A Web-based plugin for aspect extraction in sentiment analysis using Machine Learning and Natural Language Processing

Project Title

A Web-based plugin for aspect extraction in sentiment analysis using Machine Learning and Natural Language Processing

Project Area of Specialization Artificial IntelligenceProject Summary

With the explosive growth of social media and e-commerce (i.e., reviews, forum discussions, blogs and social networks) on the Web, individuals and organizations are increasingly using public reviews in these sites for their decision making. Review analysis are becoming more and more famous as they are providing loads of information about what customer exactly wants and what a company should adopt in their product to enhance their sales. As e-commerce is increasing with high rate of reviews such as Amazon have millions of reviews for their products. It is important for any e-commerce service provider to focus on reviews’ sentiment polarity, however equally important is to understand which aspect of the entity review is about. We aim to extract the aspect of any review and also perform polarity analysis whether it is a positive or negative review. Here aspect could be the battery of mobile phone, display of laptop or cooling effect of air conditioner. We will be using natural language processing (NLP) and machine learning (ML) to extract aspect and semantic analysis of the review. In conclusion we will be having a web-based plugin for e-commerce sites for review insights extraction which will be graphically showing the different aspect of reviews with their polarity. This will open new scope for e-commerce as companies could get the relevant reviews with specific aspect to work with

Project Objectives

By the time people are expressive on online shopping platforms like amazon , Alibaba and in other social websites like Twitter, Facebook etc. Due to loads of reviews on e-commerce website, it is difficult for the customer to decide whether the product is good or not. He/she can’t read all the reviews because it take lots of time. Reading few reviews may cause biased decision. As well as for the company it help to bring improvement in the product. A web-based plugins will help to decide for the customer as well as for the products owner or company by showing the summary of the reviews like number of positive reviews and number of negative reviews as well as the aspect of each reviews. This will overcome the customer difficulty to buy any product from the online store with satisfaction

Project Implementation Method

Project Approach The proposed technique is divided into three phases: data pre-processing, aspect term extraction and review polarity estimation. Each step is briefly discussed in the subsections in following. Our suggested technique for polarity estimate in customer reviews using aspect-based sentiment analysis is given in Fig. 1. 4.1 Data Pre-processing The web scraper is used to collect reviews. Stemming is used to clean up the text, and stop words are removed. Reviews are broken down into phrases and sentences. Table 1 demonstrates how reviews’ data is pre-processed

Subjectivity Segment In this study, subjective statements are those that have some positive or negative feelings, and objective statements are those that have no sentiments. Universal truths or facts are objective statements. ”I love the world,” for example, is regarded subjective, but ”the sun rises in the east” must be considered objective. As a result, we’ll need subjective reviews in the subjectivity part to do this. We use SENTIWORDNET to determine the subjectivity of reviews once they have been pre-processed. SENTIWORDNET is a text-based word list with positive, negative, and objective scores for each phrase, organised by part-of-speech key letters. SENTIWORDNET only displays a few words. To determine if a review is subjective or objective, only positive and negative scores are used. Figure 2 shows several terms’ positive, negative, and objective ratings, while Figure 3 shows the term subjectivity and sentiment score using SENTIWORDNET.[13]. Subjectivity Calculation Let pos be the word’s positive score from SENTI9 Figure 2: Terms along with their scores on SENTIWORDNET WORDNET, and neg be the word’s negative value. Equation in figure 3 is used to calculate a review or post’s subjective score. The following is the subjective score for review r: where n is the total number of words/terms in review R.

In short, this equation adds up the positive and negative weights of each term of review R, as well as the total number of words in the review or post. It’s necessary to divide by n to get the normalised subjectivity score. Even if the statement is subjective, if pos = neg, the overall score will be 0

Benefits of the Project

By the time people are expressive on online shopping platforms like amazon , Alibaba and in other social websites like Twitter, Facebook etc. Due to loads of reviews on e-commerce website, it is difficult for the customer to decide whether the product is good or not. He/she can’t read all the reviews because it take lots of time. Reading few reviews may cause biased decision. As well as for the company it help to bring improvement in the product. A web-based plugins will help to decide for the customer as well as for the products owner or company by showing the summary of the reviews like number of positive reviews and number of negative reviews as well as the aspect of each reviews. This will overcome the customer difficulty to buy any product from the online store with satisfaction.

Technical Details of Final Deliverable

Tools and Technology MERN stands for MongoDB, Express, React, Node, after the four key technologies that make up the stack[14].Figure 7 shows MERN Stack Development. • MongoDB - document database • Express(.js) - web framework 13 • React(.js) - JavaScript framework for client side • Node(.js) - JavaScript web server 5.0.1 Front End For the front end we will use React.js, the JavaScript framework the development of dynamic client-side applications in HTML. React helps us to build complex interface and for the rendering of data from the back-end server to HTML. 5.0.2 Server Tier The Express.js is server-side framework which run inside a Node.js server. Express.js is “fast, unopinionated, minimalist web framework for Node.js,” and that is indeed exactly what it is. Express.js has powerful models for URL routing and handling Hyper Text Transfer Protocol requests and responses. By making extensible markup language, HTTP Requests GETs or POSTs from React.js front-end [14]. 5.0.3 Database Tier On requirement of database the Database Tier can be used which will used MongoDB for store and retrieval of data . For the integration of python script in node.js we will create a child process that is provided by library in Node.JS to spawn a python process. 5.1 Python We will used python programming language for the implementation of machine learning modal because python is simplicity and consistency and access of machine learning libraries and framework with a platform independence.

Final Deliverable of the Project Software SystemCore Industry OthersOther Industries IT Core Technology Artificial Intelligence(AI)Other Technologies Big DataSustainable Development Goals Industry, Innovation and InfrastructureRequired Resources