http://drr.vau.ac.lk/handle/123456789/1002 Sun, 05 Apr 2026 19:38:04 GMT 2026-04-05T19:38:04Z http://drr.vau.ac.lk/handle/123456789/1040 THE APPROACHES OF PERSONALIZED ADVERTISEMENTS IN TELEVISION USING MACHINE LEARNING Anuruddha, I.D.; Suthaharan, S.S. In recent years, personalized television advertising that uses machine learning methods to boost interaction with viewers and the success of advertisers have received a lot of attention. In this investigation, we have investigated the use of machine learning models for targeted advertising, namely the LinearSVC, Polynomial Kernel, NuSVC, and igmoidal Kernel models. The purpose of this research is to examine the efficacy and viability of these models in targeting individual television viewers with relevant commercials. The feasibility of using this method to forecast consumer tastes and adapt marketing campaigns appropriately is investigated. Using higher-order polynomial transformations, the Polynomial Kernel model attempts to capture complicated connections within the data. By allowing users to adjust the parameter that determines how many support vectors are used, the NuSVC model improves upon SVM and makes it more amenable to being optimized for targeted advertising. Finally, the Sigmoid Kernel model provides a sigmoid function-based non-linear decision boundary that can adapt to a wide range of viewer preferences and send out more relevant ads. Extensive testing is performed on a dataset of audience preferences to assess the performance of these models, taking into account characteristics like demographics, watching history, and online behavior. Each model’s ability to predict the reactions of viewers to commercials is evaluated using a variety of measures, precision, recall, as well as including accuracy. The trials reveal the benefits and limitations of each model, illuminating their potential in developing targeted TV commercials. The results imply that although LinearSVC provides a simple method, it may not be able to accurately capture complicated preferences from viewers. Better performance is shown by the Polynomial Kernel model since it captures non-linear interactions; nonetheless, processing complexity may be an issue. The NuSVC model’s potential in improving model performance is highlighted by its capacity to regulate support vectors. The Sigmoid Kernel model performs well in accepting different tastes, giving it a good option for tailoring advertising to each individual. By investigating and contrasting the efficacy of LinearSVC, Polynomial Kernel, NuSVC, and Sigmoidal Kernel models, this research adds to our knowledge of personalized television commercials. Marketers and academics alike may benefit from these results since they can now use machine learning to create more successful and interesting personalized marketing campaigns Wed, 25 Oct 2023 00:00:00 GMT http://drr.vau.ac.lk/handle/123456789/1040 2023-10-25T00:00:00Z http://drr.vau.ac.lk/handle/123456789/1039 EMOTION DETECTION FROM TEXT BASED SENTENCES USING MACHINE LEARNING Jayasinghe, H.; Thirukumaran, S. In the realm of digital communication, extracting emotions from text assumes a pivotal role in deciphering human sentiment. This research delves into the significance of discerning sentiments from textual content, leveraging a diverse suite of machine learning algorithms. The study harnesses the power of Support Vector Machines, Linear Support Vector Classification, Random Forest, and Decision Trees to decode the intricate emotional nuances intertwined with language. Central to this exploration is the filtration and classification of six cardinal emotions: ‘joy’, ’fear’, ‘anger’, ‘sadness’, ‘disgust’, ‘shame’, and ‘guilt’. These emotional facets serve as the bedrock for analysis, reflecting the spectrum of humanexperiences. The study’s results reveal the prowess of these algorithms in emotion classification. Notably, Support Vector Machines, Linear Support Vector Classification, and RandomForest showcase a remarkable accuracy of 96%. On the other hand, Decision Trees set a higher benchmark with an impressive accuracy of 92%. This research amplifies the potential of machine learning in deciphering emotions from text, shedding light on the synergy of language and sentiment analysis. The outcomes extend beyond numerical metrics, enriching our understanding of human emotional expression within textual communication across diverse contexts and applications Wed, 25 Oct 2023 00:00:00 GMT http://drr.vau.ac.lk/handle/123456789/1039 2023-10-25T00:00:00Z http://drr.vau.ac.lk/handle/123456789/1038 SPAM EMAIL CLASSIFICATION AND IDENTIFICATION Keerththana, K.; Charles, E.Y.A. Email has become one of the most wide spread ways of communication in today’s society. Email spam, commonly known as junk email, spam mail, or simply spam, refers to unsolicited messages sent in large quantities through email. Even though some spam emails contain valuable information, quite often spam emails are unwanted and lead to online fraud. Hence it is necessary to filter spam emails from regular emails. An improved spam classification approach will make users’ inboxes free from spam emails while not missing any potential emails. In this research work we analyzed the classification of emails into spam and legitimate emails using the contents of the email. This work further explored the classification of the spam emails based on categories such as promotion, marketing, news, security and others. This work analyzed the applicability of the word embedding approach for spam classification. Two different kaggle datasets (sms-spam-collection-dataset, spam filter) were used in this research work. This work considered a word embedding approach for text representation and multiple classifiers (LSTM, SVM). Since there are no publicly available multiclass spam classification data sets, an incremental approach is proposed to build the classifier. Both datasets were manually categorized and used to build the multiclass classifier. This work identified the Word2Vec model with SVM classifier obtained highest accuracy of 0.86, 0.87 for both datasets. As future work, this initial classifier will be used to classify the Enron spam email dataset. With a manual analysis the results will be verified and will be used to fine tune the classifier in multiple epochs Wed, 25 Oct 2023 00:00:00 GMT http://drr.vau.ac.lk/handle/123456789/1038 2023-10-25T00:00:00Z http://drr.vau.ac.lk/handle/123456789/1037 SOLUTION OF LAPLACE EQUATION BY DIFFERENTIAL TRANSFORM METHOD Kularathna, R.S.M.; Kajan, N. The Laplace equation is a fundamental partial differential equation with wideranging applications in various scientific and engineering domains. In this research, we explore four distinct scenarios governed by the Laplace equation and devise methods to solve them effectively. To tackle these intricate situations, we employ the differential transform method (DTM), a powerful mathematical tool known for its ability to generate solutions that align closely with the underlying physical and engineering principles. By implementing the DTM, we delve into the analysis of these scenarios, each accompanied by either Dirichlet or Neumann boundary conditions. This choice mirrors real-world scenarios, enriching the practical relevance of our study. In this study, we discover that adjusting the parameters m and n can significantly enhance the accuracy of our solutions. We carefully fine-tune these parameters to strike a balance between computational efficiency and solution precision. Our code dynamically terminates based on the smallness of individual terms, signifying the convergence of the series to a sufficiently accurate result. Through systematic experimentation, we identify the optimal values for m and n that bring our solutions closest to the exact solutions. This empirical insight not only enhances the precision of our findings but also sheds light on the sensitivity of the DTM to parameter variations. In essence, our research not only provides solutions to Laplace equation-driven problems but also underscores the practicality of the differential transform method in solving complex issues with real-world implications. By emphasizing the role of parameter adjustment in optimizing solution accuracy, we contribute to the broader understanding of howmathematical techniques can be effectively harnessed to address practical challenges in physics and engineering. This study represents a significant step toward bridging the gap between theoretical mathematics and practical applications, highlighting the importance of precision and adaptability in mathematical modeling Wed, 25 Oct 2023 00:00:00 GMT http://drr.vau.ac.lk/handle/123456789/1037 2023-10-25T00:00:00Z