Nowadays, the implementation of deep learning in various fields of data mining and big data analytics has been widely used in many applications. This happens because of the ability of deep learning to be able to have excellent performance in terms of predicting cases of classification, features engineering, and clustering. Images data or texts that have large dimensions can also be processed iteratively by deep learning. In the medical field, deep learning is widely used as computer-aided detection (CAD) to provide a decision support system for radiologists or practitioners. In this research, we try to do a deep performance evaluation of several deep learning network models for the Chest X-ray disease decision support system (DSS). These X-ray images are enormous, it consists of 110,120 images, and the size is about 44 GB. Our primary interest is to get a detailed performance profile for every deep learning network model. It consists of a few evaluation aspects, accuracy performance by using areas under the receiver operating characteristic (AUROC), evaluation of training and testing time, investigation of memory usage, observation of central processing unit (CPU) usage, graphics processor power consumption, and provide some improvement solutions. We also offered a few solutions and suggestions to help the doctor or practitioner to select the most effective deep learning network model.

One of the stage that performed during a maternal and fetal health monitoring is the calculation of fetal heart rate and uterine contraction using Cardiotocography (CTG). The aim of fetal health using CTG is to avoid morbidity and mortality in fetus at risk of hypoxia. This paper propose a hypoxia detection by using classification. In this study, we improve deep learning method in order to increase its capability in detecting hypoxia. The improvement is conducted by several strategies i.e. input representation, data-scaling, data up-sampling, and adjusting classifier layers. The improvement is conducted by several strategies i.e. input representation, data-scaling, data up-sampling, and increasing classifier layers. For whole dataset that achieved by proposed method is 81%. The improved DenseNet achieved 50%, 43%, 46% precision, recall and f1-score respectively for hypoxia class that is not achieved by standard DenseNet.

The quality of higher education in Indonesia does not distribute equally, as observed from various sources of university ranking results. This calls for a solution to narrow the quality gap that leverages digital approaches. In this paper, we propose Open Tridharma, a framework for digital openness in higher education. The framework augments Tridharma, the three pillars of Indonesian higher education (i.e., education, research, and community service), by laying the digital openness foundation. We also provide qualitative analysis and discuss the benefits and challenges of Open Tridharma.

We present OD2WD, a semi-automatic pattern-based framework for populating the Wikidata knowledge graph with (originally tabular) data from Open Data portals. The motivation is twofold. One, our framework can further enrich Wikidata as a central RDF-oriented knowledge graph with a large amount of data coming from public or government sources. Two, our framework can help the discovery of data from such Open Data portals and its integration with Linked Data without forcing the Open Data portals to provide Linked Data infrastructure by themselves, as the latter may not always be feasible due to various technical, budgetary, or policy reasons. Throughout the transformation process, we identify several reengineering and alignment patterns. We implement the framework as an online service and API accessible at http://od2wd.id, currently tailored for three Indonesian Open Data portals: Satu Data Indonesia, Jakarta Open Data, and Bandung Open Data.

Interest in Semantic Web technologies, including knowledge graphs and ontologies, is increasing rapidly in industry and academics. In order to support ontology engineers and domain experts, it is necessary to provide them with robust tools that facilitate the ontology engineering process. Often, the schema diagram of an ontology is the most important tool for quickly conveying the overall purpose of an ontology. In this paper, we present a method for programmatically generating a schema diagram from an OWL file. We evaluate its ability to generate schema diagrams similar to manually drawn schema diagrams and show that it outperforms VOWL and OWLGrEd. In addition, we provide a prototype implementation of this tool.

Cultural heritage is the root of national identities, and contributes to tourism, economics, industry, and business. Digital preservation of cultural heritage is therefore crucial, particularly in a form that is easily processable by machines. Available cultural heritage information on Web sources (e.g., Wikipedia) is presented in multiple formats, such as free-form text, lists, and tables. Such formats, however, lack structures and links to other information sources. The provision of cultural heritage information as a knowledge base, that is both structured and linked, would pave new ways for consuming such information. In this paper, we propose an approach to extract entities of cultural heritage from diverse formats (i.e., text, lists, and tables), and to construct a knowledge base of cultural heritage entities, called BudayaKB, using RDF data model that provides an integrated, format-independent view. Our extraction approach follows on the observation that cultural heritage entities are often written down either with common noun descriptors (e.g., Jiwa temple) or hypernym-hyponym sentence patterns (e.g., ... Acehnese traditional weapons such as Rencong...). We evaluate our approach to Indonesian Wikipedia, and achieve a precision of 84% for extracting Indonesian cultural heritage entities. The extracted entities are then imported and linked to the Wikidata KB, allowing greater interoperability of cultural heritage information. BudayaKB is openly available at https://budayakb.cs.ui.ac.id/.

The study of infant cry recognition aims to identify what an infant needs through her cry. Different crying sound can give a clue to caregivers about how to response to the infant's needs. Appropriate responses on infant cry may influence emotional, behavioral, and relational development of infant while growing up. From a pattern recognition perspective, recognizing particular needs or emotions from an infant cry is much more difficult than recognizing emotions from an adult's speech because infant cry usually does not contain verbal information. In this paper, we study the problem of classifying five different types emotion or needs expressed by infant cry, namely hunger, sleepiness, discomfort, stomachache, and indications that the infant wants to burp. We propose a novel approach using a combination of Convolutional Neural Networks (CNN) and Recurrent Neural Networks (RNN) that acts as feature extraction and classifier method at once. Particularly, CNN learns salient features from raw spectrogram information and RNN learns temporal information of CNN obtained features. We also apply 5- folds cross-validation on 200 training data set and 50 validation data set. The model with the best weight is tested on 65 test set. Evaluation in Dunstan Baby Language dataset shows that our CNN-RNN model outperforms the previous method by average classification accuracy up to 94.97%. The encouraging result demonstrates that the application of CNN-RNN and 5-folds cross-validation offers accurate and robust result.