Accepted Papers

  • A Zoomable Interface Design for the Search Result
    Fotoon Abu-Shaqra, Ahmad Alaiad, Ismail Hmiedi,Faculty of Computer and Information Technology, J.U.S.T, Irbid, Jordan

    We can now access information everywhere, at any time, using smart phones. So a new ways of representing search results is needed to suit the screen size and limitation of those smart devices. We first present some studies of user search behavior to explore what current mobile search engines need to be more usable, then we present a new approach that exploit the small screen size in a better way and utilize more relevant results using zoomable snippets interface.

    Ahmad Alaiad1,Hassan Najadat1 Nusaiba Al-Mnayyis2,Ashwaq Khalil2 Ali Qussai Yaseen11Department of Computer Information System, Jordan University of Science and Technology, Irbid, Jordan 22 Department of Computer Science, Jordan University of Science and Technology, Irbid, Jordan

    Data Development Analysis (DEA) has been widely used in many fields. Recently, it has been adopted by healthcare sector to improve efficiency and performance of the healthcare organizations and thus reducing overall costs and increasing organizational productivity. In this paper, we demonstrate the results of applying the DEA model in Jordanian hospitals. The dataset consists of 28 hospitals and classified into 2 groups: efficient and non-efficient hospitals. We applied different association classification data mining techniques (JCBA, Weighted Classifier and J48) to generate strong rules using WEKA. We also applied The Open Source DEA (OSDEA) software and Max DEA software to manipulate the DEA model. The results showed that JCBA has the highest accuracy. However, Weighted Classifier method achieves the highest number of generated rules while the JCBA method has the minimum number of generated rules. The results have several implications for practice in healthcare sector and for decision makers.

    Souad Taleb Zouggar1,Abdelkader Adla2Department of Economics, University of Oran 2, Oran,2 Department of Computer Science, University of Oran , Oran

    We propose in this work a new function named Diversity and Accuracy for Ensemble Selection (DAES) for ensemble pruning which take into account both accuracy and diversity. A comparative study with a diversity based method and experimental results on several datasets show the effectiveness of the proposed method.

  • Generated Document Trees - Latent Dirichlet allocation
    Lawrence Master, Dakota State University

    Creating Structure or building hierarchy is a very frequently performed task. We are familiar with this act when organizing our documents in folders and subfolders on our personal computer hard drives and attempting to determine which documents are more “general” or should be in “parent” folders. When dealing with a handful of documents, the task is usually trivial. However, what about when a physician needs to diagnose a complicated disease and explore alternative treatments by attempting to sort through thousands of related article documents? In the past several years, machine learning and information retrieval techniques have been used to develop many topic models based on the extremely popular Latent Dirichlet Allocation algorithm. However, in the space of unsupervised dynamically generated document structures, this specific area of machine learning has been lacking. We propose a new method, which we call Generated Document Trees Latent Derelict Allocation. Our method does not rely on being given manually created document structures apriori and was developed to dynamically generate trees of documentsin an unsupervised fashion. We show an example of its application on the TREC clinical decision support dataset, and explore its performance.