https://dl.ucsc.cmb.ac.lk/jspui/handle/123456789/4731 Sat, 25 Apr 2026 18:06:56 GMT 2026-04-25T18:06:56Z https://dl.ucsc.cmb.ac.lk/jspui/handle/123456789/4759 Title: Psychotherapeutic cognition extractor : Lexical ontologies to facilitate cognitive behavioral therapy and trait modelling Authors: Vidanage, B.V.K.I Abstract: Abstract Use of artificial intelligence in psychotherapy has mostly, not gone beyond the level of close – ended web based questionnaires. In this research, a novel computational intelligence based approach is proposed to facilitate the psychotherapeutics procedures via creating comfortable platforms both to the consultants and patients. Lack of annotated datasets creates a major barrier in creating stable knowledge models in the arena of artificial intelligence. Amount of annotated datasets available in psychotherapy domain is almost nil. Therefore, after recognizing exclusive attributes like, rich domain modelling capability, contextual knowledge representation ability, ease of scalability and no need of a comprehensively large annotated datasets, it`s decided ontologies to be the ideal form of AI technology to be used in this research, as it characteristics suits very well with the nature of the attempted research problem to be solved. Ontologies are defined as an ideal way of encoding human intelligence to machine intelligence format. Therefore, after consulting multiple consultant psychologists and psychiatrists, covering the entire island, knowledge associated with cognitive behavioral therapy (CBT) procedures and personality trait assessment based on Big Five (Openness, Conscientiousness, Extraversion, Agreeableness and Neuroticism) OCEAN trait modelling are derived. Henceforth, pool of lexicon based ontological structures are proposed, presenting the machine readable form of knowledge extracted from human consultants. To assure smooth, operation of the proposed ontology pools in cognition segmentation, multiple clusters of Part-Of-Speech (POS) tag based rule repositories are derived considering the potential structural aspects of each of cognition expression patterns. Thereafter, each of these cognition specific POS rule sequence is mapped with a specified object property in the ontology. Further, each object property of the ontology, is linked with an annotated cognition specific lexicon pool. This mapping sequence will assure computational process enforcement of the both CBT and OCEAN based psychotherapeutic procedures. The developed application has been tested in both quantitative and qualitative forms. Thematic Analysis evaluation technique is used to assess the textual feedbacks provided by the consultants. Further, confusion matrices have been derived denoting true, positives, true negatives, false positives and false negatives associated with the cognition segmentations processed by the v prototype. With the help of these information, quantitative measures such as precision, recall, sensitivity F-measure and accuracy dimensions are derived. This research contributes to the health informatics domain, which is a very important and new discipline coming under computer science, by releasing, pool of lexicon based ontological structures which can be used for the effective segmentation of the human cognition and trait assessment, facilitating the consultants’ role towards improving the therapeutic alliance in between patient and the consultant. Further the overall architecture provided by this research prototype is an easily expandable base line architecture to cater with new knowledge requirements as well. Therefore, new knowledge aspects associated with CBT and OCEAN characteristics can be easily absorbed by the same architecture proposed by the prototype, making it a stable architecture addressing scalability of the prototype as well. The combination of all these knowledge structures and ontological mappings can be released as one knowledge api, facilitating the reusability and expandability of the structures proposed by any researcher who is interested in coupling cognition segmentation aspects to a potential healthcare information system to be developed. Tue, 01 Jan 2019 00:00:00 GMT https://dl.ucsc.cmb.ac.lk/jspui/handle/123456789/4759 2019-01-01T00:00:00Z https://dl.ucsc.cmb.ac.lk/jspui/handle/123456789/4758 Title: Transformation of Requirement Techniques to Reduce Duplication of Work in Methodologies Authors: Madanayake, R.S. Abstract: Abstract The existence of Commercial Software for many decades, has witnessed a vast assortment of diagrammatic and Non-diagrammatic software design and development techniques, leading to a tremendous amount of Duplication of Work. This has also lead to the existence of many legacy software systems which need to be transformed into modern ones to make them more efficient and scalable. Although a few transformation tools are available, none can yet comprehensively integrate most transformations. Preliminary surveys of the industry and academia done both by the researcher who authored this thesis and a previous group, regarding the Academic projects as well as the IT industries in Australia, Singapore and Sri Lanka, have revealed an enormous amount of Duplication of Work both in the industrial and academic software projects, where it also affected the documentation procedures, leading to a waste of resources and time due to duplication of work. This was due to the fact that, different modelling techniques which use the same data, are required for these projects. The above problems (Duplication of Work and the need to integrate legacy designs) were framed as a research question, – “How could the different types of transformations of Software engineering models be combined to expedite the software requirement analysis process?”. Its answer is an Ontological Framework, to define relationships of both the Unified Modelling Language (UML) and Non-UML Design techniques established upon the Ontology of philosopher Mario Bunge, which was already used in Information Systems, via a branch known as “conceptual things” which could be extended to incorporate suitable newer entities. This makes it easy to interrelate the various diagramming techniques and transform one into another, in such a way as to accomplish modular transformations effectively. From the initial surveys, it was determined that Entity Relationship Diagrams (ERDs), Class Diagrams, User Stories and Use Case Models were the most popular diagrams used as well as where the most Duplication of Work occurred, according to the data analysis conducted by the researchers. The Analysis of Opinions for transformations between different types of diagram (i.e. UML and Non-UML), were tested through three assignment based experiments given to undergraduates reading for software engineering subjects, due to their knowledge in this field. Those experiments involved ERDs, Class Diagrams, User Stories and Use Case Models, and vi they provided researchers with the most compatible UML and Non-UML diagram Transformations. An open source software was developed as the proof of concept prototype. The final evaluation of this research project involved the participation of individuals with industry experience acting as respondents using a Questionnaire and there were positive responses of over 91% for all questions in the feedback form, which inferred the success of this research, where both the major knowledge domains – the modular transformations between Use Case Models versus Epics and User Stories in Agile Software Development, as well as the modular transformations between Entity Relationship Diagrams and Class Diagrams, were evaluated. Tue, 01 Jan 2019 00:00:00 GMT https://dl.ucsc.cmb.ac.lk/jspui/handle/123456789/4758 2019-01-01T00:00:00Z https://dl.ucsc.cmb.ac.lk/jspui/handle/123456789/4757 Title: Train Traffic Optimization System to Minimize Stochastic Delays on Single Line Railway Track Authors: Kasthoori, A.P Abstract: Abstract Optimization is a far-ranging research area applicable to many areas of science and technology. Increasing demand and complexities in the domain of transportation, day by day trigger many countries in the world to pay their attention to maximizing a rapid utilization of railway transportation due to its higher degree of cost-effectiveness and environment friendliness. However, the main consideration is to utilize successfully the existing infrastructure in order to provide better and enhanced service. There is no difference in Sri Lanka, compared with other countries, facing the same problems in transportation in order to provide efficient railway services. Providing better railway service in Sri Lanka is a big challenge than other countries due to managing stochastic delays. This is the main critical problems has to be addressed in the Sri Lankan context. Annual train delays due to stochastic events are considerably high in rate than in other countries. This directly effects on total productivity of the country itself intern. The main focus of this research is to reduce delay created by stochastic events, applying optimization technologies and low-cost devices. This research focused on minimizing stochastic delay on single line rail network. Firstly in this context, the critical issue faced was gathering accurate real-time train dynamics that need to trigger rescheduling process, just after each stochastic delay, created by stochastic events. Secondly, finding flexible and adaptable algorithm in order to search optimal running time and optimal departure time is not easy, and after literature reviews, Ant Colony Optimization was applied as a methodology to the problem domain to search optimal running time and departure time that need train to move from one node to another node within the undirected graph. The main concern of research was to minimize stochastic delay by providing a realtime assisting tool to train controllers that help them in the rescheduling process. Further, this system also provides a better platform to train drivers, station masters, and train controllers to minimize stochastic delay automatically by sending auto generated instructions and establishing communication between them. As a research outcome, Computer Based Real Time Train Re-scheduler (CBRTTRS) was designed to achieve more functionalities than above, such as off-time rescheduling, data mining, railway safety monitoring, and public acknowledgment, etc. Tue, 01 Jan 2019 00:00:00 GMT https://dl.ucsc.cmb.ac.lk/jspui/handle/123456789/4757 2019-01-01T00:00:00Z https://dl.ucsc.cmb.ac.lk/jspui/handle/123456789/4756 Title: Social Sensor Networks for News Mining Authors: Fernando, M. A. I. D. Abstract: Abstract With the development of technology, many people tend to use the Internet which has resulted in an increase in usage of social networks and microblogs, inducing many organizations too to share their news in social networks and microblogs. News providers are such organizations that share large amount of news in social networks and blogs and Twitter is one such common social network, which is well known as a microblog. The short messages (Tweets) which are shared in Twitter can produce many important information. S2Net tool was developed in order to analyze these Tweets and generate useful information and present it in a suitable manner. Situations where one is interested in the news topics rather than news groups. For such cases, the clustering technique was used, in which the news was clustered into news topics. Expectation–Maximization clustering (EM Clustering) and Hierarchical Clustering were the methods used in these situations. The results show that Hierarchical Clustering with Simple Linkage function performs better than EM Clustering. The Simple Linkage function can detect the small relationships between clusters. Because of the high dimension of the features, there will be many relationships which are hard to detect. Therefore using Simple Linkage function can improve the accuracy. There can be situations where one is interested in the news topics rather than news groups. For such cases, the clustering technique was used, in which the news was clustered into news topics. EM Clustering and Hierarchical Clustering were the methods used in these situations. The results show that Hierarchical Clustering with Simple Linkage function performs better than EM Clustering. The Simple Linkage function can detect the small relationships between clusters. Because of the high dimension of the features, there will be many relationships which are hard to detect. Therefore using Simple Linkage function can improve the accuracy. These two analyzing methods were evaluated using two evaluation techniques. The classification method was evaluated using F-measure. According to the F measure, it is clear that the Random Forest method performs well than the other methods. The clustering method was evaluated by getting review comments for the each cluster. The reviewer evaluates and marks the mismatches for each cluster. According to their evaluations, EM clustering performs with 68.52% accuracy and Hierarchical clustering performs with 89.93%. Tue, 01 Jan 2019 00:00:00 GMT https://dl.ucsc.cmb.ac.lk/jspui/handle/123456789/4756 2019-01-01T00:00:00Z