem knows little about a new user. Secondly, existing rec-
ommender systems embedded in e-commerce websites can
only make recommendations when users are performing e-
commerce activities in those websites, which cannot capture
users’ instantaneous purchase intents outside those websites.
Although there has been much research work on online
product recommendation [24, 10, 14], most studies only fo-
cus on constructing solutions for certain e-commerce web-
sites and are thus constrained by the information available
there. In our work here, rather than relying on limited infor-
mation available in any specific e-commerce website, we aim
to develop a generic online product recommender system by
exploring vast amount of information available externally
such as that on social media platforms.
Online social media has already become the new arena of
our lives and involved different aspects of our so cial presence
from day to day. Through online activities such as chatting
with friends and posting short status updates, online social
networks (OSNs) have become important platforms where
users discuss their needs and desires [11], and even disclose
their personal information [3]. Thus, in this paper, we pro-
pose to capture users’ purchase intents from OSNs and de-
velop a generic product recommender system not limited to
any specific e-commerce website. In particular, we develop
our recommender system based on a microblogging service