Rohana Mahmud

An application of case-based reasoning with machine learning for forensic autopsy

We modeled the CBR technique of forensic autopsy report preparation.The CBR model was coupled with Naive Bayes learner for feature weight learning and also the outcome prediction.Feature weight learning improves the CBR system accuracy.The outcome prediction is improved with Naive Bayes prediction. Case-based reasoning (CBR) is one of the matured paradigms of artificial intelligence for problem solving. CBR has been applied in many areas in the commercial sector to assist daily operations. However, CBR is relatively new in the field of forensic science. Even though forensic personnel have consciously used past experiences in solving new cases, the idea of applying machine intelligence to support decision-making in forensics is still in its infancy and poses a great challenge. This paper highlights the limitation of the methods used in forensics compared with a CBR method in the analysis of forensic evidences. The design and development of an Intelligent Forensic Autopsy Report System (I-AuReSys) basing on a CBR method along with the experimental results are presented. Our system is able to extract features by using an information extraction (IE) technique from the existing autopsy reports; then the system analyzes the case similarities by coupling the CBR technique with a Naive Bayes learner for feature-weights learning; and finally it produces an outcome recommendation. Our experimental results reveal that the CBR method with the implementation of a learner is indeed a viable alternative method to the forensic methods with practical advantages.

Towards knowledge modeling and manipulation technologies

We identified different knowledge base modelling and manipulation techniques based on 4 categories.Compared knowledge base modelling and manipulation technologies based on their underlying theories, knowledge representation technique, knowledge acquisition technique, challenges, applications, development tools and development languages.We discussed the relevance of knowledge-based business.We proposed a promising technique for knowledge-based business management and other knowledge related applications. A system which represents knowledge is normally referred to as a knowledge based system (KBS). This article focuses on surveying publications related to knowledge base modelling and manipulation technologies, between the years 20002015. A total of 185 articles excluding the subject descriptive articles which are mentioned in the introductory parts, were evaluated in this survey. The main aim of this study is to identify different knowledge base modelling and manipulation techniques based on 4 categories; 1) linguistic knowledge base; 2) expert knowledge base; 3) ontology and 4) cognitive knowledge base. This led to the proposition of 8 research questions, which focused on the different categories of knowledge base modelling technologies, their underlying theories, knowledge representation technique, knowledge acquisition technique, challenges, applications, development tools and development languages. A part of the findings from this survey is the high dependence of linguistic knowledge base, expert knowledge base and ontology on volatile expert knowledge. A promising technique for knowledge-based business management and other knowledge related applications is also discussed.

An architecture for Malay Tweet normalization

Research in natural language processing has increasingly focused on normalizing Twitter messages. Currently, while different well-defined approaches have been proposed for the English language, the problem remains far from being solved for other languages, such as Malay. Thus, in this paper, we propose an approach to normalize the Malay Twitter messages based on corpus-driven analysis. An architecture for Malay Tweet normalization is presented, which comprises seven main modules: (1) enhanced tokenization, (2) In-Vocabulary (IV) detection, (3) specialized dictionary query, (4) repeated letter elimination, (5) abbreviation adjusting, (6) English word translation, and (7) de-tokenization. A parallel Tweet dataset, consisting of 9000 Malay Tweets, is used in the development and testing stages. To measure the performance of the system, an evaluation is carried out. The result is promising whereby we score 0.83 in BLEU against the baseline BLEU, which scores 0.46. To compare the accuracy of the architecture with other statistical approaches, an SMT-like normalization system is implemented, trained, and evaluated with an identical parallel dataset. The experimental results demonstrate that we achieve higher accuracy by the normalization system, which is designed based on the features of Malay Tweets, compared to the SMT-like system.

Using an edutainment approach of a Snake and Ladder game for teaching Jawi script

Due to rapid socio-economic changes in Malaysia, the traditional way of writing the Malay Language using Arabic characters is fading. Studies have shown that this representation, called Jawi is becoming extinct due to its difficulty in learning and teaching to the younger students. The paper proposed a multimedia edutainment game based on traditional Snake and Ladder game to teach the Jawi scripts. The aim is to provide an interactive and an engaging way to teach Jawi, in a manner it can evoke interest to primary school students. The game is tested with Year One students and their Jawi language teachers in one of the Malaysian Primary school. The result shows that students interest level in learning Jawi has increased using the proposed edutainment approach.

Issues of coherence analysis on English translations of Quran

This study shows issues of comparing English translations of Holy Quran and its Arabic text from discourse structure perspective. There are several different translations of Quran, which differ in structure and word domain. In these translations, the order of sentences, phrases, and words is different, which affects computational text analyzing results. It is a new idea to study translations of Quran from entity coherence and lexical cohesion point of view, as a method for evaluating the accuracy and equivalence of existing translations. The results of this study even can be used for machine translation in the future. This research is a preliminary stage of investigating the issues, constructing a platform and defining of some preliminary rules for comparing and evaluating discourse structure of translations.

Adaptive feedback in computer-based learning environments: a review

Adaptive support within a learning environment is useful because most learners have different personal characteristics such as prior knowledge, learning progress, and learning preferences. This study reviews various implementation of adaptive feedback, based on the four adaptation characteristics: means, target, goal, and strategy. This review focuses on 20 different implementations of feedback in a computer-based learning environment, ranging from multimedia web-based intelligent tutoring systems, dialog-based intelligent tutoring systems, web-based intelligent e-learning systems, adaptive hypermedia systems, and adaptive learning environment. The main objective of the review is to compare computer-based learning environments according to their implementation of feedback and to identify open research questions in adaptive feedback implementations. The review resulted in categorizing these feedback implementations based on the students’ information used for providing feedback, the aspect of the domain or pedagogical knowledge that is adapted to provide feedback based on the students’ characteristics, the pedagogical reason for providing feedback, and the steps taken to provide feedback with or without students’ participation. Other information such as the common adaptive feedback means, goals, and implementation techniques are identified. This review reveals a distinct relationship between the characteristics of feedback, features of adaptive feedback, and computer-based learning models. Other information such as the common adaptive feedback means, goals, implementation techniques, and open research questions are identified.

A Cognitive Knowledge-based Framework for Adaptive Feedback

Adaptive learning environments provide personalization of the instructional process based on different parameters such as: sequence and difficulty of task, type and time of feedback, learning pace and others. One of the key feature in learning support is the personalization of feedback. Adaptive feedback support within a learning environment is useful because most learners have different personal characteristics such as prior knowledge, learning progress and learning preferences. In a computer-based learning environment, feedback is considered as one of the most effective factors which influence learning. Although, there are various tools that provide adaptive feedback in learning environments, some problems still exist. One of the problems we are looking into is How to design effective tutoring feedback strategies? We propose a cognitive knowledge based framework for adaptive feedback, which combines the three facets of knowledge (pedagogical, domain and learner model) in a learning environment, using concept algebra.

Design of an Algebraic Concept Operator for Adaptive Feedback in Physics

In an adaptive learning environment, the feedback provided during problem-solving requires a means, target, goal, and strategy. One of the challenges of representing feedback to meet these criteria, is the representation of the effect of multiple concepts on a single concept. Currently, most of the methods (linguistic knowledge base, expert knowledge base, and ontology) used in representing knowledge in an adaptive learning environment only provide relationships between a pair of concept. However, a cognitive knowledge base which represents a concept as an object, attribute, and relations (OAR) model, provides a means to determine the effect of multiple concepts on a single concept. Using the OAR model, the relationships between multiple pedagogical, domain, and student attributes are represented for providing adaptive feedback. Most researchers have proposed adaptive feedback methods that are not fully grounded in pedagogical principles. In addition, the three knowledge components of the learning environment (pedagogical, domain and student models) are mostly treated in isolation. A reason for this could be the complex nature of representing multiple adaptive feedback characteristics across the main components of a learning environment. Thus, there is a need to design a concept operator that can relate the three facets of knowledge in an adaptive learning environment. Using the algebraic concept operator \( R_{i}^{in} \), the effect of multiple attributes of the three knowledge components on the student’s performance is represented. The algebraic concept operator introduced in this article will allow teachers and pedagogy experts to understand and utilize a variety of effective feedback approaches.

Ontology population: Approaches and design aspects

Ontologies provide a means to store knowledge in a machine-readable format. Ontology population is the task of updating an ontology with new facts from an input knowledge resource. These facts are represented in a structured format and integrated thereafter into the existing knowledge in the ontology. Textual resources are the dominant online knowledge resources that contain a large number of facts expressed either explicitly or implicitly. Hence, the automatic processing of the extensive knowledge available in these resources has recently gained increasing interest. This study discusses the major components of ontology population process and the different design aspects to be considered when building ontology population systems. In addition, this research explains the different approaches and techniques adopted to carry out the task of ontology population. The possible choices of the design aspects and the related issues are identified and analysed using a set of representative ontology population systems. This study concludes by describing the remaining open issues that should be further explored in ontology population.

Benchmarking Mi-AR: Malay anaphora resolution

This paper discusses benchmarking a new approach that uses the maximum entropy model and Random Forest classifier for pronominal anaphora resolution in the Malay language. Given the specific characteristics of the Malay language, such as gender-neutral pronouns, we pursued a specific two-phase methodology: (1) conducting analyses to scrutinize the features of Malay anaphors, and (2) designing, implementing, and evaluating a pronominal resolution system based on the analysis results. The approach achieved a 0.84 F-measure in testing with 9,779 tokens (i.e. 50 news and 50 non-news articles), and the results of our experiment and comparison study show that the presented approach significantly outperforms the current state-of-the-art Malay anaphora resolution systems.