Building extraction is a component of many environmental modelling and data analysis applications. It is however data and knowledge intensive. We investigate the use of publicly available data from Google Earth and OpenStreetMap and of neural networks for this task. We evaluate different candidate algorithms for the case of building extraction on the island of Bali.

An emerging and promising approach to achieving access to water and sanitation for all leverages citizen science to collect valuable data on water quantity and quality, which can assist policymakers and water utility managers in sustainably managing water resources.

Massive Open Online Course (MOOC)s are facing several challenges, and researchers in this domain are exploring several dimensions to these problems. In this paper, we address the high dropout issue faced by MOOC systems, which is a serious challenge capable of undermining the effectiveness of these systems. There is a significant body of research addressing this issue, however they are predominantly focused on identifying a student at the verge of dropout using learner and learning analytic techniques. They do not suggest any methodology for preventing learners dropout. To do so we present an agent-based model to find the conditions for which the dropouts are decreased, indicated by more and more learners completing the course as the simulation progresses in time. The simulation parameters are tuned up to ask interesting "what-if" questions, resulting in some interesting findings. It is observed that mobility (social connectivity) and initial distribution (physical placement) of agents have a deciding role in decreasing the dropout rate. We also propose an umbrella framework for analysis of the MOOC systems which goes beyond analytical modeling and is able to handle such a complex system.

Sentiment analysis for Indonesian social media text is very important because the text content in social media is very diverse and requires an accurate method that can produce an analysis describing the state of the actual data. The main problem in sentiment analysis for Indonesian text on social media is unstructured text data and the use of non-standard languages such that sentiment analysis often produces errors. This paper focuses on sentiment analysis using a hybrid model that combines lexicon based and maximum entropy methods to classify the sentiments of Indonesian public opinion on government. The method consists of extracting datasets, preprocessing, lexicon-based classification, machine learning training, machine learning classification, and result interpretation. The results of the study produce 91 classifications of neutral sentiment, 51 document negative sentiments, 39 document positive sentiments and 152 document of mix sentiments. Based on the evaluation results, the hybrid sentiment model for Indonesian Language sentiment analysis on social media produced a pretty good accuracy score of 84.31% compared to previous studies. The implication of this study is to produce a sentiment analysis system with a hybrid method for Indonesian text on social media.

Sentence boundary detection is essential for natural language processing (NLP). Sentence boundary detection in the Indonesian language has lots of problems, which includes punctuation, abbreviation, and character in the bracket. The disambiguation should be detected as sentence boundary. Thus the sentence boundary system can divide the sentences accurately. This study presents the development of a training dataset for the existing model to optimize supervised sentence boundary detection for the Indonesian language. Indonesian Translation of the Quran (ITQ) data set was used in this study by using the supervised method. The following is the process briefly: create the training data, apply sentence detection to separate sentences on ITQ, and calculate precision, recall, and F-measure. The result is quite promising, it gives as follows: Precision of 91.7%, Recall 81.6%, and F-Measure 86.4 %, respectively.

Recently, Olariu et al. [3], [7], [18], [19], [20] proposed to refer to a dynamic group of vehicles whose excess computing, sensing, communication, and storage resources can be coordinated and dynamically allocated to authorized users, as a vehicular cloud. One of the characteristics that distinguishes vehicular clouds from conventional clouds is the dynamically changing amount of available resources that, in some cases, may fluctuate rather abruptly. In this work, we envision a vehicular cloud involving cars in the long-term parking lot of a typical international airport. The patrons of such a parking lot are typically on travel for several days, providing a pool of cars that can serve as the basis for a datacenter at the airport. We anticipate a park and plug scenario where the cars that participate in the vehicular cloud are plugged into a standard power outlet and are provided Ethernet connection to a central server at the airport. In order to be able to schedule resources and to assign computational tasks to the various cars in the vehicular cloud, a fundamental prerequisite is to have an accurate picture of the number of vehicles that are expected to be present in the parking lot as a function of time. What makes the problem difficult is the time-varying nature of the arrival and departure rates. In this work, we concern ourselves with predicting the parking occupancy given time-varying arrival and departure rates. Our main contribution is to provide closed forms for the probability distribution of the parking lot occupancy as a function of time, for the expected number of cars in the parking lot and its variance, and for the limiting behavior of these parameters as time increases. In addition to analytical results, we have obtained a series of empirical results that confirm the accuracy of our analytical predictions.

This report summarizes the International Workshop on Situation, Activity and Goal Awareness held at the 13th ACM International Conference on Ubiquitous Computing, on September 18 in Beijing, China.

In ubiquitous environments a vast amount of mobile human and software entities, each with limited resources and knowledge, needs to interact with each other to achieve common and/or individual goals within a specific context. Due to their autonomy, proactiveness, mobility, social capability, and the successful implementation of agent mediated applications and services over the Web, different scenarios have been proposed in literature for the use of agents in ubiquitous environments for a wide range of applications such as user interfaces, mobile computing, information retrieval and filtering, smart messaging, telecommunication and m-commerce. In this paper, we address the problem of energy efficiency in ubiquitous environments from customer satisfaction, customer awareness and energy savings points of view. We propose an agent-based information system in which agents act on behalf of the users, are situated within the appliances, sensors and actuators, or represent different stakeholders in the energy market. Different scenarios have been proposed as the basis for the proposed information system.

The Internet, respectively the applications and services known as the Web, is undergoing significant changes. Mobile communication has enabled us with an omnipresent access to the Web, technologies such as grid computing and Web services allow us to use the Web far beyond its initial intention of document and information sharing, and intelligent agents and semantic systems aim to automate the tasks and better support us in interacting with the Web. Considering all those advances, one might ask the question how users can keep pace with all those developments and more specifically, how technology itself can help us in keeping pace. In this paper we investigate context aware adaptation and personalization as potential solutions.

Context is considered by research community in pervasive computing, and particularly in context awareness, to be subject to the available environmental sensors to retrieve the suitable information to the user. For example, the users context is based on location information or even based on other deterministic information, such as time (morning, evening, 11:00 am, etc.). The challenge is to also consider the user as a parameter influencing the context in addition to environmental information collected by sensors. This implies that human factors should be highly considered in order to formalize the ldquouser profilerdquo as an input to context awareness (context reasoning) and also as an output to the context adaptation. This vision, which we have adopted in this paper, implied to investigate both, the user actions acting in his or her living environment (Human behavior, preferences, etc.), and the context reasoning sensitive to human factors.