Mohamad Farhan Mohamad Mohsin

Design and development of portable RFID for attendance system

This paper describes an ongoing project for recording examination attendance using Radio Frequency Identification (RFID). The project is carried out to test in a university, where the system which is named Portable Examination Attendance System (PEAS) is integrated with the existing system for record extraction. The use of RFID technology enables the university management to avoid attendance forms from damages such as tear, lost, and misplaced. This paper describes about the design and development of PEAS in terms of hardware technology and software. In addition, some related works are reviewed and addressed to support this project. As a conclusion, this paper states some future works of this project.

Mining the student programming performance using rough set

One of the powerful data mining analysis is it can generates different set of knowledge when similar problem is presented to different data mining techniques. In this paper, a programming dataset was mined using rough set in order to investigate the significant factors that may influence students programming performance based on information from previous student performance. Then, the result was compared with other researches which had previously explored the data using statistic, clustering, and association rule. The dataset consists of 419 records with 70 attributes were pre-processed and then mined using rough set. The result indicates rough set has identified several new characteristics. The student who has been exposed to programming prior to entering university and obtained average score in Mathematics, English, and Malay Language subject during secondary Malaysian School Certificate (SPM) examination were among strong indicators that contributes to good programming grades. Besides that, the personality factor; the investigative and social type plus average cognitive person were also found as important factors that influence programming. This finding can be a guideline for the faculty to plan teaching and learning program for new registered student.

The preliminary design of outbreak detection model based on inspired immune system

The aim of outbreak detection model is to obtain high detection rate but at the same time maintaining the low false alarm rate. However, because of the weakness of early outbreak signal that behaves under uncertainties, it causes imbalance result between detection rate and false alarm rate. In this study, an early review of danger theory, an approach of artificial immune system is performed to seek the possibility it can be applied in outbreak detection. To investigate that, a detection model based on danger theory is developed and the model is tested with dengue outbreak dataset. The model is then evaluated in term of detection rate, specificity, false alarm rate, and accuracy. The preliminary result indicates that the proposed model has ability to generate good detection result with a balance between detection rate and false alarm rate.

The development of hashing indexing technique in case retrieval

Case-based reasoning (CBR) considers previous experience in form of cases to overcome new problems. It requires many solved cases in case base in order to produce a quality decision. Since today, database technology has allowed CBR to use a huge case storage therefore the case retrieval process also reflects the final decision in CBR. Traditionally, sequential indexing method has been applied to search for possible cases in case base. This technique is worked fast when the number of cases is small but it consumes more time to retrieve when the number of data grows in case base. To overcome the weakness, this study researches the nonsequential indexing called hashing as an alternative to mine large cases and faster the retrieval time in CBR. Hashing indexing searches a record by determines the index using only an entrys search key without traveling to all records. This paper presents the review of a literature and early stages of the integration hashing indexing method in CBR. The concept of hashing indexing in case retrieving process, the model development, and the preliminary algorithm testing result will be discussed in this paper.

Comparing the knowledge quality in rough classifier and decision tree classifier

This paper presents a comparative study of two rule based classifier; rough set (R c ) and decision tree (DT c ). Both techniques apply different approach to perform classification but produce same structure of output with comparable result. Theoretically, different classifiers will generate different sets of rules via knowledge even though they are implemented to the same classification problem. Hence, the aim of this paper is to investigate the quality of knowledge produced by R c and DT c when similar problems are presented to them. In this case, four important performance metrics are used as comparison, the accuracy of classification, rules quantity, rules length and rules coverage. Five dataset from UCI Machine Learning are chosen and then mined using R c toolkit namely ROSETTA while C4.5 algorithm in WEKA application is chosen as DT c rule generator. The experimental result shows that R c and DT c own capability to generate quality knowledge since most of the results are comparable. R c outperform as an accurate classifier, produce shorter and simpler rule with higher coverage. Meanwhile, DT c obviously generates fewer numbers of rules with significant difference.

An Evaluation of Feature Selection Technique for Dendrite Cell Algorithm

Dendrite cell algorithm needs appropriates feature to represents its specific input signals. Although there are many feature selection algorithms have been used in identifying appropriate features for dendrite cell signals, there are algorithms that never been investigated and limited work to compare performance among them. In this study, six feature selection algorithms namely Information Gain, Gain Ratio, Symmetrical Uncertainties, Chi Square, Support Vector Machine, and Rough Set with Genetic Algorithm Reduct are examined and their effectiveness to represent dendrite cell signal are evaluated. Eight universal datasets are chosen and assessing their performance according to sensitivity, specificity, and accuracy. From the experiment, the Rough Set Genetic Algorithm reduct is found to be the most effect feature selection for dendrite cell algorithm when it generates a consistent result for all evaluation metrics. In single evaluation metrics, the chi square technique has the best competence in term of sensitiveness while the rough set genetic algorithm reduct is good at specificity and accuracy. In the next step, further analysis will be conducted on complex dataset such as time series data set.

NFC-based Data Retrieval Device

This paper describes the design and development of data retrieval system using near field communication (NFC) protocol to read and transfer data from the device storage panel located at the recycle bin. In existing systems, data are manually collected from the storage device using the SD card and sent for upload to regional workstation of the data center, which is located at the central server. The device automatically establishes the NFC connection with the recycle bin panel and once the connection is established, data will automatically be transferred to the device and the current data storage in the recycle bin will be erased. Next, the collected data will be uploaded to the server through regional workstation.

Experimenting the Dendrite Cell Algorithm for Disease Outbreak Detection Model

The characteristics of early outbreak signal which are weak and behaved under uncertainties has brought to the development of outbreak detection model based on dendrite cell algorithm. Although the algorithm is proven can improve detection performance, it relies on several parameters which need to be defined before mining. In this study, the most appropriate parameter setting for outbreak detection using dendrite cell algorithm is examined. The experiment includes four parameters; the number of cell cycle update, the number of dendrite cell allowed to be in population, weight, and migration threshold value. To achieve that, an anthrax disease outbreak is chosen as a case study. Two artificial anthrax datasets known as WSARE7 and WSARE58 are taken as experiment data. The experiment is measured based on five metrics; detection rate, specificity, false alarm rate, accuracy, and time taken to produce result. Besides that, a comparison is made with Cumulative Sum, Exponentially-weighted Moving Average, and Multi Layer Perceptron. From the experiment, the best parameter setting for anthrax outbreak using dendrite cell algorithm is identified whereby it proven can helps the model to produce a good detection result between detection rate and false alarm rate. Since each outbreak disease carries different outbreak characteristic, the parameter setting for different outbreak might be different.

Discovering Web Server Logs Patterns Using Generalized Association Rules Algorithm

With the explosive growth of data available on the World Wide Web (WWW), discovery and analysis of useful information from the World Wide Web becomes a practical necessity. Data Mining is primarily concerned with the discovery of knowledge and aims to provide answers to questions that people do not know how to ask. It is not an automatic process but one that exhaustively explores very large volumes of data to determine otherwise hidden relationships. The process extracts high quality information that can be used to draw conclusions based on relationships or patterns within the data. Using the techniques used in Data Mining, Web Mining applies the techniques to the Internet by analyzing server logs and other personalized data collected from customers to provide meaningful information and knowledge. Web access pattern, which is the sequence of accesses pursued by users frequently, is a kind of interesting and useful knowledge in practice (Pei, 2000). Today web browsers provide easy access to myriad sources of text and multimedia data. With approximately 4.3 billion documents online and 20 million new web pages published each day (Tanasa and Trousse, 2004), more than 1 000 000 000 pages are indexed by search engines, and finding the desired information is not an easy task (Pal et al., 2002). Web Mining is now a popular term of techniques to analyze the data from World Wide Web (Pramudiono, 2004). A widely accepted definition of the web mining is the application of data mining techniques to web data. With regard to the type of web data, web mining can be classified into three types: Web Content Mining, Web Structure Mining and Web Usage Mining. As an important extension of data mining, Web mining is an integrated technology of various research fields including computational linguistics, statistics, informatics, artificial intelligence (AI) and knowledge discovery (Fayyad et al., 1996; Lee and Liu, 2001). Srivastava et al. (2002) classified Web Mining into three categories: Web Content Mining, Web Structure Mining, and Web Usage Mining (see Figure 1). 11