Opinion and suggestion mining on customer reviews is a system providing both organizations and customers with valuable insights into the quality of products and services. The system operates on two perspectives, the customer, and the organization. For new customers, the system analyses reviews from previous customers to help them determine whether the product or service offered by the organization is worth purchasing. For organizations, the system performs suggestion mining on negative reviews to provide actionable feedback for improvements. Additionally, the system provides organizations with the percentage of customers who have liked or disliked their products or services, offering valuable insights into customer satisfaction. Overall, this project offers a valuable tool for organizations looking to improve their offerings and for customers seeking reliable information to inform their purchasing decisions.
In today's world for an organization to grow it is very important to satisfy the needs and expectations of the customer. To understand what a customer expects from the product or the service that the organization provides it is very important for the organization to analyze the customer reviews and get working to improve the quality of service or product according to the reviews.
Also, when we talk on the part of the customer, customer gets confused over a wide range of reviews given by the people who've already used the service or the product.
This wide range of reviews may contain some positive as well as negative reviews. In order for a new customer to analyse if he/she should buy the product or service or not they have to go through a bulk of reviews and that is nearly impossible to be analysed by an individual and come to a conclusion.
The main motive of this project is to help the organization to analyse the percentage of positive and negative reviews over their particular product or the service and based on the negative reviews our project gives them improvement tips and suggestions using which they can improve the quality of the service or the product.
II. AIMS AND OBJECTIVE
Our aim is to provide good improvement suggestions for the organizations so that they can work upon the improvement suggestions provided based upon the negative reviews that the customers have given for their product or service and act accordingly in applying those improvement suggestions on their product or service. On the customer perspective our aim is to provide a new customer an easy way to analyze the bulk of reviews and help them make a decision if they can opt to buy service or the product that the organization offers. A ready-to-use model that works as stated above on the organization as well as the customer's perspective is made for industries to analyze different customer reviews according from the variant opinions made for the product or the service that the organization offers.
III. LITERATURE REVIEW
Opinion mining is a computational technique used to extract subjective information from textual data. In recent years, opinion mining has become increasingly important due to the proliferation of social media platforms, which generate a large volume of unstructured data containing opinions, reviews, and feedback. Previously there have been many projects based on opinion mining for customer review which gave you the sentiment of a single review.
The following are the review papers on opinion mining for single customer review
In 2020, Zhang et al. proposed a novel approach to improve the accuracy of opinion mining in social media posts by incorporating domain-specific knowledge. They used a graph-based approach to represent domain-specific knowledge and achieved significant improvements in sentiment classification accuracy compared to traditional methods.
In 2021, Jha et al. presented a study on emotion detection in Indian social media data that uses a deep learning-based approach. They used a multi-layer perceptron network to classify emotions and achieved high accuracy on a Hindi emotion detection dataset.
In 2022, Liu et al. proposed a novel approach to aspect-based sentiment analysis that integrates human attention mechanisms and domain-specific knowledge. They used a hierarchical attention mechanism to capture the importance of different aspects and achieved state-of-the-art results on several benchmark datasets.
In this study, we employed a dataset made up of 38932 reviews which has happy and unhappy sentiments associated to each.
Opinion mining and suggestion mining is a machine learning program in which we have used the classification algorithm and trained our machine learning model based on reviews associated to happy and unhappy classes.
G. Model Evaluation
We evaluated the trained model using a test dataset that contains customer reviews with sentiment labels as ‘happy’ and ‘not happy’. The evaluation metrics we used included accuracy, precision, recall, and F1-score.
V. WORKING OF MODEL
Our project titled by opinion mining of customer review deals with providing solutions to two perspective the customer perspective and the organizations perspective
Customer reviews play a major role in identifying if the service or the product provided by organization is good to use or not.
???????A. Model Working Based on Customer Perspective
Our project helps the new customers to know if the product or service is good or not , based upon the bulk of customer reviews that have previously used the product or the service of the organization. It becomes hard for the new customers to analyze if the product or service is good or not if they have to go through each and every review and sit analyzing if the product has positive reviews to the maximum or negative reviews to the maximum.
In conclusion, the implementation of our machine learning model for opinion and suggestion mining has proven to be effective in providing valuable insights to both customers and seller organizations. The model accurately analyzes and presents the percentage of positive and negative reviews for products to customers, allowing them to make informed decisions when purchasing. At the same time, organizations can get the percentage of positive and negative reviews for product, additionally model provide benefit from the suggestions provided by customers, which can help them improve their products and services enabling organizations to address specific concerns or issues that may arise. The accuracy of the suggestions provided by the model can help organizations make data-driven decisions, leading to more satisfied customers and increased profits. Overall, this project highlights the potential of machine learning in improving customer satisfaction and enhancing the performance of seller organizations.
 Opinion Mining Using Multi-Dimensional Analysis (IEEE 2023)- SATARUPA BISWAS AND G. POORNALATHA - This review addresses emerging literature in the field of Opinion Mining with a particular emphasis on user-generated textual content
 Opinion Mining using Machine Learning Approaches: A Critical Study (IEEE 2020) - Jayabrabu Ramakrishnan, Dinesh Mavaluru, Karthik Srinivasan, Azath Mubarakali
 Opinion Mining on Food Services using Topic Modelling+ and Machine Learning Algorithms (IEEE 2020) - R. Akila; S. Revathi; G. Shreedevi - real time McDonald\'s dataset has been used for this work to determine the positive, negative and neutral reviews.
 \"Customer opinion mining for product improvement using machine learning\" by S. R. Wadhwa et al. (2020) - The authors propose a machine learning-based approach to customer opinion mining that can be used to improve product development and customer satisfaction.
 \"Sentiment analysis of social media data using natural language processing techniques\" by S. Singh et al. (2020) - The authors explore the use of natural language processing techniques for sentiment analysis of social media data.
 \"Opinion mining using machine learning techniques: A review\" by P. Kumar et al. (2020) - This paper provides a comprehensive review of the literature on opinion mining using machine learning techniques.
 \"Opinion mining of news articles using machine learning techniques\" by A. Singh et al. (2021) - This paper investigates the use of machine learning techniques for opinion mining of news articles.
 \"Opinion mining for political analysis: A case study of Indian election data\" by S. Goyal et al. (2021) - This paper presents a case study of opinion mining for political analysis using Indian election data.