M. Ali Fauzi

Eye Movement as Navigator for Disabled Person

Eyes is one of human organs which mostly still functions properly in disabled people when other parts of the body are disabled. This research propose a new framework to recognize and detected eye movement for handling position by considering the decision of both left and right eye. The sophisticated algorithm, Haar Cascade Algorithm was used for observing the area of eyes, then thresholding image using morphology is used to obtain the focus of eyes. The Hough Circle Transform with several rules could decide the handling position of eye movement. The performance of the pro-posed algorithm could reach over 80% in all dataset.

A laser-vision based obstacle detection and distance estimation for smart wheelchair navigation

The aim of the research is to present an approach of obstacle distance estimation and navigation for smart wheelchair. The smart wheelchair is an electric wheelchair equipped with camera and line laser to navigate and avoid an obstacle. The camera was used to capture images from the environment to sense the pathway condition. The line laser was used in combination with camera to recognize an obstacle in the pathway based on the shape of line laser image in certain angle. A blob method detection was applied on the line laser image to recognize the pattern of the detected obstacles. The line laser projector and camera were mounted in fixed-certain position to make sure a fixed relation between blobs-gaps and obstacle-to-wheelchair distance. A simple linear regression from 16 obtained data was used to represent this relation as the estimated obstacle distance. As a result, the average error between the estimation and actual distance was 1.25 cm from 7 data testing experiments. The experiments result indicates that the proposed method is able to estimate well the distance between wheelchair and obstacle. Later, the smart wheelchair needs to decide further action whether it should turn left, right or just walk straight when facing certain obstacle to avoid it.

Adaptive human tracking for smart wheelchair

People with impairment and having difficulties to walk, even impossible to move due to illness, injury, or disability need assistance tool. One assistance tool to help those people is wheelchair. With current technological developments, conventional wheelchair can be improved. Conventional wheelchair which operated by hand cannot be used by people with hand-foot impairment, as well as electric-powered wheelchair that need to be controlled with hand. For those with hand-foot impairment, conventional wheelchair can be assisted by assistant to help pushing and to maneuver. One drawback with this approach is the assistant will have limited movement and will have fatigue from pushing a wheelchair. This research try to overcome this drawback so that the wheelchair can move semi-autonomously. Proposed approach incorporates human tracking algorithm that later will be used to make the wheelchair moving independently without assistant to push from behind. This paper propose a framework that combines keypoint descriptors for human tracking: ORB, KAZE, AKAZE, BRISK, SIFT, and SURF. Each keypoint descriptors are given a score which is used to choose which descriptor is used until the minimum number of keypoints is fulfilled. If the best in the method list does not suffice, then the second best will be selected to generate keypoints, and so on. The result of the framework obtained high precision, 0.93 and 0.89 from two videos with different environments.

Comparative analysis of string similarity and corpus-based similarity for automatic essay scoring system on e-learning gamification

Essay assessment within e-learning need to be conducted manually by human expert. This process takes time and costly. Hence, automatic essay scoring is needed. Since the scoring system will be integrated to the e-learning, we need a computationally lightweight method that still does not rule out the accuracy of the assessment. In this paper, we propose an automatic scoring system for essay examination using unsupervised approaches. We compare and analyze two similarity measure methods, cosine similarity and latent semantic analysis. The parameters that was used to measure the performance of the methods are the computational complexity — measured by the amount of CPU and memory usage, and page load time — and accuracy — measured by Pearson Correlation and Mean Absolute Error. The results showed that both algorithm consumed same amount of memory. For CPU usage, LSA consumption is 0.13% and cosines is 0.06%. For page load time, cosine similarity is faster than LSA which is 0.2 second and 0.5 second consecutively. Based on the correlation measure with Pearson, LSA is more superior to the cosine similarity by 0.59 to 0.49. LSA also has less MAE than cosine similarity which is 5.69 compared to 5.33. From that result, LSA and Cosine Similarity has a very competitive result in accuracy. However, Cosine has a better server performance so that preferred to be implemented in e-learning automatic essay scoring system.

Improving Sentiment Analysis of Short Informal Indonesian Product Reviews using Synonym Based Feature Expansion

Sentiment analysis in short informal texts like product reviews is more challenging. Short texts are sparse, noisy, and lack of context information. Traditional text classification methods may not be suitab le for analyzing sentiment of short texts given all those difficulties. A common approach to overcome these problems is to enrich the original texts with additional semantics to make it appear like a large document of text. Then, traditional classification methods can be applied to it. In this study, we developed an automatic sentiment analysis system of short informal Indonesian texts using Naïve Bayes and Synonym Based Feature Expansion. The system consists of three main stages, preprocessing and normalization, features expansion and classification. After preprocessing and normalization, we utilize Kateglo to find some synonyms of every words in original texts and append them. Finally, the text is classified using Naïve Bayes. The experiment shows that the proposed method can improve the performance of sentiment analysis of short informal Indonesian product reviews. The best sentiment classification performance using proposed feature expansion is obtained by accuracy of 98%.The experiment also show that feature expansion will give higher improvement in small number of training data than in the large number of them.

Ensemble Method for Indonesian Twitter Hate Speech Detection

Received Dec 04, 2017 Revised Jan 11, 2018 Accepted Apr 15, 2018 Wireless Body Area Networks (WBANs) are fundamental technology in health care that permits the information of a patient’s essential body parameters to be gathered by the sensors. However, the safety and concealment defense of the gathered information is a key uncertain problem. A Hybrid Key Management (HKM) scheme [13] is worked based on Public Key Cryptography (PKC)-authentication scheme. This scheme uses a oneway hash function to construct a Merkle Tree. The PKC method increase the computational complexity and lacking scalability. Additionally, it increases expensive computation, communication costs and delay. To overcome this problem, Robust Security for Protected Health Information by ECC with signature Hash Function in WBAN (RSP) is proposed. The system employs hash-chain based key signature technique to achieve efficient, secure transmission from sensor to user in WBAN. Moreover, Elliptical Curve Cryptography algorithm is used to verifies the authenticate sensor. In addition, it describes the experimental results of the proposed system demonstrate the efficient data communication in a network.A Weblogs contains the history of User Navigation Pattern while user accessing the websites. The user navigation pattern can be analyzed based on the previous user navigation that is stored in weblog. The weblog comprises of various entries like IP address, status code and number of bytes transferred, categories and time stamp. The user interest can be classified based on categories and attributes and it is helpful in identifying user behavior. The aim of the research is to identifying the interested user behavior and not interested user behavior based on classification. The process of identifying user interest, it consists of Modified Span Algorithm and Personalization Algorithm based on the classification algorithm user prediction can be analyzed. The research work explores to analyze user prediction behavior based on user personalization that is captured from weblogs.Wireless Body Area Networks (WBANs) are fundamental technology in health care that permits the information of a patient’s essential body parameters to be gathered by the sensors. However, the safety and concealment defense of the gathered information is a key uncertain problem. A Hybrid Key Management (HKM) scheme [13] is worked based on Public Key Cryptography (PKC)-authentication scheme. This scheme uses a oneway hash function to construct a Merkle Tree. The PKC method increase the computational complexity and lacking scalability. Additionally, it increases expensive computation, communication costs and delay. To overcome this problem, Robust Security for Protected Health Information by ECC with signature Hash Function in WBAN (RSP) is proposed. The system employs hash-chain based key signature technique to achieve efficient, secure transmission from sensor to user in WBAN. Moreover, Elliptical Curve Cryptography algorithm is used to verifies the authenticate sensor. In addition, it describes the experimental results of the proposed system demonstrate the efficient data communication in a network.

Automatic Essay Scoring System Using N-Gram and Cosine Similarity for Gamification Based E-Learning

E-Learning is one of the great innovations in teaching methods. In the E-learning, there are several assessment methods; one of them is the essay examination. Essay assessment takes a long time if corrected manually. Therefore, researches on automatic essay scoring have been growing rapidly in recent years. The method that is usually used for automatic essay scoring is Cosine Similarity by utilizing bag of words as the feature extraction. However, the feature extraction by using bag of words did not consider to the order of words in a sentence. Meanwhile, the order of words in an essay has an important role in the assessment. In this study, an automatic essay scoring system based on n-gram and cosine similarity was proposed. N-gram was used for feature extraction and modified to split by word instead of by letter so that the word order would be considered. Based on evaluation results, this system got the best correlation of 0.66 by using unigram on questions that do not consider the order of words in the answer. For questions that consider the order of the words in the answer, bigram has the best correlation value by 0.67.

Development of computer vision based obstacle detection and human tracking on smart wheelchair for disabled patient

People with physical disability such as quadriplegics may need a device which assist their mobility. Smart wheelchair is developed based on conventional wheelchair and is also generally equipped with sensors, cameras and computer based system as main processing unit to be able to perform specific algorithm for the intelligent capabilities. We develop smart wheelchair system that facilitates obstacle detection and human tracking based on computer vision. The experiment result of obstacle distance estimation using RANSAC showed lower average error, which is only 1.076 cm compared to linear regression which is 2.508 cm. The average accuracy of human guide detecting algorithm also showed acceptable result, which yield over 80% of accuracy.

Optimizing K-means text document clustering using latent semantic indexing and pillar algorithm

Document clustering is an important tool to help managing the vast amount of digital text document. This paper introduces a new approach to cluster text document. First, text is preprocessed and indexed using inverted index. Then the index is trimmed using TF-DF thresholding. After that, Term Document Matrix is built based on TF-IDF. Next step uses Latent Semantic Indexing to extract important feature from Term Document Matrix. The following process is selecting seeds via Pillar algorithm. Based on determined seeds, K-Means clustering is performed. Experiment result proves that this approach outperforms standard K-Means document clustering.