An Unsupervised Neural Model for Aspect Based Opinion Mining

The Growing Hierarchical Self Organizing Map (GHSOM) was introduced as a dynamical neural network model that adapts its architecture during its unsupervised training process to represents the hierarchical relation of the data. However, the dynamical algorithm of the GHSOM is sensitive to the presence of noise and outliers, and the model will no longer preserve the topology of the data space as we will show in this paper. The outliers introduce an influence to the GHSOM model during the training process by locating prototypes far from the majority of data and generating maps for few samples data. Therefore, the network will not effectively represent the topological structure of the data under study. In this paper, we propose a variant to the GHSOM algorithm that is robust under the presence of outliers in the data by being resistant to these deviations. We call this algorithm Robust GHSOM (RGHSOM). We will illustrate our technique on synthetic and real data sets.

Effects of sentiment quantity, dispersion, and dissimilarity on online review forwarding behavior: An empirical analysis

Clustering plays an important role in data mining of large data and helps in analysis. This develops a vast importance in research field for providing better clustering technique. There are several techniques exists for clustering the similar kind of data. But only very few techniques exists for clustering mixed data items. The cluster must be such that the similarity of items within the clusters is increased and the similarity of items from different clusters must be reduced. The existing techniques possess several disadvantages. To overcome those drawbacks, Self-Organizing Map (SOM) and Extended Attribute-Oriented Induction (EAOI) for clustering mixed data type data can be used. This will take more time for clustering; the usage of SOM has the inability to capture the inherent hierarchical structure of data. To overcome this, a Growing Hierarchical Self-Organizing Map (GHSOM) is proposed in this paper. The experimentation is done by using UCI Adult Data Set.KeywordsAttribute-Oriented InductionPattern DiscoverySelf-Organizing Map (SOM)Growing Hierarchical Self-Organizing Map (GHSOM)

Social media is emerging rapidly on the internet. This media knowledge helps people, company and organizations to analyze information for important decision making. Opinion mining is also called as sentiment analysis which involves in building a system to gather and examine opinions about the product made in reviews or tweets, comments, blog posts on the web. Sentiment is classified automatically for important applications such as opinion mining and summarization. To make valuable decisions in marketing analysis where implement sentiment classification efficiently. Reviews contain sentiment which is expressed in a different way in different domains and it is costly to annotate data for each new domain. The analysis of online customer reviews in which firms cannot discover what exactly people liked and did not like in document-level and sentence-level opinion mining. So, now opinion mining ongoing research is in phrase-level opinion mining. It performs finer-grained analysis and directly looks at the opinion in online reviews. The proposed system is based on phrase-level to examine customer reviews. Phrase-level opinion mining is also well-known as aspect based opinion mining. It is used to extract most important aspects of an item and to predict the orientation of each aspect from the item reviews. The projected system implements aspect extraction using frequent itemset mining in customer product reviews and mining opinions whether it is positive or negative opinion. It identifies sentiment orientation of each aspect by supervised learning algorithms in customer reviews.

Aspect-based opinion mining is one among the thought-provoking research field which focuses on the extraction of vivacious aspects from opinionated texts and polarity value associated with these. The principal aim here is to identify user sentiments about specific features of a product or service rather than overall polarity. This fine-grained polarity identification about myriad aspects of an entity is highly beneficial for individuals or business organizations. Extricating these implicit or explicit aspects can be very challenging and this paper elaborates copious aspect extraction techniques, which is decisive for aspect-based sentiment analysis. This paper presents a novel idea of combining several approaches like Part of Speech tagging, dependency parsing, word embedding, and deep learning to enrich the aspect-based sentiment analysis specially designed for Twitter data. The results show that combining deep learning with traditional techniques can produce excellent results than lexicon-based methods.

Social media has become an essential part of everyday communication, including within the library context. Twitter, in particular, provides an Application Programming Interface (API) that enables real-time and comprehensive text-mining analyses of users’ perceptions and experiences. This study examines efforts to enhance library service quality by identifying and interpreting user sentiments on Twitter in a more detailed manner, enabling libraries to formulate targeted and effective service improvement strategies. The research consists of several stages: a literature review on sentiment analysis, data collection from Twitter using API-based data crawling, and a series of pre-processing steps, including data cleaning, case folding, tokenization, numerization, stopword removal, and stemming. Data were analyzed using the Aspect-Based Sentiment Analysis (ABSA) method. The findings indicate that the Collection and Facilities aspects receive the highest levels of positive sentiment, while Accessibility and Service attract considerable negative sentiment, signaling priority areas for improvement. Based on these results, libraries may consider extending operational hours, improving digital access, updating collections, and renovating facilities. Additionally, strengthening staff competencies and interpersonal skills is crucial for improving service quality and increasing user satisfaction.

With the development of the Internet, social media has been the main platform for human to express opinions about products/services, key figures, socio-political and economic events… Besides the benefits that the platform offers, there are still various security threats relating to the fact that most extremist groups have been abusing social media to spread distorted beliefs, to incite the act of terrorism, politics, religions, to recruit, to raise funds and much more. These groups tend to include sentiment leading to illegal affairs such as terrorism, cyber-attacks, etc. when sharing their opinions and comments. Therefore, it is necessary to capture public opinions and social behaviors in social media content. This is a challenging research topic related to aspect-based opinion mining, which is the problem of determining what the exact opinions on specific aspects are rather than getting an overall positive or negative sentiment at the document level. For an entity, the main task is to detect all mentioned aspects of the entity and then produce a summary of each aspect’s sentiment orientation. This paper proposes an aspect-based opinion mining model to address the problem of estimating public opinion in social media content. The model has two phases: 1 - extracting aspects based on double propagation techniques, and 2 - classifying opinions about the detected aspects with the consideration of the context of review sentences using the hybrid approach of machine learning and lexicon-based method.

Optimizing aspect-based sentiment analysis using sentence embedding transformer, bayesian search clustering, and sparse attention mechanism

A novel approach for image segmentation is proposed in this paper. This approach is based on the growing hierarchical self-organizing map (GHSOM), which consists of a hierarchical architecture composed of growing self-organizing maps (SOMs). The SOMs have shown to be successful for the analysis of high-dimensional input data as in data mining applications. The hierarchical architecture of the GHSOM is more flexible than a single SOM in the adaptation process to input data, mirroring inherent hierarchical relations among input data. Image hierarchical segmentation can be achieved by using this neural network model, where the hierarchical structure of segmented regions is captured. In order to evaluate the performance of this segmentation method, an application for hierarchical background modeling in video sequences is provided. Therefore, foreground detection is achieved. Experimental results show that the proposed approach is promising for applications where hierarchical segmentation is required.

Opinions are key influencers of almost all human practices. One can easily find a number of opinions about any product or services in the form of product reviews. These product reviews are available in a tremendous amount. It is not feasible or even impossible to go through each review and make a concise decision about any product. Aspect-based sentiment analysis (ABSA) comes as a solution to this problem. It gives an approach to examine online reviews and provides a summary based on these reviews. Machine learning techniques have been broadly utilized for ABSA. Recently with the evolution of processing power of computers and digitization of the society, deep learning is taking off. Deep learning methods produced state-of-the-art results in various NLP tasks without intensive feature engineering. In this chapter, we present an introduction about ABSA following a comprehensive overview of various deep learning models used in the field of ABSA.

The analysis of the opinions of customers and users has been always of great interest in supporting decision-making in many fields, especially in marketing. Sentiment analysis (SA) is the umbrella term for techniques and approaches that analyze user’s sentiments, emotions, opinions in text or other media. The need for a better understanding of these opinions paved the way to novel approaches that focus on the analysis of the sentiment related to specific features of a product, giving birth to the field of aspect-based sentiment analysis (ABSA). Although the increasing interest in this discipline, there is still confusion regarding the basic concepts of ABSA: terms like sentiment, affect, emotion, opinion, are used as synonyms while they represent different concepts. This often leads to an incorrect analysis of the users’ opinions.This work presents an overview of the state-of-the-art techniques and approaches for ABSA, highlighting the main critical issues related to current trends in this field. Following this analysis, a new reference model for SA and ABSA, namely the KnowMIS-ABSA model, is proposed. The model is grounded on the consideration that sentiment, affect, emotion and opinion are very different concepts and that it is profoundly wrong to use the same metric and the same technique to measure them. Accordingly, we argue that different tools and metrics should be adopted to measure each of the dimensions of an opinion. A qualitative case study, regarding product reviews, is proposed to motivate the advantages of the KnowMIS-ABSA model.

Consumer opinions on purchased products significantly influence the number of purchases on ecommerce platforms. Sentiment analysis of these opinions assists both consumers and companies in making informed decisions. Currently, consumers judiciously evaluate different aspects of a product to guarantee it meets their requirements. Consequently, performing sentiment analysis on various product aspects can aid consumers select the precise product and support companies in meeting on areas that prerequisite upgrading. This study recommends a systematic and expandable method to aspect-based sentiment analysis of user reviews for electrical devices like computers, mobile phones, and headphones. Through an amalgamation of advanced preprocessing, sentence-level segmentation by means of transformer-based models, and similarity-based aspect detection, the system is competent of extracting fine-grained sentiment for each product aspect. A hybrid sentiment classification strategy employing both BERT and VADER certifies robustness and accurateness in sentiment detection, while the structured transformation of data into a multi-label format permits for efficient model training. The proposed Bi-LSTM architecture with an integrated attention mechanism boosts sentiment prediction by concentrating on the utmost pertinent parts of each sentence, directing to better-quality performance across evaluation metrics, including accuracy, precision, recall, and F1-score. The exploration offers a comparative study of the performance of aspect-based sentiment analysis using a standard Bi-LSTM model versus a Bi-LSTM model with an attention mechanism, applied to analyse the reviews of three diverse products. The outcomes explains that the Bidirectional Long Short-Term Memory (Bi-LSTM) model with an attention mechanism outpaces the Bidirectional Long Short-Term Memory (Bi-LSTM) in terms of accuracy, precision, recall, and F1-score. The investigation validates the efficiency of merging NLP techniques and deep learning to deliver understandings from product reviews, presenting beneficial assistance to consumers by aiding to make informed decisions and focus on companies directing to improve their offerings based on targeted feedback.

The majority of internet users are currently searching the internet before buying certain products. One consideration of prospective buyers is product reviews (product review). Prospective consumers can decide to buy a product because it is influenced by reviews with positive sentiments, or decide not to buy a particular product because it is influenced by a negative sentiment review. Product reviews are a way of delivering consumer opinions and sentiments to a product online. In essence, the product review data mined directly from the database is unbalanced, between positive sentiment and negative sentiment. This condition makes it difficult for machine learning algorithms to perform classification and clustering functions. In this study, sentiment analysis was conducted based on Trendy Shoes products from Denim Shoes. The stages of sentiment analysis consist of data collection, initial processing, data transformation, feature selection and classification stages using SMOTEBoost. Initial processing applies the stages of text mining namely case folding, non alpha numeric removal, stop words removal, and stemming. The results of sentiment analysis were measured using the criteria of Accuracy, G-Mean, and F-Measure. By applying the test to two types of sentiment data, the results show that SMOTEBoost can classify sentiments well. SMOTEBoost's performance is compared to other ensemble techniques namely ADABoost, RUSBoost, and SMOTEBagging. Classification results of review_1 data, SMOTEboost is better in accuracy and G-Mean. While for the review_2 data, SMOTEBoost has better results for all criteria, both accuracy, F-Measure and G-Mean

Overview: The rise of e-commercehas created a more interactive and competitive environment for both sellers and buyers. In recent times, a heavy reliance on online product reviews posted on the e-commerce website by existing and verified users has been witnessed. From the extant literature, it is evident that the impact of non–textual or quantitative aspects of online product reviews (OPR), such as volume, valence, star rating, helpful votes, etc., on buying behaviour, purchase intention, sales, etc. has been vastly studied. But since the past decade, the importance of analysing and understanding the textual data of OPR has been strongly recommended and explored.Aim: Reviewers express their opinions through non-structured and unfiltered reviewswhere the information presented goes beyond mere words.The review text has embedded emotions and sentiments also. The current research aims to investigate and comprehend the implicit information conveyed by consumers in the unstructured and heartfelt reviews on Amazon.in. The study seeks to identify key semantic aspects affecting OPRs' assessment by highlighting latent topics, sentiments and intricacies within the textual content of the reviews. Methodology: Amazon is the most popular e-commerce site in India, and its review system is widely seen as reliable, clear, and standardised. Its multi-dimensional format—textual reviews(review statements and title statements), aggregate star ratings which guarantee consistency across many product categories and offers a solid foundation forqualitative research. This study employed a multistage sampling technique to extract 5,900 reviews from 59 top-selling products across three best-selling categories: beauty, fashion, and electronics from Amazon India. Web scraping has been used to extract the review data using Python as a programming language. A qualitative content analysis, topic modelling, sentiment analysis and sentiment score analysis have been employed using various packages and functions in R Studio.Findings: The research reveals that both negative and positive sentiments have significant effects on product ratings. The word count analysis indicated a predominant use of positive words such as ‘good’, ‘product’, ‘quality’, ‘nice’, and ' price’. The three distinct topics identified through topic modelling demonstrate that reviews are shaped by a combination of functional utility, sensory experience, and detailed product attributes. The sentiment analysis revealed that positive sentiment is more common than negative sentiment. Additionally, emotions like trust, anticipation and joy were predominant, while negative emotions such as disgust and anger were less frequently observed. A key finding is the asymmetrical effect, where negative sentiment has a notably stronger negative impact than positive sentiment.Although statistically significant, the relatively low R-squared value suggests that sentiment scores alone account for only a small part of the variance in product ratings, indicating that ratings are a complex outcome influenced by multiple factors beyond expressed sentiment. Implications: The study highlights the multifaceted role of online product reviews in shaping consumer behaviour on e-commerce platforms. The predominance of positive emotions emphasises the value of building strong consumer confidence. At the same time, the asymmetrical effect of negative sentiment urges the need for businesses to address the dissatisfaction promptly. For practitioners, the findingssuggest that effective product management involves not only mitigating negative reviews but also actively leveraging positive consumer emotions and experiences to strengthen brand loyalty. Strategically, e-commerce businesses should adopt a dual approach—proactively managing negative feedback while amplifying positive narratives—to enhance customer satisfaction and trust. Academically, the research contributes by evidencing the interplay of emotional, semantic, and many other factors in review analysis, offering pathways for future studies to incorporate richer variables and advanced text mining techniques.

Clustering is an unsupervised classification method that divides a data set in groups, where the elements of a group have similar characteristics to each other. A well-known clustering method is the Growing Hierarchical Self-Organizing Map (GH-SOM), that improves the results of an ordinary SOM by controlling the number of neurons generated. In this paper it is proposed a optimization of the typical GH-SOM, using a cluster validation index to verify the quality of partitioning.