Comparative Analysis of Machine Learning Techniques for Opinion Mining in Web Texts Using Artificial Intelligence - research report

Aug 5, 2026

Upload by Rowan

This paper presents a comparative analysis of machine learning techniques for opinion mining (sentiment analysis) in web texts using artificial intelligence. It evaluates traditional classifiers, including Naïve Bayes, k-Nearest Neighbor, and Random Forest, alongside neural-network models such as LVQ, Elman, and FFNN. An ensemble strategy combining Naïve Bayes and SVM is proposed to raise classification accuracy, and the KINN neural model is introduced for enhanced sentiment performance. Experiments on a Kaggle dataset show improved accuracy for sentiment classification, supporting better product and service quality using customer feedback.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages5

Reading Time13 min

AccessFree to download

Introduction

Importance of Opinion Mining

Challenges in Opinion Mining

Research Objectives

Review of Literature

Methodology

Data Collection

Data Preprocessing

Feature Extraction

Comparative Analysis of Machine Learning Techniques for Opinion Mining in Web Texts Using Artificial Intelligence Ram Chandra Pal 1 and Dr. Suresh S. Asole2 1,2Department of Computer Science & Engineering, Dr. A. P. J. Abdul Kalam University, Indore, MP, India [rcpal07@gmail.com](rcpal07@gmail.com1 suresh_asole@yahoo.com2)[1](rcpal07@gmail.com1 suresh_asole@yahoo.com2)[ suresh_asole@yahoo.com](rcpal07@gmail.com1 suresh_asole@yahoo.com2)[2](rcpal07@gmail.com1 suresh_asole@yahoo.com2) Abstract : This paper presents a comparative analysis of machine learning techniques for opinion mining in web texts using artificial intelligence. Opinion mining, also known as sentiment analysis, involves extracting and classifying opinions from textual data found on the web, such as reviews and blogs. The study evaluates the effectiveness of traditional data mining classifiers Naïve Bayes, k-Nearest Neighbor, and Random Forest and neural network classifiers LVQ, Elman, and FFNN. A novel approach combining Naïve Bayes and SVM in an ensemble method is proposed to enhance classification accuracy. The KINN algorithm, a neural network-based model, is introduced, demonstrating improved performance over existing methods. Experimental results using a dataset from Kaggle show that the proposed methods achieve higher accuracy in sentiment classification, offering valuable insights for improving product and service quality based on customer feedback. Keywords: KINN, FFNN, NLP, AI INTRODUCTION Natural Language Processing (NLP) has revolutionized the way textual data is processed, enabling the transformation of text into machine-readable formats. When combined with Artificial Intelligence (AI), this transformation allows for the extraction of opinions from textual data on the web, a process known as opinion mining or sentiment analysis. Opinion mining encompasses a range of tasks such as sentiment classification, feature-based sentiment classification, and opinion summarization. Importance of Opinion Mining The proliferation of user-generated content on platforms like social media, blogs, and review sites has made opinion mining crucial for businesses and organizations. By analyzing these opinions, companies can gauge public sentiment towards their products, services, and brand image. This feedback loop is vital for strategic decision-making and improving customer satisfaction. Challenges in Opinion Mining Opinion mining involves several challenges: • Ambiguity in Text: Human language is inherently ambiguous, with words and phrases often having multiple meanings depending on context. • Subjectivity Detection: Distinguishing between subjective and objective statements can be difficult. • Feature Extraction: Identifying and extracting relevant features from text requires sophisticated algorithms. These challenges necessitate the use of advanced machine learning techniques to accurately classify and summarize opinions. Machine Learning Approaches Various machine learning algorithms are employed in opinion mining, each with its strengths and limitations. Techniques such as Naive Bayes, Support Vector Machines (SVM), and ensemble methods combine multiple algorithms to improve classification accuracy. These methods leverage large datasets to train models capable of understanding and predicting sentiment with high precision. RESEARCH OBJECTIVES This thesis aims to: • Analyze different machine learning algorithms used for opinion extraction. • Evaluate the performance of these algorithms in terms of accuracy, efficiency, and scalability. • Propose improvements or new approaches to enhance opinion mining techniques. REVIEW OF LITERATURE Chen and Qi (2011) report about the job of informal organization in online customer's choice procedure when they scan for an unpractised item to purchase. They get a lot of framework derivation and coordinated into three-phase framework engineering. For choice stages, straight chain co

FAQs

What problem does the paper address in web data analysis?

The paper addresses opinion mining, also called sentiment analysis, which extracts and classifies opinions from web text such as reviews and blogs.

Which machine learning methods are compared for sentiment classification?

The study compares traditional classifiers (Naïve Bayes, k-Nearest Neighbor, Random Forest) and neural network classifiers (LVQ, Elman, FFNN).

What improvements does the paper propose to enhance classification accuracy?

It proposes an ensemble combining Naïve Bayes and SVM, introduces the KINN neural network model, and reports higher sentiment-classification accuracy from experiments on a Kaggle dataset.

Aug 5, 2026

Upload by Rowan