Scribd
Upload
8 views

4-2 Project

This document summarizes a presentation on sentiment analysis of Yelp reviews using machine learning. It discusses using machine learning algorithms like Naive Bayes, logistic regression and support vector machines to classify Yelp reviews as positive or negative. The presentation proposes a novel approach using deep learning and transfer learning to improve on existing methods by better understanding context and reducing dependency on labeled data. It then outlines the proposed method and compares accuracies of different algorithms. The presentation concludes that satisfactory accuracy can be achieved through supervised learning for sentiment analysis of Yelp reviews.

Uploaded by

nakkasiri333
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
8 views

4-2 Project

This document summarizes a presentation on sentiment analysis of Yelp reviews using machine learning. It discusses using machine learning algorithms like Naive Bayes, logistic regression and support vector machines to classify Yelp reviews as positive or negative. The presentation proposes a novel approach using deep learning and transfer learning to improve on existing methods by better understanding context and reducing dependency on labeled data. It then outlines the proposed method and compares accuracies of different algorithms. The presentation concludes that satisfactory accuracy can be achieved through supervised learning for sentiment analysis of Yelp reviews.

Uploaded by

nakkasiri333
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 18

WELCOME ALL

PROJECT TITLE :

Sentiment Analysis of Yelp Reviews by Machine Learning

Guide: K Karunakar sir

03/23/2024 PRESENTATION TITLE 1


1. Sentiment Analysis
• Sentiment Analysis is a process of analyzing a piece of text
written by a writer to identify and classify the opinions writer

about the topic is positive , negative , or neutral.


• Yelp is a review from which provides reviews on local business.
• Users from anywhere in the world can post reviews and rate any business in this social
networking site.
• In this process , the textual yelp reviews of businesses are analyzed to assign a
probability for the review as having positive or negative sentiment.
Agenda
1.Existing System and its Disadvantages:
• The existing systems for sentiment analysis of Yelp reviews typically rely on machine
learning algorithms such as Naive Bayes, Support Vector Machines (SVM), While these
approaches have shown reasonable performance, they suffer from several disadvantages
1. Limited Context Understanding
2. Dependency on Labelled Data
3. Difficulty in Handling Domain-specific Vocabulary
4. Scalability Issues

03/23/2024 PRESENTATION TITLE 3


2.Proposed System and its Advantages:
To address the limitations of existing systems, we propose a novel approach for sentiment
analysis of Yelp reviews, leveraging advanced techniques in natural language processing and
machine learning. Key advantages of our proposed system include:

1. Contextual Understanding with Deep Learning


2. Semi-supervised and Transfer Learning
3. Domain-specific Adaptation
4. Scalability and Efficiency

03/23/2024 PRESENTATION TITLE 4


2.1 Contextual Understanding with Deep Learning

Our system incorporates state-of-the-art deep learning architectures such


as Transformer-based models (e.g., BERT) to better capture the contextual
nuances in Yelp reviews. These models excel at understanding complex
language structures and can handle sarcasm, irony, and ambiguity more
effectively.

03/23/2024 PRESENTATION TITLE 5


2.2 Semi-supervised and Transfer Learning

To alleviate the dependency on labelled data, our system utilizes semi-


supervised and transfer learning techniques. By leveraging pre-trained
language models and fine-tuning them on a smaller labelled dataset
specific to Yelp reviews, we can achieve competitive performance with
reduced data annotation efforts.

03/23/2024 PRESENTATION TITLE 6


2.3 Domain-specific Adaptation

We employ techniques for domain adaptation and customization to tailor


the sentiment analysis model to the specific vocabulary and language
patterns found in Yelp reviews. This ensures better accuracy and relevance
in sentiment classification across diverse business domains.

03/23/2024 PRESENTATION TITLE 7


2.4 Scalability and Efficiency

Our system is designed with scalability in mind, utilizing distributed


computing frameworks and optimized algorithms to process large volumes
of Yelp reviews efficiently. By leveraging parallel processing and cloud-
based infrastructure, we can handle increasing data loads without
compromising on performance.

03/23/2024 PRESENTATION TITLE 8


Primary goal
Literature Survey
• Sentiment Analysis is a major tool for a machine to understand
human psychology.
• This technology is being studied extensively in order to
implement in the fields where humans were needed to detect
sentiment or feeling.
• It is of major importance in assistant chatbots and combined
with speech recognition technology.
• It can also be used to substitute humans in call centers.

03/23/2024 PRESENTATION TITLE 10


PROPOSED
METHOD
PROPOSED METHOD
• This paper enlightens the implementation of Machine Learning algorithms which
processes textual, statistical data provided by the Yelp dataset as a part of the Yelp
Dataset Challenge.
• It is an application of supervised learning.
• The aim is to classify the business reviews into graded categories such that they can be
roughly sorted in an order from bad to good or negative to positive.
• There is a sequence of steps involved from taking raw data to predicting whether a
review is of positive or negative sentiment and to what degree.
• Tokenizing, part of speech tagging, assigning probabilities of good or bad on the basis of
occurrence to every adjective that occurs within the reviews are just some of the major
steps to be followed

03/23/2024 PRESENTATION TITLE 12


The following algorithms have been
used:
1. Naive Bayes
2. Multinomial Naive Bayes
3. Bernoulli Naive Bayes
4. Logistic Regression
5. Linear SVC (Support Vector Clustering

03/23/2024 PRESENTATION TITLE 13


ACCURACIES IN DIFFERENT
ALGORITHMS

Algorithm Accuracy
Naïve Bayes 79.12

Multinomial Naïve Bayes 78.92

Bernoulli Naïve Bayes 73.22

Logistic Regression 78.88

Linear SVC (Support Vector Clustering) 75.32

Proposed 78.44

03/23/2024 PRESENTATION TITLE 14


Limitations and Future Scope

1 2 3 4 5
Quality of
It does not deal
Accuracy and Training set training set
It does not with
less diversity of suppuration is determine the
detect sarcasm food ,ambiance,
categorization not an easy task efficiency of
cost, service
each algorithm

03/23/2024 PRESENTATION TITLE 15


Conclusion

• Analysis of the results shows that we have successfully been able to get a
satisfactory level of accuracy in the classification of Yelp reviews using
supervised learning.

• By attaining higher accuracy for such experiments we can see a day in the
near future where it will be normal for day to day machines to understand
our sentiments by words straight out of our mouth.

03/23/2024 PRESENTATION TITLE 16


Meet our team
RAJYA LAKSHMI
MUNGARA Mirjam Nilsson
Team lead Chief Executive Officer

Flora Berggren Rajesh Santoshi


Chief Operation Officer VP Marketing

03/23/2024 PRESENTATION TITLE 17


Thank you
From
Batch-3
CSE-B
SIET

You might also like