Course Content

Abstract :

Yelp is a review forum which provides reviews on local businesses. Users from anywhere in the world can post reviews and rate any business in this social networking site. The growth of social media in recent years has led to an increase in online reviews that reflects consumer opinions. Firms benefit greatly from making this information available in order to respond more effectively to consumer dissatisfaction and to exploit market opportunities by observing standards that may represent unsatisfied needs. The present study aims to address this problem through a survey based on the Yelp platform. To this end, 14,000 comments related to different tourism products were used and text mining techniques and topic models were applied to find the main latent topics discussed in the online comments and their associated sentiments.

An algorithm that can predict the review rating of a potential business with only existing information about the location and business categories would be an invaluable tool in making investment decisions. Utilizing the Yelp business dataset, we built a model, that can do as such, by classifying whether a potential business belongs to a positively-reviewed class (star ratings greater than or equal to 4) or a negativelyreviewed class (star ratings less than 4) given its location in latitude and longitude and the categories the potential business belongs to.

In recent years, online reviews have been playing an important role in making purchase decisions. This is because, these reviews can provide customers with large amounts of useful information about the goods or service. However, to promote factitiously or lower the quality of the products or services, spammers may forge and produce fake reviews. Due to such behavior of the spammers, customers would be misleaded and make wrong decisions. Thus detecting fake (spam) reviews is a significant problem.

So we proposed a system with the help of machine learning techniques, NLP techniques like count vectorizer, TF-IDF Transformer and Machine learning  algorithms like Logistic Regression, KNN, SVC, Random Forest ,Decision Tree and Naïve Bayes to predict the review is positive or negative and also we will detect whether the review is fake or real based on review data entered by the user in the front end.

Here we also do the review translation that is the review data entered by the user can be converted to different Indian languages like kannada, telgu, hindi, tamil etc and also the user will get audio ouput of the review in the translated language only.

 

 

 

Introduction:

Sentiment analysis has become an important research area for understanding people’s opinion on a matter by analyzing a large amount of information. Millions of people express their thoughts about various services or products using social networking sites, blogs or popular reviews sites. The active feedback of the people is valuable not only for companies to analyze their customers’ satisfaction and the monitoring of competitors, but is also very useful for consumers who want to research a product or a service prior to making a purchase. Yelp users’ reviews express opinions and sentiments about businesses and service providers among a given rating, scaled from 1 to 5, which is used as a general metric review. Yelp challenge dataset contains information about 61 000 local businesses, 1.6 million reviews and 366 000 users in 10 cities across 4 countries on the globe. Yelp challenge dataset is much larger than previous released Yelp academic dataset, which contains 15 585 business and 335 022 users’ reviews.

There are thousands of reviews online, which makes it convenient for people to make decisions, but the amount of data makes it difficult to sort through . The real value of online reviews is in its content and the certainty that reviewer indeed received products or services prior to writing the review. Promotion or demotion of the products and services is one of the main reasons for deceptive reviews. At times, to create better ratings for the venue, hotel owners pay employees to fabricate false reviews . Alternatively, some reviewers write negative reviews for malicious reasons, like to distort the reputation of the business reviewed. Yelp.com is one of the biggest online review sites. It uses a filtering algorithm to detect fake reviews

Nowadays with the advent of the e-commerce, an increasing number of people are taking pleasure of shopping online, and then sharing their opinions on the electronic business website. These online opinions may be used by customers and merchants when they make purchase and other decisions. For the online reviews, positive reviews play a stimulative role in reapping economic benefits and well-deserved reputation for merchants’ businesses. Thus, it makes merchants have strong intentions to manoeuvre their fame and employ specialized imposters posting higher opinions on the shopping sites. Besides, there exists competition between online merchants. Consequently, the employed fraudsters may post negative opinions to defame their rivals, resulting in bad sales of products and services. Such individuals are called opinion spammers, and their behavior is called opinion spamming . Because of the above activities, the online customers might be misleaded by the deceptive opinions. Therefore, opinion spam has attracted important attention from both business and research circles. And the research purposes mainly aim to identify fake reviews and let online opinions recover reliability and facticity.

So we proposed a system with the help of machine learning techniques, NLP techniques like count vectorizer, TF-IDF Transformer and Machine learning  algorithms like Logistic Regression, KNN, SVC, Random Forest ,Decision Tree and Naïve Bayes to predict the review is positive or negative and also we will detect whether the review is fake or real based on review data entered by the user in the front end.

Here we also do the review translation that is the review data entered by the user can be converted to different Indian languages like kannada, telgu, hindi, tamil etc and also the user will get audio ouput of the review in the translated language only.

 

1.1. Aim and Objective

 

1.1.1. Aim

The aim of our project is to predict the review is positive or negative and also to detect the fake or real review. Here we also aimed to do review translation with is help of machine learning and NLP techniques and machine learning algorithms.

 

1.1.2. Objective

The objective of this project is to facilitate the user to find the sentiment of the review and helps to detect fake reviews and user can do also review translation.

 

 

 

 

 

 

 

 

 

 

1.2Existing System

Yatani et al. and Huang et al. designed different  interfaces for Yelp  that show top  frequent adjectives used  to  describe  a business. These  interfaces  also visualized  overall  sentimental scores  of reviews  in various  colors.  The authors  did not provide any evidence to show whether using adjectives  are more effective than other words. It is also not clear whether  using  sentimental  scores  have  more  advantages  over  raw reviews' text.

 

1.2.1 Disadvantages

 

In the existing system only sentimental analysis or fake review detection is done not both together and is not that efficient and accuracy of individual system is not satisfying.

 

1.3 Problem statement

The growth of social media in recent years has led to an increase in online reviews that reflects consumer opinions. Firms benefit greatly from making this information available in order to respond more effectively to consumer dissatisfaction and to exploit market opportunities by observing standards that may represent unsatisfied needs.

In recent years, online reviews have been playing an important role in making purchase decisions. This is because, these reviews can provide customers with large amounts of useful information about the goods or service. However, to promote factitiously or lower the quality of the products or services, spammers may forge and produce fake reviews. Due to such behavior of the spammers, customers would be misleaded and make wrong decisions. Thus detecting fake (spam) reviews is a significant problem.

 

1.4 Proposed system

We proposed a system with the help of machine learning techniques, NLP techniques like count vectorizer, TF-IDF Transformer and Machine learning  algorithms like Logistic Regression, KNN, SVC, Random Forest ,Decision Tree and Naïve Bayes to predict the review is positive or negative and also we will detect whether the review is fake or real based on review data entered by the user in the front end.

Here we also do the review translation that is the review data entered by the user can be converted to different Indian languages like kannada, telgu, hindi, tamil etc and also the user will get audio ouput of the review in the translated language only.

1.4.1 Advantages

•         The proposed system is cost effective.

•         Here review sentiment analysis, fake review detection and review translation all together occurs in the same single system.

•         Reduces human effort and intervention.

•         Drastically reduces time compared to manual detection.

•         Accuracy in prediction.

•         Easy to use