How to Identify Inauthentic “Add to Favorites” Activity in e-Commerce

This article is part of the Academic Alibaba series and is taken from the paper entitled “Detecting Crowdturfing “Add to Favorites” Activities in Online Shopping” by Ning Su, Yiqun Liu, Zhao Li, Yuli Liu, Min Zhang and Shaoping Ma, accepted by The Web Conference 2018. The full paper can be read here.

Image for post
Image for post

The “Add to Favorites” button is a popular function on online shopping sites which helps users to make a record of potentially interesting items for future purchases. It is usually regarded as a type of explicit feedback signal for item popularity and therefore also adopted as a ranking signal by many shopping search engines.

With the increasing usage of crowdsourcing platforms, some malicious online sellers also organize crowdturfing activities to increase the numbers of “Add to Favorites” for their items in order to obtain higher ranking positions.

A portmanteau of “crowdsourcing” and “astroturfing”, crowdturfing is a relatively new spamming phenomenon that mobilizes large numbers of users to artificially boost support for, or the reputations of, companies, organizations, products, or even opinions.

This new kind of malicious activity proposes challenges to traditional search spam detection efforts because the crowdsourced workers involved are, most of the time, also normal online shopping users making use of functions such as the “Add to Favorites” button.

To tackle this problem, a team of researchers from Tsinghua University and the Alibaba Adversarial Intelligence Group have proposed a novel detection framework based on a Probabilistic Graphical Model. By comprehensively analyzing the spamming activities from the perspective of behavior, user, and item, “Add to Favorite” activities are modeled using an Activity Factor Graph as follows:

Image for post
Image for post
Activity Factor Graph Model
Image for post
Image for post

By capturing the characteristics of “Add to Favorite” spamming activities, the above model can learn how to distinguish spam from normal activities. With real-world testing, the researchers found that 60% of the spam activities can be detected in the top 1% test records. The method can achieve an Area Under the Curve (AUC) of 0.903, which is a more than 50% boost compared with traditional methods such as Logistic Regression, SVM, ant Random Forest etc.

These results show that the graphical model performs well in in countering spam on E- commerce platforms. The detection system has already been deployed on Taobao, China’s largest domestic e-commerce platform, in order to suppress spamming activities and therefore provide better and more accurate information for authentic customers.

Read the full paper here.

Alibaba Tech

First-hand and in-depth information about Alibaba’s latest technology → Search “Alibaba Tech” on Facebook

Written by

First-hand & in-depth information about Alibaba's tech innovation in Artificial Intelligence, Big Data & Computer Engineering. Follow us on Facebook!

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store