Fake Review Detector
Uncover the truth in reviews—shop smarter, not harder. Analyzes e-commerce reviews, filters fake feedback, and provides data-driven purchase recommendations based on authentic customer experiences
Created on 5th April 2025
•
Fake Review Detector
Uncover the truth in reviews—shop smarter, not harder. Analyzes e-commerce reviews, filters fake feedback, and provides data-driven purchase recommendations based on authentic customer experiences
The problem Fake Review Detector solves
In today's digital marketplace, online reviews heavily influence purchase decisions. However, these reviews aren't always reliable due to:
-- Fake reviews posted by sellers to boost ratings
-- Incentivized feedback that skews product perception
-- Review bombing that unfairly damages product reputation
-- Outdated reviews that no longer reflect current product quality
-- Fake Review Detector addresses these challenges by:
-- Filtering misleading content - Our system identifies patterns indicative of fake or manipulated reviews.
-- Providing authentic rating averages - We calculate ratings based only on verified, legitimate feedback.
-- Analyzing sentiment accurately - Beyond star ratings, we examine the actual content of reviews to gauge customer -
satisfaction.
-- Offering clear recommendations - Our verdict system simplifies decision-making with straightforward advice based on real
data.
-- Visualizing review distributions - Interactive charts show how ratings are distributed, revealing patterns that might be
hidden in averages.
Our Fake Review Detector transforms the chaotic world of online reviews into actionable insights, saving consumers time and protecting them from misleading information. Instead of wading through hundreds of reviews, users get immediate clarity on product quality based on authentic customer experiences.
Challenges we ran into
1)Scraping Dynamic E-commerce Sites
Flipkart's website uses dynamic JavaScript rendering and frequently changes its HTML structure, making consistent review extraction challenging. We implemented:
--Multiple fallback CSS selectors
--Regex pattern matching for ratings
--A specialized scraper that adapts to different page structures
2)Accurate Rating Extraction
Reviews displayed ratings in various formats (stars, text, numbers), causing inconsistent data. We solved this by:
--Creating a multi-tiered extraction system that tries different methods
--Implementing robust error handling to prevent crashes
--Adding fallback calculations for when direct extraction fails
3)Rating Distribution Visualization
Initially, our distribution bars wouldn't render correctly when certain reviews had zero counts. We fixed this by:
--Adding minimum width styling for bars with non-zero counts
--Implementing position absolute with z-index to ensure proper rendering
--Creating custom CSS animations for a better user experience
4)Integration of frontend with backend(with trained models)
Tracks Applied (4)
First Prize
Second Prize
Third Prize
HashHacks 7.0
Technologies used