https://dl.ucsc.cmb.ac.lk/jspui/handle/123456789/4552 2026-03-31T09:53:18Z https://dl.ucsc.cmb.ac.lk/jspui/handle/123456789/2822 Title: An Affective Computing Approach to Identify the Potential for Mental Health Problems by Measuring User Affect Authors: Karunaratne, G.T.I. Abstract: Emotional Intelligence (EI) is a key indicator of successful and social individuals. Being emotional intelligent they become both intra-personal intelligent, and inter-personal intelligent. They become capable of understanding not only their own feelings and needs, but also of others’. EI is not merely intuitive; it could be enhanced with self-reflections, and by being in company with others. With the proliferation of Computer Mediated Communication (CMC) and digital media, the means of social contact has been drastically altered. It has ensured better social connectedness, but shattered strengths of social ties, marking a downfall of emotional intelligent capabilities of many individuals. There is an inherent after effect: individuals find it difficult to regulate their emotions, seek help from others, and also to provide emotional support to others. The social isolation, and therefore the psychological distress among individuals have been kindled. It may not be possible and worthwhile to bring the pieces back by wiping out CMC and digital media. Instead, it is to be explored how to use this inevitable social change to address the issue positively. Accordingly, it has been investigated how the way the computer users interact with computers can be used to identify their psychological distress. In order to avoid deliberate alterations to the interaction patterns, a non-intrusive mechanism was adopted. The research was conducted using Action Research (AR) method following a progressive development in three consecutive research cycles. During these research cycles carried out, it has been learned that the stress is reflected in the way the users interact with computers; these interactions could be monitored and recorded non-intrusively using an activity logger; and more importantly these measures are actually correlated with the psychological distress. Predicting the prevalence for mental health problems based on the human computer interaction patterns is thereby become viable. Two interesting indicators have been exposed during these efforts. The prediction model does not fit into a linear mathematical equation, nor could make reliable predictions by applying a pre-defined rule set. Instead, the model which gave more promising results used machine learning techniques, and adopted case-based reasoning. This sets out new line of thinking: Is emotional intelligence an ability of case-based learning and reasoning of feelings and needs? 0030-09-14T00:00:00Z https://dl.ucsc.cmb.ac.lk/jspui/handle/123456789/2821 Title: Landslides Disaster Prediction: A Case Study for Badulla District Authors: Subhashini, L.D.C.S. Abstract: Landslides are the one of the most recurrent and prominent disaster in Sri Lanka. Sri Lanka has been subjected to a number of extreme landslide disasters that resulted in a significant loss of life, material damage, and distress. It is required to explore a solution towards preparedness and mitigation to reduce recurrent losses associated with landslides. Artificial Neural Networks (ANNs) and Hidden Markov Model (HMM) are now widely used in many computer applications spanning multiple domains. This research examines the effectiveness of using Artificial Neural Networks and Hidden Markov Model in landslides predictions and the possibility of applying the modern technology to predict landslides in a prominent geographical area in Sri Lanka. A thorough survey was conducted using a questionnaire with the participation of resource persons from several national universities in Sri Lanka to identify and rank the associated factors for landslides in Sri Lanka. Then, data was analyzed using the SPSS software. It shows that rainfall is the factor which associated the most. Twelve factors were identified and the identified factors were divided into internal factors and external factors. The landslide related factors which include external factors (Rainfall, Number of Previous Occurrences and Influence of Construction) and internal factors (Soil Material, Geology, Land Use, Curvature, Soil Texture, Slope, Aspect, Soil Drainage, and Soil Effective Thickness) were extracted. The ArcGIS 10 software was used to extract the internal factors; slope, lithology, soil material, distance from drainage by using digital elevation models which were created using existing maps. External factors were collected by using past data. These factors were used to recognize the possibility to occur landslides by using an ANN and HMM. The models acquire the relationship between the factors of landslide during the training session. These Models with landslide related factors, the inputs are trained to predict three classes namely, ‘landslide occurs’, ‘landslide does not occur’ and ‘landslide likely to occur’. Once trained, the models predict the most likely class for the prevailing data. Then, a prototype was built using the Mathlab® software. Finally compared two models with regards to prediction accuracy, False Acceptance Rates and False Rejection rates. This research indicates that the Artificial Neural Network could be used as a strong decision support system to predict landslides efficiently and effectively than Hidden Markov Model. Factors were identified with regard to all the districts in the country so this model can be applied for any District. To the best of our knowledge it is the first time that such a research has been conducted in Sri Lanka. Therefore, the outcome of this research will be of immense beneficial to the country. 0030-09-14T00:00:00Z https://dl.ucsc.cmb.ac.lk/jspui/handle/123456789/1820 Title: Automatic Text Summarization for Sinhala Authors: Welgama, W.V. Abstract: With the rapid development of information and communication technology, people are surrounded with vast amounts of information albeit with less and less time or ability to make sense of it. The field of automatic summarization which has been in existence since the 1950’s is anticipated to find solutions to this issue. With the adaptation of Unicode technology in 2004, the Sinhala language began to appear in computers rapidly and Sinhala language users also began to experience the above issue. This research on Automatic Text Summarization in Sinhala is carried out to find the possible approaches to address the above issue with the minimum linguistic resources. The field of automatic text summarization began with some classical approaches which attempted to indentify the most salient information of an article using some thematic features. This research was intended to indentify such features for the Sinhala language with the most suitable approach to define each of these features for achieving accurate summaries. In order to benefit from all these features, this research proposes a best possible linear combination of identified features. The proposed method was evaluated by comparing the machine generated and human extracted summaries based on the primary assumption that the human summaries are perfect. Results show that the sentence location feature is the best individual feature for extracting most informative sentences from Sinhala articles while the linear combination of keyword feature, title words feature and the sentence location feature giving the best performance for a summarizer. Results revealed some equations to define the flow of information over a Sinhala article which can be used in many such applications. Further, this research provides a benchmark for future research on Sinhala automatic text summarization. 0004-01-14T00:00:00Z https://dl.ucsc.cmb.ac.lk/jspui/handle/123456789/1819 Title: Automatic Word Clustering in Application of Open-Ended Response Categorization N P Authors: Medagoda, N.P.K. Abstract: Open ended questions are an essential and important part of survey questionnaires. They provide an opportunity for researchers to discover unanticipated information regarding the domain of study. However, they are problematic for processing since they are unstructured questions to which possible answers are not suggested, and the respondent is free to answer in his or her own words. This thesis presents novel methods of categorizing such open ended survey responses. A document clustering technique is employed in this study to categorize responses to open-ended survey questions. Supervised and unsupervised methods of categorizing open ended responses are tested in the study. Initially the author proposed a hierarchical clustering based algorithm as the unsupervised method to code the open-ended responses which were not labelled at all. The algorithm employs several natural language processing techniques to extract a classification of responses automatically. Naive Bayes classification was proposed as the supervised solution. This Naive Bayes algorithm was proposed for the open ended responses which were partially labelled. Two experiments were carried out to determine the accuracy of the proposed algorithms which proved to be promising. Hierarchical clustering based algorithm shows more than 70% accuracy when compared with the manually coded responses. The proposed Naive Bayes algorithm didn’t not illustrate the results as it expected. Therefore Positive Naive Bayes algorithm was introduced and it achieved an overall performance of 80% 0004-01-14T00:00:00Z