Download as pptx, pdf, or txt
Download as pptx, pdf, or txt
You are on page 1of 30

DHANALAKSHMI SRINIVASAN ENGINEERING COLLEGE

PERAMBALUR - 621212

ONLINE FAKE REVIEWS DETECTION IN E-


COMMERCE

SUBMITTED BY:
GUIDED BY:
CONTENTS

 ABSTRACT

 OBJECTIVE

 EXISTING SYSTEM

 DISADVANTAGES

 PROPOSED SYSTEM

 ADVANTAGES

 LITERATURE SURVEY

 DATA FLOW DIAGRAM


 ARCHITECTURE DIAGRAM

 HARDWARE / SOFTWARE REQUIREMENTS

 MODULES

 OUTPUT

 CONCLUSION

 FUTURE ENHANCEMENT

 REFERENCE
ABSTRACT

• This project proposed to identify fake reviews on e-commerce sites


using the text, rating properties, and other information from a review.
• The project also proposes to classify the reviews as positive or negative
based on the text used in the reviews, ratings given to the product so on.
• This system will find out fake reviews made by posting fake comments
about a product by identifying the IP address along with review posting
patterns.
• To find out the review is fake or genuine, system will find out the IP
address of the user if the system observes fake review send by the same
IP Address many a times it will inform the admin to remove that review
from the system.
OBJECTIVE

The objective is to develop the information regarding fine-grained aspects


in identifying fake reviews. model the relationships between users and
products and use these relationships as a regularization term to redefine the
model’s objective function.
EXISTING SYSTEM
• Recommender Systems are indispensable to provide personalized services on
the Web. Recommending items which match a user’s preference has been
researched for a long time, and there exist a lot of useful approaches. First,
discuss existing collaborative filtering methods with explicit feedbacks.
• Collaborative filtering with explicit feedbacks that both positive and negative
feedbacks are observed in the dataset. The collaborative filtering methods can
be divided into the memory-based method, the model based method and the
combination of the two.
• The memory-based method includes the neighbourhood method, which
calculates the similarity of the users or items. The model-based method
includes the matrix factorization model, the probabilistic model and cluster
based model.
DISADVANTAGES

• Only analyzed ratings from user reviews


• Fake reviews can’t be analyzed by existing work
• User can’t be identify genuine reviews
• Handle only limited number of product reviews
PROPOSED SYSTEM

• A recommendation system has been implemented based on hybrid approach of


stochastic learning and context based engine.
• They have tried to combine the existing application for recommendation to come up
with a hybrid one.
• It improves the performance by overcoming the drawbacks of traditional
recommendation systems. Recommender systems being a part of information filtering
system are used to forecast the bias or ratings the user tends to give for an item.
• Among different kinds of recommendation approaches, collaborative filtering
technique has a very high popularity because of their effectiveness.
ADVANTAGES

• System helps the user to find out correct review of the product
• Handle large number of contextual information
• User easily buy genuine products
• Recommend the positive products based on user reviews
• Automatic decision making system in product recommendation
LITERATURE SURVEY

TITLE AUTHORS AND TECHNIQUES PROBLEM DESCRIPTION


YEAR

The Effect of Fake Reviews on e- Detecting fake online reviews


Commerce During and After Covid-19 outperforms on the yelp dataset and
Hina Tufail, 2022 K-Nearest Neighbor (KNN)
Pandemic: SKL-Based Fake Reviews accuracy depends on the quality of the
Detection data

Fake Online Reviews: A Unified Logistic Regression, Naïve starting with intra-domain classification
Mujahed Abdulqader,
Detection Model Using Deception Bayes, Decision Tree, and and extending to cross-domain
2022
Theories Random Forest classification

Fraud Detection in Online Product Fraud Aware Heterogeneous FAHGT is able to handle heterogeneous
Review Systems via Heterogeneous graphs with multi-relation and multi-
Graph Transformer Songkai Tang, 2021 Graph Transformer node type without designing meta-path
(FAHGT) manually.

Detecting Spammer Groups From Product


Reviews: A Partially Supervised Learning partially supervised learning
Lu Zhang, 2017 Detection process is low.
Model model (PSGD)

election of unlabeled data- It is essential


Revisiting Semi-Supervised Learning for
Jitendra Kumar to select the correct unlabeled data as it
Online Deceptive Review Detection semi-supervised learning
Rout,2017 might hamper the performance of the
model.
LITERATURE SURVEY

TITLE AUTHORS AND TECHNIQUES PROBLEM DESCRIPTION


YEAR
Do Reviewers’ Words and Behaviors
Help Detect Fake Online Reviews and It involves lots of arbitrary
Thi-Kim-Hien Hierarchical logistic
Spammers? Evidence From a Le,2022 decisions, it does not work with
Hierarchical Model regression (HLR) missing data

An Ensemble Model for Fake Online


Review Detection Based on Data It may cause data complexity that
Resampling, Feature Pruning, and Jianrong Yao, Data resampling, Feature
2021 pruning
deteriorates classification
Parameter Optimization
accuracy.

Detecting Fake Reviews Using


Multidimensional Representations Meiling Liu, Neural network
With Fine-Grained Aspects Plan loss of credibility
2020. model

Fake Review Detection Based on


Jingdong Wang, Deep Neural Network Lighting Condition Background
Multiple Feature Fusion and
2020.
Rolling Collaborative Training

Rami CNN Detecting accuracy is less


Fake Reviews Detection
Mohawesh,2021.
DATA FLOW DIAGRAM

LEVEL 1

Online e-commerce framework

Admin login User login

Post the products with


View the products Buy the products
review datasets

Provide rate, review


and emoticons to
products
DATA FLOW DIAGRAM

LEVEL 2

Input: Reviews, Ratings Extract Star


and Emoticons Ratings

Extract the Output: Correlated


Reviews datasets

Extract the
emoticons
DATA FLOW DIAGRAM

LEVEL 2

Input: Datasets Star values


collections calculations

Output: Positive
sentiments
Text mining for
reviews
Output: Negative
sentiments
Emoticons
symbol values
ARCHITECTURE DIAGRAM
HARDWARE REQUIREMENTS

Processor : Dual core processor 2.6.0 GHZ


RAM : 1GB
Hard disk : 160 GB
Compact Disk : 650 Mb
Keyboard : Standard keyboard
Monitor : 15 inch color monitor
SOFTWARE REQUIREMENTS

Operating system : Windows 10


Front End : Python
Back end : MYSQL Server
Tool : Python 2.7 IDLE
Modules
• Login

• Upload review dataset

• Text processing

• Sentiment analysis

• Classification

• Fake review detection


Module Description
• Login- In this module allows admin to the system for login process. Admin
can enter the username and password to login the system.

• Upload review dataset- After login process, admin collect the user reviews
and upload dataset in to the system. This dataset is stored in the database
which contains user reviews on the e-commerce website’s.

• Text processing- This module user reviews are taken from the dataset. The
reviews are in text format. First text is preprocessed using some text
processing methods.

• Sentiment analysis- After text processing, particular keywords are extracted


from the text then sentiment is analyzed from the text which is means positive
or negative review.
Module Description
• Classification- The sentiment analyzed from the text which can be classified
such as positive or negative based on the extracted keywords.

• Fake review detection- In this module admin can identify the fake review of
users. Multiple fake reviews are detected by same user account that is detected
using this approach.
SCREENSHOT
SCREENSHOT
SCREENSHOT
SCREENSHOT
SCREENSHOT
CONCLUSION

• In this project, proposed a product recommendation system based on hybrid


recommendation algorithm.
• The main advantages of method are a visual organization of the data based on the
underlying structure, and a significant reduction in the size of the search space per
result output.
• And user can easily search the products anywhere and anytime.
• Ratings, reviews and emoticons are analyzed and categorized as positive and negative
sentiments.
FUTURE ENHANCEMENT

• Further enhancements can be made in designing the screens. Some more

forms can also be added so as to better retrieve the feedback details.

• Various other options can also be added for the better usability of project
REFERENCE
• Hina Tufail “The Effect of Fake Reviews on e-Commerce During and After Covid-19
Pandemic: SKL-Based Fake Reviews Detection” 2022
• Mujahed Abdulqader “Fake Online Reviews: A Unified Detection Model Using
Deception Theories” 2022.
• Songkai Tang “Fraud Detection in Online Product Review Systems via Heterogeneous
Graph Transformer”2021.
• Lu Zhang “Detecting Spammer Groups From Product Reviews: A Partially Supervised
Learning Model” 2017.
• Jitendra Kumar “Revisiting Semi-Supervised Learning for Online Deceptive Review
Detection”2017.
REFERENCE

• Thi-Kim-Hien “Do Reviewers’ Words and Behaviors Help Detect Fake


Online Reviews and Spammers? Evidence From a Hierarchical Model”
2022.
• Jianrong Yao “An Ensemble Model for Fake Online Review Detection
Based on Data Resampling, Feature Pruning, and Parameter
Optimization” 2021.
• Meiling Liu “Detecting Fake Reviews Using Multidimensional
Representations With Fine-Grained Aspects Plan” 2020.
• Jingdong Wang “Fake Review Detection Based on Multiple Feature
Fusion and Rolling Collaborative Training” 2020.
• Rami Mohawesh “Fake Reviews Detection” 2021.
THANKYOU

You might also like