2009
Bamboo Garden - Visualization of Instanting Messaging - Yang Ning (Lewis), Zhang Ting (Crystal), Zhang Xieping (David) [main supervisor: Dr. Robert P. Biuk-Aghai].
Abstract: Nowadays, more and more people are willing to use instant messaging to keep their social network or to do business. The Bamboo Garden is motivated by an interest in designing a visualization that can help us reflect our social relationships using the existing personal chat history of instant messaging. The chat history contains the issued time, the message come from, where to send and the content of the message. The idea of Bamboo Garden is to use the components of a bamboo to visualize this data.
LiteMiner III: A Data Mining Application - by Chang Lam Mao (Cholis), Chan Lap Choi (Richard), Wong Keng Kio (Dylan), Chan Weng Tim (Tim).
Abstract: Data mining, also known as knowledge discovery (KDD), is the process of extracting hidden predictive information and patterns from large amount of data. Data mining is becoming an increasingly important tool to transform this data into information. The core of data mining is the data mining algorithms. They use data provided to build a descriptive model that describing the natural of the data itself or a predictive model that can be used to forecast unseen data. In this report, we introduce LiteMiner 3, a continuation of the project LiteMiner, started in 2005. LiteMiner is a data mining application, which provides functions such as data preparation, model building, evaluation, knowledge deployment, and visualization. LiteMiner includes several algorithms for performing data mining tasks. In LiteMiner1, seven algorithms were implemented. In LiteMiner2, four algorithms and data visualizations functions were added. In LiteMiner 3, we had added seven algorithms, a new data visualization function and optimized some of the previous functions. Besides, we had developed a website that would allow people to download, utilize and comment LiteMiner3. Both of LiteMiner 1 and LiteMiner 2 were written in Visual C++ 2005 with .Net Framework 1.1. To keep the track, LiteMiner 3 was developed under the same choices. However, LiteMiner was totally updated to Visual C++ 2008 with .Net framework 3.5., the latest versions of those software.
2008
AMS - Asset Management System - by Lam Chi Wa (Isaac), Wong Ka Kit (Kit), Mak Chan Seng (Victor)
Abstract: What is Asset Management? Asset management can be defined as a comprehensive and structured approach to the long term management of assets as tools for the efficient and effective delivery of community benefits. The World Road Association (PIARC) defines asset management as: "A systematic process of effectively maintaining, upgrading and operating assets, combining engineering principles with sound business practice and economic rationale, and providing the tools to facilitate a more organized and flexible approach to making decisions necessary to achieve the public's expectations." In the light of these broad definitions, fixed asset management means managing an asset life cycle (Control, Audit & Reconcile) to satisfy the requirements of business and asset users, at the lowest possible cost over a long period of time as well as maximization of the asset utilization. The main streams in asset management are: • Identification of need for the asset, in the light of community requirements • Provision of the asset, including its ongoing maintenance and rehabilitation to suit continuing needs • Operation of the asset • Disposal of the asset when the need no longer exists or it is no longer appropriate for the asset to be retained.
Automatic Extraction of Semantic Relationships from the Web - by Leong Hou Meng (Eriol), Wan Chi Fung (Theo) [main supervisor: Prof. Gong ZhiGuo].
Abstract: In our world, there are relations between objects and there are different forms of relations that are used in our daily life such as “cat is a kind of animal”, “milk consists of water” and “shark is a creature”. We, humans, know that there exist relations between cat and animal, milk and water, shark and creature. Nowadays, the World Wide Web is becoming a new repository of information, containing a lot of text documents with plenty of such relations. Computers cannot recognize relations between words as humans. The problem is related with the fact that text documents are written for human consumption, not for machine consumption. The understanding of semantic relations is very useful for Natural Language Processing and Machine Learning problems such as question answering and textual entailment. For example, we have a sentence with a lexico-syntactic pattern like “X is a Y”, then a hyponymy relation between X and Y can be identified, or “X consists of Y”, then there is a hypernymy relation between X and Y. There are various other semantic relations, such as meronymy (i.e. x is part of y), synonymy (i.e. x denotes the same as y), and antonymy (i.e. x denotes the opposite of y), and so forth. Recognizing and managing these kind of semantic relations in text documents will cost a lot of time and manpower. There are more and more text documents being added to the WWW with a high-speed development. In order to extract these semantic relations from the WWW, we had implemented a Java application to do this extraction automatically with minimum human participation by interacting with any search tool such as Google.
2006
LiteMiner II: A Data Mining Application - by Chan Ip Fa (Eva), Lou Wai Fong (Thomas)
Abstract: Data mining, also known as knowledge discovery (KDD), is the process of extracting hidden predictive information and patterns from large amount of data. It automated the processes that traditionally required extensive hands-on analysis. The core of data mining is the data mining algorithms. They use data provided to build a descriptive model that describing the natural of the data itself or a predictive model that can be used to forecast unseen data. In this report, LiteMiner Version 2, which is the continuation of LiteMiner Version 1, is introduced. Throughout this report, the term LiteMiner 1 is refered to LiteMiner Version 1 while LiteMiner 2 is refered to LiteMiner Version 2. LiteMiner 1, is a data mining tools, which provides several functions such as pre-processing, apply data mining algorithms on data and to apply predictive models on unseen data. LiteMiner 1 contains seven data mining algorithms which include C4.5 decision tree, multi-layer perceptron, multi-linear regression, Naïve Bayes classifier, K-Means clustering, EM clustering and Apriori. In LiteMiner 2, the four new algorithms and data visualization function are added. The four new algorithms added are Multiple Logistic Regression, Logistic Model Tree, Density-based Fuzzy Clustering and Tertius. The data visualization function provides a visualization of the analyzed result. It includes some bar graph, pie chart, doughnut chart, tree view chart and some other specific graph for the corresponding algorithms. The quality of LiteMiner 2 is compared to other data mining software called Weka which is an open source software implemented in Java by University of Waikato. The compared results are similar and satisfactory.
Marketing Analysis and Management System - by Wong Cheong Fok (Kent), Wong Man Seng (Aaron)
Abstract: This is a marketing management and analysis system. It was developed for a middle-small selling company. The system is easy to operate, have fully sale’s functions, several useful management functions and several reports. This system combines sale, management and analysis functions. It has 5 modules: Customer, Product, Sale, Purchase and Report. Customer: Manage customer and deliver man information. It can show credit status of customers as well as the bucket status. The deliver man function controls the information of the staff in charge of the delivery of the products as well as generate the ranking of deliveries done. Product: Manage products, can see product’s information and show all products to ranking them. Find out the most sell product. Sale: The system’s core part. Make an order and record down all transactions to the Database. The Report function made use of these data to generate the reports. Purchase: Manage the purchases of products from the suppliers. Report: Generate different reports of customers, products, and credit. In the report part, data can be even exported to MS Excel. This system can handle the sales operation, control of the inventory, the purchase from suppliers, customer’s credit. Those records will be manage by the system and provide useful reports to the management.
2005
LiteMiner: A Data Mining Application - by Ho Wen Kin (Raymond), Cheang Hoi Key (Jacky), Wong Tin Lok (Titus)
Abstract: Data mining, also known as knowledge discovery (KDD), is the process of extracting hidden predictive information and patterns from large amount of data. It automated the processes that traditionally required extensive hands-on analysis. The core of data mining is the data mining algorithms. They use data provided to build a descriptive model that describing the natural of the data itself or a predictive model that can be used to forecast unseen data. In this report, a data mining software, named as LiteMiner, is proposed. LiteMiner provides several functions such as pre-processing, applying data mining algorithms on data and to applying predictive models on unseen data. Seven of the most popular data mining algorithms were implemented in LiteMiner. They are C4.5 decision tee, multi-layer perceptron, multi-linear regression, Naïve bayes classifier, K-Means clustering, EM clustering and Apriori. The algorithms are tested using well known datasets used by researchers in this area. We had also compared our results with a famous application: WEKA which is an open source data mining software implemented in Java by University of Waikato and the results were satisfactory.
Air Cargo Scheduling Application - by Chan Nga Chun (Nesta), He Kan (Kenneth), Lai Chi Ieng (Kenson) [co-supervised with Dr. Simon Fong].
Abstract: This paper is to optimize the scheduling of the packages with the aircrafts’ loading capacities, which are simulated. The optimization criteria are evaluated by customer satisfaction and maximize the usage and profit of the aircrafts. Three algorithms for the batch delivery scheduling problem are developed to find the optimal air cargo shipment. These algorithms are genetic algorithm with earliest due date method, extended due date method and genetic algorithm with extended due date method. The performances of these algorithms are compared to first come first serve and earliest due date scheduling method. The performance of genetic algorithm is analyzed by its fitness function. Air cargos which are handled within Chinese cities is based on flight schedules of nine airline companies including Air Macau, EVA Airways, Cathay Pacific, China Southern Airlines, China Eastern Airlines, Air China, Dragon Air, China Airlines and Mandarin Airlines. The flight schedule information is available on public and the web sites of these airlines are listed in the reference page.
2004
Ease-Form - An Online E-Form System - by Fan Chi Kong (Clark), Chang Kam Seng (Arcturus)
Abstract: Traditional paper forms are being replaced, or complemented, by eletronic forms. Everything you can do with a traditional paper form you can do with an e-form like mailing or signing. Besides, e-forms have several advantages such as reducing printing services, saving time to to distribute and collect forms, reducing rrors in input and so on. E-form solutions increase productivity, improve service and reduce operating costs for a variety of different forms. As our final year project, we decided to develop a Web-based Electronic Form System. Our system provides three main parts: Form Generator, Form Viewer and web application system facility.
2002
EC-Dict - A Palm English and Chinese Dictionary - by Lam Heng Un (Jason), Lok Chi Keung (Danny) - [main supervisor: Dr. Robert P. Biuk-Aghai]
Abstract: EC-Dict is a Chinese English Dictionary that running on the Palm Operating System. The whole dictionary contains two difeerent parts, one is the data file and the other is the dictionary main program. Dictionary data file stored all the dictionary information, like the word, Pinyin, the word's defintion and the strcuture of managed data. The dictionary main program is an executable file, it can run directly in the Palm Operating System. Inside the dictionary main program, it has a searching function, different user interface and it supports Taditional Chinese and English.