https://dl.ucsc.cmb.ac.lk/jspui/handle/123456789/1693 2026-11-10T10:25:50Z https://dl.ucsc.cmb.ac.lk/jspui/handle/123456789/1732 Title: Semantically Enhanced Search Mechanism for Web Service Registries Authors: Gunarathna, M.R.D.I.A. Abstract: Service Oriented Architecture (SOA) has emerged as one of the leading technologies that is being used by most of the business entities today in order to carry out most of their day to day business operations. It facilitates di erent business applications operating under in di erent platforms to interact with each other using standard protocols. As the amount of web services developed daily increase drastically new problems like service usage monitoring and automated service discovery emerged. Service discovery is one of the critical factors in SOA in order to get the maximum usage of its features like composition and interoperability. To get the maximum usage of the provided facilities of a service one needs to locate the service correctly. With the rapid increment of the number of available web services, nding out the appropriate service becomes a hugely challenging task. This gives rise to web service discovery mechanisms and di erent service registry mechanisms. Semantic based web service discovery is one such mechanism that could be used to enhance the service discovery phase of a particular registry. The work presented in this research consists of design, implementation and evaluation of a semantic based enhanced service discovery method that could be plugged in to any particular web service registry. 2013-12-19T00:00:00Z https://dl.ucsc.cmb.ac.lk/jspui/handle/123456789/1734 Title: Recognition of Human Facial Expression in 3D Space Authors: Gunathilaka, D.M. Abstract: Facial expression is one of the most powerful resources for people to coordinate conversation and communicate emotions and other mental, social, and physiological cues. Studying the facial expression is one successful way of studying the emotional state as the facial expression represents ones emotional state successfully. We propose a method to identify the facial expression represented by human face in this thesis. Facial expression is an important channel of nonverbal communication as well. If the facial expression can be recognized automatically then the results can be applied in many important areas like human computer interaction which can be used to enable the communication between human and computers in a more natural way. Facial expressions were studied mostly using 2D images or 2D video sequences. Recent advances in imaging technology and ever increasing computing power have opened up new areas of automatic facial expression recognition. We have proposed a method to recognize the facial expression by analyzing the 3D geometry of the human face. The geometry of the human face is studied in 3 dimensional spaces and the facial expression is identi ed. There are six kinds of universally recognized facial expressions: happiness, sadness, fear, anger, disgust, and surprise. The classi cation of the facial expressions is done using an Arti cial Neural Network. 2013-12-19T00:00:00Z https://dl.ucsc.cmb.ac.lk/jspui/handle/123456789/1733 Title: A Configurable Tool to Generate Parallel Programming Applications on a Hybrid DM/DSM Cluster Authors: Gunasekara, H.A.D.R.R. Abstract: As the popularity of parallel programming gets increased day by day more people tries to use parallel computing and get the bene ts. According to user requirements and the type of the problem, di erent parallel programming models should be used in order to get the maximum performance. Among di erent kinds of parallel programming models hybrid distributed memory (DM) and distributed shared memory (DSM) model can be mention as relatively new model. Since this is a combination of two parallel programming models users have to eventually use two di erent programming languages and mix them properly. Although there are tools available to assist programmers on creating parallel computing applications, none of them support developing hybrid DM/DSM applications. So the objective of this project was to create such a tool. Using the implemented tool users can creates virtual sub-clusters which use DSM approach for internal nodes communication. Size of the sub-cluster is con gurable and can be changed according to user requirements. Connectivity among sub-cluster can be established using message channels (DM). Once users enter the logic into a single sub-cluster it can be replicated for others and execute in a parallel manner. All the relevant synchronization parts will be handled by the tool. 2013-12-19T00:00:00Z https://dl.ucsc.cmb.ac.lk/jspui/handle/123456789/1731 Title: Semantic Web Approach to Block Pornography Websites Authors: Dissanayake, D.H.S. Abstract: With the rapid growth of the internet a vast industry for the production and consumption of pornography has grown all over the globe. To keep people away from this kind of o ensive things, di erent kinds of Web ltering techniques are implemented. Today there are many Web lters available to solve this problem. They use di erent techniques to lter the Web. Blacklists & white-lists, keyword blocking, using rating systems and image based algorithms are the most popular ones among them. Keeping and updating a list is quite promising. Because there are thousands of new web pages are publishing in the inter- net. So checking all of them and listing is not an easy task. Sometimes using keyword blocking techniques, legitimate pages can be blocked too. The prob- lem of this is that rating is not compulsory. So it may not always available. Apart from that because of the self rating, ratings may not always accurate. Other than that the image scanning mechanisms may be very costly due to the huge processing time it needs for the classi cation. The main goal of the project is to develop an automated Web ltering mechanism, which extract the semantic web features of the web page for the classi cation process. The proposed system considers URL of a web page and the web page content to lter the page. Here new URL scanning & scoring mechanism as well as in-line page keyword scanning mechanism is introduces throughout the project. The web ltering engine is implemented as a extension for the Firefox browser. The project aims at minimizing over blocking rate while provide a fast and accurate ltering mechanism. 2013-12-19T00:00:00Z