Search

Implementation of digital technology in healthcare often fails, especially when involving patients, because it should be supported by the community's readiness to adopt the technology. This study describes patients' internet behaviour, eHealth literacy and its correlation with patients' digital health adoption in primary health care (PHC). This study had 800 samples from 8 of 37 PHC in Semarang, Indonesia. The survey distributed a questionnaire to measure internet behaviour, eHealth literacy, and patient adoption in online registration in 2019. The questionnaire modified the previous Computer Email, Web (CEW) questionnaire and eHealth literacy scale (eHeals). This study reveals that age had a negative correlation in anxiety $(\mathrm{r}=-0.204,\ {,} \mathrm{p}=0.000)$ and intention $(\mathrm{r}=0.000, \mathrm{p} =-0.196)$ using online registration. eHealth literacy correlates with the anxious $(\mathrm{p}=0.01,\ \mathrm{r}=-0,085)$ and intention toward online registration $(\mathrm{p}=0.00,\ \mathrm{r}=0.334)$ . Patients should have digital health literacy to adopt digital health in PHC, and patients should get educated before implementing digital health.

South Sumatra has a great potential for renewable energy development, including solar energy. South Sumatra enjoys year-round sunshine due to its location on the equator. South Sumatra has open-pit coal mining, which creates open land suitable for solar power plant development. Given that the open-pit mining area is exposed to mining dust, this potential must be continuously explored and investigated to determine if the benefits outweigh the drawbacks. Environmental conditions strongly influence electrical generation via the photovoltaic effect, where shading and partial shading due to dust and soil deposition can reduce PV panel performance. This paper is part of research to study the feasibility of harvesting solar energy in open-pit mining. The experiments were conducted in six months; starting from June, the cleaned PV panel had higher average power and efficiency of 1.85 W and 0.94% than the soiled panel. The same condition also occurred in July (2.05 W and 1.06% higher), August (3.63 W and 1.30%), and December; it is 3.2 W and 1.57% higher than the soiled panel. The six-month experiments show that the collection of dust and soil can reduce the power output and efficiency; however, with regular cleaning maintenance, installing solar farms in an open-pit mining area is beneficial due to open space without shading and higher irradiance potential.

Massive natural disaster not only causing big casualty and loss property, but it may interference or damage the infrastructure of communication network. Since communications system is needed on victim search and rescue process, thus we proposed an Interplanetary Overlay Network on Delay Tolerant Network (ION - DTN) as transmission network based on Feedback Unmanned Aerial Vehicle (UAV) system for emergency communication during natural disaster. Delay Tolerant Network (DTN) can be one of suitable solution for intermittent connectivity. DTN is a network that designed for the challenging conditions, where the Internet Protocol Suite could not handle it properly. QoS test results on the implementation of DTN network in this project, shows that the best Delivery Probability is 1, the smallest Delay is 37.7 second, the smallest jitter is 0.534 second, and the biggest throughput is 69.5 kbps.

Hate speech is a form of communication which contains hatred by doing things, such as inciting, insulting, disparaging, or demeaning a person or group. Hate speech issues in Indonesia often have linkages to politics. In 2018 and 2019, for example, the hate speech relates to the local leader and presidential elections. The hate speech actors commonly use social networks, such as Instagram, to spread their hatred words. About 60% of hate speech is found in the comments of the posts and it will be a real threat if not quickly detected. Our study aims to detect hate speech in Instagram comments. We propose the use of a word2vec method with skip-gram models and a modified TextCNN to learn and detect hate speech texts. Furthermore, random oversampling, random under sampling, and class weight was used to solve imbalanced dataset problems. The results show that the best accuracy, in term of F-score, is 93.70%, gained from a combination of word2vec skip-gram with window size 15, a modified TextCNN, and random oversampling methods.

The quality of the internship activities can contribute to the participants as experiences it can be used as provisions for student self-development. The Good management consistency needs to be done by the school, so the implementation of the apprenticeship can run as well and provide valuable experience. Moreover, good management certainly requires careful and measured planning. The preparation of participants have to able to adjust the conditions of the internship before implementation needs to considered, so the participants can adaptation with the new environment. The Architectural modeling have to be able on handling the apprenticeship management system on the create right model on solving problem. Through out the Framework test results applied to the apprenticeship data management model, the Zachman Framework which was applied in the development of the apprentice management system architecture model proved to able a positive solution for the implementation of the apprenticeship model. This is determined based on the results of user testing where 73.79% expressed satisfaction with the internship service that is the correlation value between the variable X to Y has a degree of closeness of 0.512 which has a strong meaning.

The aim of developing Tengger ethnic culinary training materials is to develop online training teaching materials for the HORECA industry manager. Training materials consist of Tengger cuisine recipes as a reference in food and beverage services for HORECA guests. Teaching material is the result of Tengger ethnic culinary standardization as one of the culinary tourism or eco-culinary tourism around the TNBTS tourist area. The method used to develop Tengger culinary training materials was ADDIE model consisting of five steps, while the learning targets were managers of the HORECA industry in the region of Tengger. The results of this development are: 1) analysis phase of teaching materials, including the classification of Tengger local culinary based on the results of the research survey which is classified into staple foods, vegetable dishes, side dishes, condiments/sambal and snacks; 2) design phase, based on the material analysis phase by setting training objectives using a life-based learning approach. The learning resources used are in the form of text (pdf), videos and online learning called "kulinera training"; and 3) development phase, which is the compilation of Tengger culinary training materials as a training service from instructures to create and maintain online-based learning.

Assessment is an important part of learning that should be not only to assess students’ knowledge, but also any misconceptions they might make. The teacher may have difficulty in identifying it because of the diversity of students’ abilities. To solve this problem, we propose an Intelligent Tutoring System (ITS) that offers adaptive exercises. In adaptive exercises, the next problem to be solved by a student is chosen by considering her performance during learning. Unlike former ITS that is mostly based on students’ knowledge, the proposed ITS uses students’ knowledge and misconceptions to perform adaptation. This paper discusses the implemented learning scenario and domain model structure to support adaptive exercises. Furthermore, it discusses information about the student recorded in the student model and how it is organized and inferred to find the most appropriate problem.

This paper proposes an approach to following good-learners' paths in Intelligent Tutoring Systems (ITS). Based on a theory which states that learning can be acquired through imitation of competent models, we observed and analyzed the exercises completed by good learners to predict an appropriate sequence of questions for the students to follow. We applied the Hidden Markov Model to represent students' skills and the generative Markov Decision Process to model the exercise experience of good learners. The resulting model represents recommendations of dynamic sequences of questions. A preliminary study has been conducted to elicit a real-world dataset of a programming course. It aims to measure the reliability of the proposed approach, in comparison with the conventional approach where teachers determined the questions. The approach performance is indicated by increased percentages of good learners. The experiment results show that the proposed approach has a better reliability than the conventional approach.

Until recently, recommender systems have been widely applied. Commonly, a single account is used by only one user when interacts with a recommender system. In fact, it is possible that users share their accounts with other users, for example, a single shopping account in an online book store is used by three users in a household. Most recommender systems fail where multiple preferences are mixed in one shared account without contextual information for splitting the shared account. This paper discusses our study and implementation of COVER disambiguating item-based algorithm, a solution for shared account problems because of the absence of contextual information. It applies an item-based top-N collaborative filtering algorithm as a base algorithm. The algorithm aims to improve the item-based top-N collaborative filtering algorithm for tackling the generality, dominance, and presentation problems of the shared account. It has been tested using BookCrossing and Amazon Review datasets and it generates recommendations based on binary and positive-only feedback. To conclude, the proposed COVER disambiguating item-based algorithm has been able to reduce the score of fraction at zero recall. In addition, it can increase the identifiability score of recommendations, in comparison with the item-based top-N collaborative filtering algorithm.

ICT is well known in the information society. The influence of ICT is like a pressure on life that impossible to be ignored. ICT appears in the habits of the community to communicate, interact, information sharing, especially in school community. The utilization of information and communication technology (ICT) in the school community has been extensively stated by many prominent studies. However, many of them show that there is no single universal model suitable to be adopted by all schools. This paper presents a modeling of school ICT utilization that is built by statistical causality test that reflects from empirical schools' cases. Testing the relationships among seven components of ICT utilization in schools is performed by using Structural Equation Modeling (SEM). The seven components get their respective path coefficient levels. each hypothesis is based on research references and expert judgement. The model shows seven principal components as building which reflected from ICT utilization in schools.

Collaborative learning is widely applied in education. One of the key aspects of collaborative learning is group formation. A challenge in group formation is to determine appropriate attributes and attribute types to gain good group results. This paper studies the use of an improved ant colony system (ACS), called Multi Objective Ant Colony System (MOACS), for group formation. Unlike ACS that transforms all attribute values into a single value, thus making any attributes are not optimally worth, MOACS tries to gain optimal values of all attributes simultaneously. MOACS is designed for various combinations of attributes and can be used for homogeneous, heterogeneous or mixed attributes. In this paper, sensing/intuitive learning styles (LSSI) and interests in subjects (I) are used in homogeneous group formation, while active/reflective learning style (LSAR) and previous knowledge (KL) are used for heterogeneous or mixed group formation. Experiments were conducted for measuring the average goodness of attributes (avgGA) and standard deviation of goodness of attributes (stdGA). The objectives of MOACS for homogeneous attributes were minimum avgGA and stdGA, while those for heterogeneous attributes were maximum avgGA and minimum stdGA. As a conclusion, MOACS was appropriate for group formation with homogeneous or mixed.

Recommender systems aim to recommend items for certain users based on rating prediction, user preferences, or other parameters. Along with their advantages in guiding users to appropriate items, recommender systems potentially experience a popularity bias. It is a condition where popular items will often be recommended, while non-popular items will never be recommended. Controlling popularity bias is necessary to introduce new, but appropriate, items. It can be done by making a fair proportion of popular and non-popular items to be recommended to users. One of the methods is by applying fairness aware regularization. This method can increase the proportion of medium-tail recommendations of non-popular items, without sacrificing ranking performance on the recommender system. We combined this method with a learning-to-rank algorithm and applied it in a recommender system for e-commerce. Based on the testing results by using a dataset of Amazon review, the method is able to increase the proportion of medium-tail recommendation of non-popular items, up to 54.1%, on the measurement of Average Percentage of Tail (APT) and gain 0.281 on the measurement of Normalized Discounted Cumulative Gain (NDCG) score. In addition, the method is also able to introduce different medium-tail non-popular items to users.

Intrinsic motivation is essential in learning programming. Gamification can be used to increase the intrinsic motivation of students in the learning process. Gamification is a technique of applying game-elements into learning activities. A drawback of gamification is that it is one-size-fit-all, which means that the same gamification element is applied to all students. Hence, it is necessary to apply adaptation that consider the characteristics of students, so that students can get learning materials that suit their needs and abilities. In our research, we have combined adaptive navigation, in the forms of direct guidance, link hiding, link annotation and link generation, and gamification for learning basic algorithm. The results indicate that our approach can increase the intrinsic motivation of students.

One of the challenges in adaptive learning systems is to manage learner models that include learners' profiles and activities and interaction between learners and the learning tools. In the context of lifelong learning, the learner model is a user learning pattern, which can be modelled all the time based on user interaction with various sources. The main issue in lifelong learner model design is what type of information needs to be stored in the learner model and how the information granularity is involved. The type and level of information granularity of a user will determine how appropriate the pedagogical policies are to be provided by the system for a user. Lifelong learner model design should be in line with the principles of lifelong learning, namely sociality, interoperability, and scrutability. This paper discusses our research on learner model ontology for lifelong learning. The goal expected to be achieved is that there is a lifelong learner model ontology to be accessed and updated by adaptive learning systems. This research classifies learners' attributes into two groups, which are static and dynamic attributes. Static attributes tend not to change or very rarely change. They include learner's profile, special needs, aptitude, preference, learning approach, learning style, personality, and cognitive capability. On the other hand, dynamic attributes include learning history and study plan. Through conducting some experiments, it can be concluded that LifeOn supports ubiquitous and lifelong learning, and can be used in adaptive learning systems with various learner models.

Cyberbullying is a repeated act that harasses, humiliates, threatens, or hassles other people through electronic devices and online social networking websites. Cyberbullying through the internet is more dangerous than traditional bullying, because it can potentially amplify the humiliation to an unlimited online audience. According to UNICEF and a survey by the Indonesian Ministry of Communication and Information, 58% of 435 adolescents do not understand about cyberbullying. Some of them might even have been the bullies, but since they did not understand about cyberbullying they could not recognise the negative effects of their bullying. The bullies may not recognise the harm of their actions, because they do not see immediate responses from their victims. Our study aimed to detect cyberbullying actors based on texts and the credibility analysis of users and notify them about the harm of cyberbullying. We collected data from Twitter. Since the data were unlabelled, we built a web-based labelling tool to classify tweets into cyberbullying and non-cyberbullying tweets. We obtained 301 cyberbullying tweets, 399 non-cyberbullying tweets, 2,053 negative words and 129 swear words from the tool. Afterwards, we applied SVM and KNN to learn about and detect cyberbullying texts. The results show that SVM results in the highest f1-score, 67%. We also measured the credibility analysis of users and found 257 Normal Users, 45 Harmful Bullying Actors, 53 Bullying Actors and 6 Prospective Bullying Actors.

One of the issues in collaborative learning is forming groups based on criteria that have been determined before such as grades, learning style, free time, and others. Computer-Supported Group Formation (CSGF) is a research field which purpose is to automate this process so group formation can be done efficiently and effectively. This paper discusses a research on CSGF to form homogeneous groups using a Fuzzy C-Means Clustering method. The parameter used is learning styles according to Felder-Silverman model. The goal of the clustering is that all students can be grouped, without orphan students, the learning styles of learners among all members in each formed cluster are as similar as possible. The proposed method has been applied in two classes of 42 and 39 undergraduate students. The results show that the clustering goals can be achieved.

ICT has been widely known to support educational process in the school environment. In this research, there are components of ICT adoption in education which build the framework. Framework has structures such as components' names, weights, relation amongst them and also their arrangement. This paper contains a summary and continuation of previous research to finalize the framework by expert judgement. Experts with different backgrounds such as school, government, academics, and practitioners in charge of ICT implementation in schools were invited to participate in this study,. The review was executed using Delphi method and run for several days on each review round. The experts' profound experiences enriched the information about ICT adoption in education framework from several angles. The result of this research is a proposed ICT Adoption in Education Framework (IAEF) that can be used as a reference for schools in implementing ICT.

In recent years, with the rapid increases of books, finding relevant books has been a problem. For that, people might need their peers' opinion to complete this task. The problem is that relevant books can be gained only if there are other users or peers have same interests with them. Otherwise, they will never get relevant books. Recommender systems can be a solution for that problem. They work on finding relevant items based on other users' experience. Although research on recommender system increases, there is still not much research that considers user personality in recommender systems, even though personal preferences are really important these days. This paper discusses our research on a hybrid-based method that combines attribute-based and user personality-based methods for book recommender system. The attribute-based method has been implemented previously. In our research, we have implemented the MSV-MSL (Most Similar Visited Material to the Most Similar Learner) method, since it is the best method among hybrid attribute-based methods. The personality factor is used to find the similarity between users when creating neighborhood relationships. The method is tested using Book-crossing and Amazon Review on book category datasets. Our experiment shows that the combined method that considers user personality gives a better result than those without user personality on Book-crossing dataset. In contrary, it resulted in a lower performance on Amazon Review dataset. It can be concluded that user personality consideration can give a better result in a certain condition depending on the dataset itself and the usage proportion.

Social Tagging Systems (STS) are very popular web application such that millions of people join the systems and actively share their contents. These enormous number of users are flooding STS with contents and tags in an unrestrained way in that threatening the capability of the system for relevant content retrieval and information sharing. Recommender Systems (RS) is a known successful method for overcome information overload problem by filtering the relevant contents over the nonrelevant contents. Besides manage folksonomy information, STS also handle social network information of its users. Both information can be used by RS to generate a good recommendation for its users. This work proposes an enhanced method for an existing hybrid recommender system, by incorporating social network information into the input of the hybrid recommender. The recommendation generation process includes Random Walk with Restart (RWR) alongside Content-Based Filtering (CBF) and Collaborative Filtering (CF) methods. Some parameters were introduced in the system to control weight contribution of each method. A comprehensive experiment with a set of a real-world open data set in two areas, social bookmark (Delicious.com) and music sharing (Last.fm) to test the proposed hybrid recommender system. The outcomes exhibit that it can give improvement compared to an existing method in terms of accuracy. The proposed hybrid achieves 24.4% more than RWR on the Delicious dataset, and 53.85% more than CBF on Lastfm dataset.

Web-based learning environment has led to the creation of a big number of digital learning materials that offers various topics. Learners spend much time for browsing and filtering information that suits their needs. Limited time can prevent learners from finding useful learning materials. One of the most successful such technique to solve the problem is recommender system which can select items the user is interested in from a large amount of data. Previous research has attempted to use content-based filtering combined with good learners' ratings method (CBF-GL method). This research proposes an improved method for an existing e-learning recommender system with combination of content-based filtering and collaborative filtering with good learners' ratings (CBF-CF-GL method). Adding a collaborative filtering method intended that only ratings from good learners with certain similarity with the active learner are used in rating recommendations. Experiments were conducted for measuring and comparing the Mean Absolute Error (MAE) scores of CBF-CF-GL and CBF-GL methods using the same dataset. The experiment shows that the CBF-GL method has produced a MAE score of 0.542, while the CBF-CF-GL method has given a MAE score of 0.447. It has shown that the inclusion of collaborative filtering considering good learners' ratings has improved the recommendation accuracy.

Nowadays, learning styles are increasingly incorporated into adaptive learning systems. Previous studies have proved that learning styles can make learning navigation or presentation adaptive to learners' needs. A challenge in adaptive learning is that to present learning materials relevant to learners based on their learning styles, considering there are a huge number of open accessed learning materials in the web. This paper discusses our research on investigating the effect of learning styles in recommending learning materials. Hybrid filtering, which combines collaborative filtering and content-based filtering, is used in recommender systems by taking into account individual competency learners and the similarity with other learners who have learned the learning materials previously. Commonly, the similarity is taken from correlation among learners, such as rating they have given to learning materials. In our work, we combine rating and learning styles similarities to recommend learning materials. We use Felder-Silverman Learning Styles Model (FSLSM). The experiments conducted is quantitative, in which 44 undergraduate students who have taken Algorithm and Programming Basic course are involved. We make a comparison between MAE scores resulted from recommender when it applies collaborative filtering, learning style similarity filtering, or combined collaborative and learning style similarity filtering. The experiment results indicate the prediction score using rating similarity is the best among the three methods.

A recommender system has an important role in decision support by identifying items that will be useful to users in various domains. Development of recommender systems does not only focus on individual users but also groups in various domains. In this paper we present group recommender systems with a collaborative filtering method. Collaborative filtering is a widely used method for recommender systems, easily found in previous studies, that finds similar preferences on the basis of other users' explicit feedback for recommendations. The problem with recommendations for groups using collaborative filtering still is finding the best aggregation technique for group recommender systems. Aggregation techniques in group recommender systems are used to construct an individual model. There are several aggregation techniques that we discuss in this research, such as average, least misery and most happiness. Observations of the recommendation give us the conclusion that the use of least-misery techniques results in the best recommendations according to F1. Furthermore, giving more weight to group neighborhood can also improve the recommendation.

Adaptive educational hypermedia (AEH) has been an alternative tool to replace conventional learning tools. It mainly offers adaptive navigation and presentation based on a wide range of characteristics, preferences and understanding of learners. The adaptation is enabled by a user model which records users' characteristics, domain models which represent all concepts taught in a course and relationships among them, and goal and adaptation models that infer domain and user models to produce adaptation. The main adaptation offered in AEH deals with where a learner can go next. However, it does not inform learners how far they are from the goals. This paper addresses the problem of finding the shortest path to reach a learning goal. We propose a modified Dijkstra algorithm applied to a graph-based domain model. It takes account of learners' knowledge, the weight of topics in a course and the influence scores of each topic to learn the other topics to predict the possibility of success in learning a topic. Based on the success possibility scores, learning paths are recommended to students. Since students' knowledge is progressive, the success possibility scores are dynamics and they will result in a dynamic learning path adaptive to students' progress.