|
|
The IEEE reserves the right to exclude a paper from distribution after the conference (including its removal from IEEE Explore) if the paper is not presented at the conference.
| Papers |
I. Kovačević, I. Mekterović (FER, Zagreb, Croatia)
Novel BI Data Architectures 
Most modern companies utilize business intelligence (BI) engines and tools to acquire new insights into their customers and operations and consequently make better business decisions. The traditional approach to BI consists of a data warehouse containing historical data records, an OLAP engine performing data analysis, and various reporting and visualization tools used to present the results. With the advent of big data and cloud technologies, the BI approaches have started to diverge to accommodate various usage scenarios. Examples for such scenarios include small-scale analytics, which can be performed offline on personal computers, and large-scale, often near real-time, big data analytics. Additionally, several vendors have started offering cloud-based and mobile BI tools. This paper gives an overview of the novel BI approaches and associated BI solutions, and presents a feature comparison of those solutions.
|
V. Fomichov, Moscow, Russian Federation), A. Razorenov (National Research University Higher School of Economics (HSE), Moscow, Russian Federation)
Theory of K-representations as a Tool for Designing File Managers with a Natural Language Interface
During last decade, semantic parsing of the instructions in natural language (NL) has become a significant branch of the studies aimed at creating semantics-oriented NL processing systems. A topical problem of the kind is designing file managers with a NL-interface. The principal attention in our previous papers was paid to creating new, more effective theoretical foundations of designing semantic parsers of NL-instructions. In particular, we suggested in a very concise form an original and broadly applicable algorithm of semantic parsing. This paper, firstly, illustrates a correspondence between input instructions and their semantic representations. Secondly, the main attention is given to describing mathematical foundations of executing NL-instructions by computer systems. The methodological basis for these results is the theory of K-representations (knowledge representations), Its basic formal model introduces a system consisting of ten partial operations on conceptual structures. There are solid grounds for conjecturing that, combining these operations in arbitrary order, it is possible and convenient to build step by step a semantic representation of arbitrarily complex sentence or discourse in NL. The stated theoretical results have become the basis for designing a file manager with a NL-interface NLC-2 (Natural Language Commander - Version Two).
|
K. Černjeka, D. Jakšić (Department of Informatics, University of Rijeka, Rijeka, Croatia), V. Jovanovic (Georgia Southern University, Statesboro, United States)
NoSQL Document Store Translation to Data Vault Based EDW
With evolution of technology and Web 2.0 tools NoSQL stores appeared as a common solution to data storage and management demands in modern databases and applications. Relational databases were not designed to cope with agility challenges, the scale, commodity storage and processing power demanded by modern applications. Many industries today are choosing NoSQL database technology over relational databases (or at least combining them) in order to gain needed flexibility and scalability. Due to the limitations of relational databases, most researchers are oriented toward relational database transformations and data migration to different types of NoSQL stores. Our research goes in the opposite direction – we aim to develop a metamodel for translating a NoSQL document store (MongoDB) into a data vault based enterprise data warehouse. The reason we do so is to integrate different data sources into a data vault central repository and develop a new data warehouse system catalog that will track changes in both relational and NoSQL schemas. The integration of relational database and NoSQL store would help extract wider knowledge through BI tools, enable data traceability, trend discovery and will accommodate auditing process. The main contributions of this paper are translation rules that accommodate the translation between NoSQL MongoDB document store and data vault based enterprise data warehouse.
|
M. Pejić Bach, N. Vlahović, J. Pivar (University of Zagreb Faculty of Economics & Business, Zagreb, Croatia)
Self-Organizing Maps for Fraud Profiling in Leasing
Fraud is intended and planned activity aimed at achieving material or immaterial gains against interests of an organization or person. It often occurs in financial industries, such as banking, insurance, and leasing. The goal of this paper is to present a novel approach to profiling fraudulent behavior in leasing companies, using self-organizing maps. Dataset of one leasing company that consists of both fraudulent and non-fraudulent transactions has been analyzed. Cluster analysis has been applied using the self-organizing maps algorithm, with the support of Viscovery SOMine software. Five clusters were identified, that have a different structure according to the industry of the client, previous experience with the client, type of leasing object, and status of leasing object (new or used). Clusters were compared using chi-square test according to the proportion of fraudulent and non-fraudulent transactions, resulting in profiles of clients and leasing objects that are more prone to fraudulent behavior.
|
M. Ptiček, B. Vrdoljak (Faculty of Electrical Engineering and Computing, Zagreb, Croatia)
Semantic Web Technologies and Big Data Warehousing
Research in data science field has pointed out analytical potentials contained in big data on numerous occasions. The new paradigms in data storage and processing emerged with the goal of handling big data, but also pushing traditional, already present systems out of focus and creating a gap between the old and the new. That was also the case with traditional data warehouses and emergence of NoSQL data stores, whose integration has shown to be quite problematic due to NoSQL and big data features. Latest research has been trying to overcome this gap in various ways, mostly by system integration on various levels or introducing completely different new ones, but due to the data warehouse's particular nature and strict process of data modelling and acquisition, this problem must be approached on a conceptual level, capturing the essence of the domain which is to be analysed. The lack of schema in NoSQL databases makes them less comprehensible for integration and analysis, which motivated the idea of employing semantics to enrich NoSQL database contents, making them more suitable for integration. This paper analyses application of semantics in bringing together data warehouses, NoSQL databases, and big data.
|
A. Yadegari Ghahderijani, M. Razzazi (Amirkabir University of Technology, Tehran, Iran)
Author Clustering on Persian Text
Several authorship analysis tasks require the decomposition of multi-authored text into its authorial components. In this paper, we focus on unsupervised authorship analysis and provide an evaluation framework and a baseline to compare different approaches. In this work, given a collection of short documents, we approach the author clustering task to determine which documents are written by the same author. The number of clusters is determined through the computation of silhouettes for some approaches. Several approaches are compared but Affinity Propagation has the best result with 0.51 average BCubed F-score. Several features including POS tags frequency and function words are extracted from data. Text data are gathered from 6 different Persian newspaper authors.
|
S. Peker (Atilim University, Ankara, Turkey), A. Kocyigit, P. Eren (Middle East Technical University, Ankara, Turkey)
An Empirical Comparison of Customer Behavior Modeling Approaches for Shopping List Prediction
Shopping list prediction is a crucial task for companies as it can enable to provide a specific customer a personalized list of products and improve customer satisfaction and loyalty as well. To predict the customer behaviors in the next transaction, many studies in the literature have employed customer behavior modeling approaches which are individual-level and segment-based. However, pervious efforts to predict customers’ shopping lists have mainly employed the individual-level approach, which constructs a prediction model for each customer by only considering the transactional data of the individual customer. Further, no prior research compared the performance of these approaches to determine which of them is better in the shopping list prediction. This paper introduces the segment-based approach relying on the data of a customer group into the shopping list prediction and then presents an empirical comparison of the individual-level and the segment-based approaches in this problem. For this purpose, well-known machine learning classifiers and customers’ purchase history are employed, and the comparison is performed on a real-life dataset by conducting a series of experiments. The results suggest that there are no clear winners in this comparison and the performance of predictive modeling approaches depends on the machine learning algorithm implemented. The study can help researchers and practitioners to understand different aspects of using customer behavior modeling approaches in the shopping list prediction.
|
A. Murtić, M. Maljić, S. Gruičić, D. Pintar, M. Vranić (Sveučilište u Zagrebu Fakultet elektrotehnike i računarstva, Zagreb, Croatia)
SNA-Based Artificial Call Detail Records Generator
Research involving Big Data often has to deal with the problem of data availability. Real-life data involving people and their activities is usually tied with various issues of privacy, security and secrecy, which results in difficult barriers which need to be overcome before the research can even start. In this paper we suggest an approach which can reliably provide researchers with an arbitrary amount of synthetic Call Detail Records (CDR) data which would exhibit a high level of similarity with a corresponding real-life dataset. We base our approach on a simulator whose functionality is derived on results of an exploratory analysis performed on a real-life dataset which represents a social network of users with records of their activities. In this paper we concentrate on generating CDR data used in telecommunications industry, although the approach is applicable in the other domains too.
|
L. Brkić (Faculty of Electrical Engineering and Computing, Department of Applied Computing, University of Zagr, Zagreb, Croatia), K. Tomičić-Pupek (Fakultet organizacije i informatike, Sveučilište u Zagrebu, Varaždin, Croatia), D. Pavlić (Procesna inteligencija, Split, Croatia)
Consulting Experience Regarding Business Process Management Software in Croatia – A Case Study
Within the project of the Croatian Science Foundation "Process and Business Intelligence for Business Excellence - PROSPER" a research on the application of business process management systems was conducted. The research was based on case studies and gathering insights on consulting and customer experience by using structured questionnaires. The goal of research activities within this broader project part was to systematize, analyze and evaluate the functionalities of business process management software and systems, in order to define guidelines for their selection, and further on to develop a guide that would be used by organizations in selecting business process management software. In this paper we present one of case studies dealing with BPMS from our research as well as key experience factors recognized in Croatia that influence later BPMS selection process.
|
T. Mijač, M. Jadrić, M. Ćukušić (Ekonomski fakultet. Sveučilište u Splitu, Split, Croatia)
The Potential and Issues in Data-Driven Development of Web Personas
The concept of personas that represent the key audience segments of a service, product, site, or a brand has been widely used in marketing. As of lately, personas are often used in digital services design and development as well. However, there are some frequent errors and issues that designers and developers come across in the process of developing realistic personas. Namely, personas may be developed based on irrelevant data such as statistically insignificant dataset, or have been developed based on assumptions, inadequate analyses or, most often, estimated intuitively. Another issue is that once created, personas cannot be easily updated. There is great potential in big data technologies to tackle these issues since it enables analyses of extremely large amounts of data to gain insights into real user behavior patterns that lead to better business decisions. This paper explores the possibilities for developing data-driven web personas based on real user-data thus saving time in comparison with current collection methods and also providing easier updates. Data mining process using a dataset from a e-business site is performed. The research results highlight the potential of developing data-driven web personas based entirely on interactions between real users and web service.
|
K. Rabuzin, M. Šestak (Fakultet organizacije i informatike, Varaždin, Croatia)
Grafovske baze podataka – pregled istraživanja i budućih trendova
Koncept NoSQL baza podataka obuhvaća niz tehnologija za alternativni (nerelacijski) pristup upravljanju podacima. Kao jedna od kategorija NoSQL baza podataka, grafovske baze podataka specifične su po reprezentaciji podataka u obliku čvorova i veza u grafu nad kojima se prilikom izvršavanja upita izvršavaju različiti algoritmi iz teorije grafova. Društvene mreže najčešći su scenarij korištenja grafovskih baza podataka, budući da je osobe i veze između njih „najprirodnije“ reprezentirati kao čvorove i veze u grafovskim bazama podataka. U ovom ćemo radu napraviti pregled dosadašnjih istraživanja vezanih uz grafovske baze podataka općenito, sustave za upravljanje graf bazama podataka i upitne jezike za interakciju s graf bazama podataka, te ćemo posebnu pozornost obratiti na buduće trendove razvoja ove zanimljive i sve popularnije kategorije NoSQL baza podataka.
|
| Break |
| Papers |
D. Oreščanin, T. Hlupić, I. Sorić (Poslovna inteligencija d.o.o., Zagreb, Croatia)
Predictive Models for Digital Broadcasting Recommendation Engine
Communication operators are offering to their consumers many services that enable them to consume video content, using different fixed and mobile technologies through different devices, either in their homes or mobile. For operators it is of vital importance to analyze behavior of the consumers in real-time and over longer period of time, to be able not only to maximize revenue and minimize costs, but also to serve consumers better and to reach highest possible level of customer experience. For that reason, operator are using sophisticated recommendation engines to propose to their consumers’ content that may be interesting for them. This paper will explain in detail the implementation of such engine in modern content analytics system.
|
S. Peker (Atilim University, Ankara, Turkey), A. Kocyigit, P. Eren (Middle East Technical University, Ankara, Turkey)
A Methodology for Product Segmentation Using Sale Transactions
This paper presents a novel methodology for product segmentation using cus-tomers’ transactions on products. The proposed methodology utilizes the features of FMC model and clustering algorithms to group products into segments. The applicability of the proposed approach has been demonstrated on a supermarket chain. The results show that the proposed methodology provides an efficient tool to identify different product segments and to gain valuable insights about these distinct groups. We believe that obtained product segments by this methodology can help managers in the inventory management and developing marketing strate-gies.
|
V. Brkljač, L. Brkić (University of Zagreb, Faculty of Electrical Engineering and Computing, Zagreb, Croatia)
Navigating Pre-Recorded Off-Road Routes Using a Mobile Device
Most navigation systems, during calculation of user’s route, use a directed graph whose edges represent streets and roads. Using shortest path algorithms they find an adequate route. But as most bicycle paths, macadam roads and forgotten old paths are not digitally mapped, these algorithms become useless. In this paper we present a solution for tracking user’s route while driving through the unmarked area as well as navigating the user through one of the recorded routes from the repository of a stored routes. During tracking, user can mark the route by adding navigational cues. Upon completion of the tracking, previously marked route can be modified and shared with others by uploading and publishing in a shared repository. To successfully navigate user through the selected route, we developed a position determination algorithm and user-friendly interface for mobile devices. Navigation part of the solution is responsible for visual and voice instructions about the upcoming navigational cues at the right time.
The software is intended primarily to cyclists but can be used, for example, by hikers and mountaineers for safe exploration of nature.
|
T. Pełech-Pilichowski (AGH University of Science and Technology, Krakow, Poland)
On Adaptive Prediction of Nonstationary and Inconsistent Large Time Series Data
The use of time series prediction results in benefits for an organization. Forecasting efficiency relies on applied prediction formula and quality of data received from technical devices and manually inputted. They are often of low quality, with inconsistencies. However, high data quality is crucial for efficient forecasting/prediction purposes (also event detection from time series and pattern recognition), in particular during large data sets processing (often heterogeneous, including data obtained from IoT devices). Such processing should cover inconsistency analysis, interpolation of missing/lacking data, as well as the use of data pre-transformations.
The paper presents problems of inconsistent, nonstationary data prediction on the example of stock level daily forecasting. Selected methods of time series interpolation are outlined. Results of implementation of algorithms for short-term time series prediction are illustrated and discussed. Prediction quality measured based on errors values calculated both in total and in a moving window is discussed. A concept of an adaptive algorithm based on a change in the prognostic formula depending on short-term characteristics of time series is outlined.
|
V. Bosilj Vukšić (Faculty of Economics and Business, Department of Informatics University of Zagreb, Zagreb, Croatia), K. Tomičić-Pupek (Faculty of Organization and Informatics, Department of Information Systems Development, University o, Varaždin, Croatia), M. Pejić Bach (Faculty of Economics and Business, Department of Informatics University of Zagreb, Zagreb, Croatia)
Simulation Based Business Processes Analysis in Higher Education Institutions
Business process management (BPM) based on the usage of simulation modelling allows organizations to sustain and improve their operational performance. Discrete event simulation (DES) can help identify resource’s bottlenecks, such as staffing, estimate process execution or delay times, test improvement suggestions in the process design phase, and overall assist in understanding how processes might impact performance in various scenarios. While many research papers report on the industry applications of BPM and DES, we focus to higher educational institutions (HEIS). Goals of the paper are: (1) to review the research of BPM and DES in HEIs, and (2) to discuss the impact of BPM and DES to process performance in one Croatian HEI. The case study methodology has been used in order to shed some light on the issue of applying discrete event simulation in BPM in the HEIs, especially for the purpose of collaboration improvement and impact assessment.
|
T. Gregurović , R. Penco (Ericsson Nikola Tesla / R&D, Zagreb, Croatia)
Sales Configuration Creation for Complex Telecommunication Solutions
The increasing complexity of telecommunication configuration solutions has driven a rising interest in effective object oriented modeling of solutions with linear complexity scaling. In order to accomplish this, the decision was to implement and develop new configuration engine for the official Ericsson Product Catalog used by sales and marketing personnel of Ericsson. Telecommunication sales requirements are to achieve simple and fast orders for solutions based on different complexity levels. This paper gives an overview of the whole advanced system for assembly and sales process developed to support all specific requirements.
|
D. Pavlić, M. Jadrić, M. Ćukušić (Ekonomski fakultet u Splitu, Split, Croatia)
Discrete Simulation Modeling of Intelligent Passenger Boarding
The aim of this paper is to explore the application of intelligent passenger boarding systems for airports. The research was based on discrete simulation performed at Split Airport to investigate smart systems implementation implications for process and organizational unit optimization, as well as customer experience improvement. Discrete simulation model included passenger boarding process in 2 variants: 1. Validated As-is process simulation; 2. To-be process simulation with intelligent passenger boarding technology. Significant degree of optimization was achieved when intelligent passenger boarding simulation experiment was conducted, which included self-boarding solutions. This research demonstrated the potential of discrete simulation for exploring benefits of smart technology integration in transport (airport operations) services.
|
S. Tovernić, V. Banović, Z. Hrastić, K. Plantić, A. Šandić, M. Baranović (FER, Zagreb, Croatia)
Solution for Detecting Sensitive Data Inside a Data Lake
This paper is the result of a project realized by a team of current master’s degree students. The team created an algorithm for recognition of sensitive data, primarily name, surname and OIB (Croatian personal identification number). Same algorithm iterates across given unstructured texts and appoints tags for documents considering the existence of specific sensitive data. This process offers a way for companies to narrow down the search for personal information if a client demands removal of his data. Similar algorithm was implemented for working with server logs as well, which are represented as data streams and analysed in real time. To provide insight on the quantity of sensitive information and how it is distributed across different types of documents the team created a dashboard that shows statistical data accumulated by developed algorithms. The solution is stored on Cloudera, Apache Hadoop-based open source platform designed for data management and analytics, which is deployed on Microsoft Azure cloud infrastructure.
|
R. Rujevčić, R. Penco (FER, Zagreb, Croatia)
General Purpose Modeling Tool
The purpose of this document is to describe the functionality of the General Purpose Modeling Tool application with the goal of presenting it’s four key components, which serve different purposes and have their own unique implementations. The focus will be on showing the solution approach, regarding the challenges of making various aspects of the tool a reality, without going into details concerning the implementation itself.
|
|
Basic information:
Chairs:
Boris Vrdoljak (Croatia), Mirta Baranović (Croatia), Matteo Golfarelli (Italy), Mihaela Vranić (Croatia)
Program Committee:
Marko Banek (Croatia), Mirta Baranović (Croatia), Bartosz Bebel (Poland), Ladjel Bellatreche (France), Ljiljana Brkić (Croatia), Alfredo Cuzzocrea (Italy), Todd Eavis (Canada), Enrico Gallinucci (Italy), Dragan Gamberger (Croatia), Paolo Garza (Italy), Matteo Golfarelli (Italy), Igor Mekterović (Croatia), Damir Pintar (Croatia), Rok Rupnik (Slovenia), Zoran Skočir (Croatia), A Min Tjoa (Austria), Mladen Varga (Croatia), Mihaela Vranić (Croatia), Boris Vrdoljak (Croatia)
Registration / Fees:
|
REGISTRATION / FEES
|
Price in EUR
|
|
Up to 7 May 2018
|
After 7 May 2018
|
| Members of MIPRO and IEEE |
180
|
200
|
| Students (undergraduate and graduate), primary and secondary school teachers |
100
|
110
|
| Others |
200
|
220
|
The discount doesn't apply to PhD students.
Contact:
Boris Vrdoljak
University of Zagreb
Faculty of Electrical Engineering and Computing
Unska 3
HR-10000 Zagreb, Croatia
Phone: +385 1 6129 532
Fax: +385 1 6129 915
E-mail: boris.vrdoljak@fer.hr
The best papers will get a special award.
Accepted papers will be published in the ISBN registered conference proceedings. Papers written in English and presented at the Conference will be submitted for posting to IEEE Xplore.
Location:
 Opatija, with its 170 years long tourist tradition, is the leading seaside resort of the Eastern Adriatic and one of the most famous tourist destinations on the Mediterranean. With its aristocratic architecture and style Opatija has been attracting renowned artists, politicians, kings, scientists, sportsmen as well as business people, bankers, managers for more than 170 years.
The tourist offering of Opatija includes a vast number of hotels, excellent restaurants, entertainment venues, art festivals, superb modern and classical music concerts, beaches and swimming pools and is able to provide the perfect response to all demands.
Opatija, the Queen of the Adriatic, is also one of the most prominent congress cities on the Mediterranean, particularly important for its international ICT conventions MIPRO that have been held in Opatija since 1979 gathering more than a thousand participants from more than forty countries. These conventions promote Opatija as the most desirable technological, business, educational and scientific center in Southeast Europe and the European Union in general.
For more details please look at www.opatija.hr/ and www.visitopatija.com.
|
|
|
|
| Currently there are no news |
|
|
|
|
|