ICWSM-2013-Geli-burst

Sep 3, 2026

Upload by Paura

This document presents a study on detecting and classifying spam and non-spam reviews, focusing on a method called "Geli-burst". The study evaluates the performance of different classification algorithms, including k-means, and presents results in various tables and figures. The key metrics reported are precision, recall, F-score, and accuracy, applied to both general reviews and specifically "burst" reviews, which are assumed to be reviews posted in a short, concentrated period. The tables detail the performance of "spammer", "non-spammer", and "mixed" categories against these algorithms, with a particular focus on the "With local" approach showing impressive spam classification. The Kappa statistic is also provided to measure inter-rater agreement. The research appears to be a technical paper from the ICWSM 2013 conference, likely detailing a novel approach to review analysis and spam detection within online platforms, with an emphasis on computational methods and quantitative evaluation.

LanguageEnglish

CategoryGeneral

FormatPDF

Pages10

Reading Time4 min

AccessFree to download

Results

Table 1: Spam vs. Non-Spam Classification by Algorithm

Table 2: Performance Metrics for All Reviews and Burst Reviews (Scenario 1)

Table 3: Performance Metrics for All Reviews and Burst Reviews (Scenario 2)

Table 4: Performance Metrics for All Reviews and Burst Reviews (Scenario 3)

Table 5: K-Means Clustering Results with Different Approaches

FAQs

What is the primary focus of this document?

The document focuses on detecting and classifying spam and non-spam reviews using a method called "Geli-burst" and evaluating different classification algorithms.

What metrics are used to evaluate the performance of the classification methods?

The document reports on precision, recall, F-score, and accuracy for various scenarios, and also includes the Kappa statistic for inter-rater agreement.

Which approach showed the most effective spam classification according to the tables?

The "With local" approach, particularly in the k-means clustering results, demonstrated the most effective spam classification, with a high number of correctly identified spam reviews and a low number of non-spam reviews.

Sep 3, 2026

Upload by Paura