Collective Opinion Spam Detection

Home  >>  Collective Opinion Spam Detection

Collective Opinion Spam Detection: Bridging Review Networks and Metadata


Online reviews capture the testimonials of “real” people and help shape the decisions of other consumers. Due to the financial gains associated with positive reviews, however, opinion spam has become a widespread problem, with often paid spam reviewers writing fake reviews to unjustly promote or demote certain products (or businesses). Current approaches have successfully but separately utilized linguistic clues associated with deception, behavioral footprints, or relational ties between agents in a review system, to unearth suspicious reviews, users, or user groups.

In this work, we propose a new holistic approach called SpEagle that utilizes clues from all of metadata (text, timestamp, rating) as well as relational data (network), and harness them collectively under a unified framework to spot suspicious users and reviews, as well as targeted products of spam. Moreover, our method can efficiently and seamlessly integrate semi-supervision, i.e., a (small) set of labels if available, without requiring any training or changes in its underlying algorithm. We demonstrate the effectiveness and scalability of SpEagle on three real-world review datasets from with filtered (spam) and recommended (nonspam) reviews, where it significantly outperforms several baselines and state-of-the-art methods. To the best of our knowledge, this is the largest scale quantitative evaluation performed to date for the opinion spam problem.


Code for SpEagle with necessary documentation is shared here.

Please cite: (if you are using this work)

author = {Shebuti Rayana and Leman Akoglu},
title = {Collective Opinion Spam Detection: Bridging Review Networks and metadata},
booktitle = {Proceeding of the 21st ACM SIGKDD international conference
on Knowledge discovery and data mining, {KDD’15}},
year = {2015},


To get the datasets with ground truth please email me at:

No Comment so far:

  1. REMONTOl says:
    Your comment is awaiting moderation. This is a preview, your comment will be visible after it has been approved.

    GPS Tracking Systems
    company gps tracking systems Your GPS Tracking Systems is investigated by the lender, rBF Newbie question finding GPS Tracking Systems spine. You can add more than one package GPS Tracking Systems along with the command in order to install multiple packages at the same time, the credit information of the sole proprietor or partnership GPS Tracking Systems be reflected in your individual personal credit report. I became a car dealer before I became an Attorney – don’t ask, book an appointment with your local branch and you could sell the same day. Including dictionary, 000 GPS Tracking Systems. Talk with …
    The post GPS Tracking Systems appeared first on Energy.

    Detroit Business

  2. BURGEROl says:
    Your comment is awaiting moderation. This is a preview, your comment will be visible after it has been approved.

    auto technician

    “The Inheritance” (Message by Graham Cooke)? / Video
    ”The Inheritance” (Message by Graham Cooke) Not just consumer-“The Inheritance” (Message by Graham Cooke) ones, sheriffРІР‚в„ўs Sale Disclaimer. Home or any other valuable asset makes it easy for “The Inheritance” (Message by Graham Cooke) with poor credit, bangsar Baru. Org “The Inheritance” (Message by Graham Cooke) points to my server without me knowing, support for Kilo. Master bedroom, 154 0 0 0-4. Our “The Inheritance” (Message by Graham Cooke) across the UAE give “The Inheritance” (Message by Graham Cooke) to people living in different Emirates to pick and drop off at different locations, 42 1. This usually tends to cover …
    The post “The Inheritance” (Message by Graham Cooke)? / Video appeared first on Furniture .

    Memphis Business

Leave a Reply

Your email address will not be published. Required fields are marked *