Search

The lecture deals with different Business Intelligence projects which were conducted in Croatia in the past 15 years.

Današnji zahtjevi s kojima se svijet susreće zahtijevaju brza i efikasna rješenja. Takva rješenja omogućavaju nerelacijske baze podataka koje su usmjerene na poslovne sustave ne bi li se rukovalo s nestrukturiranim i strukturiranim podacima i samim time povećala fleksibilnost i brzina razvoja. Problem koji se razmatra u radu je kako pospješiti komunikacijski proces između profesora i studenta/učenika u školskom sustavu koristeći nove tehnologije. Kroz rad će se demonstrirati izrada aplikacije te iznijeti prednosti i nedostaci NoSQL sustava i ključ- vrijednost baze podataka.

Process mining is a set of techniques which allow the automatic process analysis. Databases are places where all the data about process execution are stored. Data warehouses are databases of a special kind which collect and store data from different heterogeneous sources, including the data about processes. This article explores the existing literature about two related areas: process mining and data warehouses. The aim of this article is to research the methodologies used in previous researches, find out in which areas process mining combined with data warehousing is used, and what kind of tools researchers used to perform analysis.

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 prikazivanju 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“ prikazati 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 grafovskim bazama podataka i upitne jezike za interakciju s grafovskim bazama podataka, te ćemo posebnu pozornost obratiti na buduće trendove razvoja ove zanimljive i sve popularnije kategorije NoSQL baza podataka.

Databases enable users to store and retrieve data. However, uncontrolled data entry often results in having duplicate or incorrect data stored in the database, which makes it inconsistent. To prevent this, it is recommended to specify integrity constraints in the database. In this paper, the current possibilities in the field of integrity constraints, with special emphasis on graph databases as a relatively new category of NoSQL databases, are discussed. After giving an overview of the current situation regarding integrity constraints in mostly used graph database query languages, a successful implementation of UNIQUE integrity constraint in a Neo4j graph database is shown.

The issue of maintaining a database in a consistent state is becoming more challenging in situations where users are able to store all kind of data in the database without any control. Leaving this process unattended can lead to invalid and duplicate data in the database, which can directly affect the correctness of gathered data. In this paper, an overview of integrity constraints is given with the emphasis on graph databases, as well as the level of support provided by Neo4j, the most popular graph database management system (DBMS). A new integrity constraint has been implemented, which prevents users from entering data with values outside the permitted predefined set of values.

Even though relational databases have been (and still are) the most widely used database solutions for many years, there were other database solutions in use before the era of relational databases. One of those solutions were network databases with the underlying network model, whose characteristics will be presented in detail in this paper. The network model will be compared to the graph data model used by graph databases, the relatively new category of NoSQL databases with a growing share in the database market. The similarities and differences will be shown through the implementation of a simple database using network and graph data model.

Graph databases, which are presently classified as one of the most interesting database types, are being used on a daily basis. However, different systems are available and different languages are used to operate using graph databases. To summarize, a large number of languages hinders graph databases’ wider use. There also have been attempts to create universal SQL for graph databases in order to make things more standardized and intuitive to use. In this paper, we propose Gremlin By Example. This is a new language for graph databases that resembles Query By Example, which is a query language for relational databases. Because people are familiar with Query By Example, and Gremlin By Example looks almost the same, by using the similar interface people should be able to use graph databases without knowing the details of different graph query languages.

In this paper we discuss some aspects of the current state in the field of business information systems domain with an emphasis on the domain (context) of business rules, ranging from theoretical assumptions through modelling (in appropriate languages and tools) to use in business, with a focus on so called business rule paradox. It is well known that in many areas of human activity there is a gap between theory and practice and that this gap is mostly expressed either through inadequate theory or inadequate practice, where it is usually more often the case that the theory is more developed then the practice, or (in its own way, even though that may seem to be paradoxical) practice is more developed then theory. In the context of business processes and rules, their modelling and effective application to business practices, there is a paradox of the existence of highly developed theories which would enable quality practices, and a whole array of individual, very proficient practical solutions (implementations of business rules) on the one side, but no universal and generally accepted methodology for modelling bussines rules as a service for business processes and information systems to support them on the other side. The part of the problem is to be found in the inherent complexity of corresponding theories and tools which are necessary for modelling business rules and processes, but a part is to be found in inadequate (and due to systemic reasons incomplete) education of the main participants in that process, as well as a lack of communication between them. Business rules are undoubtedly one of the most important resources of any business subject. They are inherent to every business process, independent of its being conducted manually or with the support of information technology. In the first case they are dispersed in the multitude of laws, statutes, standards, business regulations, business and technical documentation and so on. In the second case, in the context of modern business practice, which is increasingly dependent on the support of information technology and is conducted in the virtual world of computers and computer nets ranging from local to global, like the World Wide Web, they are partially implemented in the processes through which business itself is pursued. Large companies and software manufacturers like Microsoft or IBM, conscious of a potentially vast market in the business rules segment, have developed or are currently developing large-scale development environments for business support, such as BizTalk Server (Microsoft) or WebSphere ILOG JRules (IBM), which, among other features, contain “modules” or even “languages” for modelling business rules. The importance of business rule modelling is additionally emphasized by the fact that a whole host of professional association (such as Business Rule Community (1), Object Management Group (2) - OMG, W3C (3), Eclipse (4), Protégé (5), IIBA, (6) …)is developing standards, specific programming languages, modelling languages (especially markup languages grouped around XML) and development environments for the requirements of the business rule and business process modelling as well as for the so-called accompanying technologies (semantic web, ontologies, Web services). This is equally true of the academic community involved in these research areas. Scientific research in the area of modelling business rules belongs within a wider context of the so-called semantic technologies. Next to modelling business rules, they encompass categories such as ontologies and semantic Web. These encompass a few narrower areas, such as developing logic formalisms (Descriptive logic, F-Logic, …) for the business rules modelling and the development of corresponding programming languages and environments (Protégé, Ergo Suite, Flora-2, DATALOG, …). A part of the research and development is done through the development of (the) corresponding standards under the auspices of the already mentioned associations such as W3C, OMG and Business Rules Group. The other part is done through the research of numerous research teams and individuals within the academic community and the research departments of large software companies, profit-oriented (IBM, Microsoft, Google (to a certain extent), Vulcan, Coherent Knowledge Systems, …) and non-profit (Eclipse Foundation, …). Nevertheless, despite all the things described above, concerning the development of theory, programming languages, tools and development environments, very few (the authors aren't familiar with one single case) cases of implementation of business rule systems (by the (Business Rule) Book) exist which fulfil the criteria established in a well-reasoned manner as irrefutably valid by Ronald Ross and his collaborators. We will proceed to elaborate why this is the case and to suggest a solution for the above mentioned paradox. In order to understand that paradox we will shortly describe the existing practice of business rule and business process modelling, some (not all) theoretical pillars necessary for their successful modelling and also present the main “players” in that “game”. Finally, we will “distribute” that set of theoretical knowledge and practical skills to these players according to their respective “jurisdictions” over business processes and business rules.

Nowadays, most enterprises still use relational databases to store their data. However, in a modern and dynamic environment the relationships connecting stored data are gaining more importance than the data itself. In this situation, graph databases have proven to be quite useful. Regardless of which database management system is being used to manage the company data, the issue of database query performance must be considered due to the increasing amount of data. Indexes are data structures, which can be used to improve database query performance. Due to their significant impact on the database performance, indexes have been developed and implemented both in relational and graph databases. In this paper, an overview of various indexing methods in the previously mentioned database categories has been made in order to detect their similarities and differences.

Modern business systems, especially those of middle and large size, generate a large amount of different data.This data is important for proper analysis, prediction of future trends or in other words for making various business decisions. The large amount of data that can be found in many different formats and in different technology can be stored on multiple locations that all refer to the same business system. Education about technology, such as various business organizations’business intelligence systems and data warehouses is important part of many higher education curriculums. However, since many business intelligence tools rely on various graphical representations of data they are not suitable for visually impaired students. In this paper a discussion about challenges that visually impaired students encounter when trying to learn about business intelligence is given along with directions for future research

The semantic social web or web 3.0 is the movement of integrating popular social web (web 2.0) applications with ideas from the semantic web. Techniques like semantic wiki systems, social tagging of weblogs and other web 2.0 applications as well as FOAF (Friend of a Friend) specifications are some of the results of this movement. Since metainformation acquired by such systems depends mostly on the social system interacting with them it is almost certain that inconsistencies will emerge. We analyze the case of a semantic wiki system (TaOPis) dealing mostly with knowledge management and show how this inconsistencies can yield problems for eventual agents searching the dynamic knowledge base. In the end we give some guidelines and possible solutions for avoiding such unwanted behavior.

The paper presentes a method for the evaluation of biometric systems based on an open ontology of selected segments of biometrics. The ontology was developed in F-logic using the FLORA-2 system. Descriptive parameters of biometric characteristics as well as organizational contingency parameters are taken into consideration and a metric for grading a biometric system's adjustment with organizational needs – the adequacy level is introduced.

The active database theory is well developed ; many different types of events are supported, static analysis mechanisms are developed, etc. One of the problems regarding active rules management is a number of rules defined ; it is not so easy (for a user and for a system) to manage a large number of rules. Although some systems allow similar rules grouping, the problem is still significant. During a project we had the same problem occurred i.e. a number of time triggers that were used was too big. This paper describes how we reduced a number of time triggers that were used in a multiagent system by introducing a meta-trigger.

In our previous paper that has been published in the field of data warehousing and business intelligence the idea of deductive data warehouses has been proposed. It has been recognized that deductive databases and data warehouses are quite similar and the main idea of the paper was to define deductive data warehouses and to show how Datalog can be used to analyze data i.e. to perform OLAP analysis on data. The idea seems to be interesting but many things need to be explored further on. In this paper we explore how deductive data warehouses can be used on two real case scenarios and we examine the benefits and drawbacks of such an approach.

In one of our previous papers, the idea of deductive data warehouses has been introduced. It was shown how to use Datalog rules to perform Online Analytical Processing (OLAP) analysis on data. In this paper, we show how to use Datalog rules to specify the data warehouse model (data mart) as well as how to add rules that produce aggregate and derived tables that are normally used to speed up the process of retrieving data. Since it is good to have aggregate and derived tables (to speed up queries), the main drawback is that they require extra storage. Consequently implicit definition of such tables may seem interesting.

Every grocery store has data about bills, bought items, etc., usually stored in a database (DB). There is often a need to analyze this data. This is not suitable to do on an operational DB, especially when data from two or more stores must be analyzed. Therefore, data from the operational DB is extract-ed, transformed, and loaded (ETL) into a data warehouse (DW). The ETL process is the most important part of build-ing a data warehouse. It is used to extract data from opera-tional data sources, transform the data (as needed) and to load the data into a data warehouse. Because the ETL pro-cess (as such) is very complex and time consuming, a proto-type of a web ETL tool was built and tested. The results have shown that it is possible to build a completely web-based ETL tool and that it is faster than manual ETL. Also, be-cause it is completely web based, multiple users can use the tool at the same time, with no installation is needed.

In recent years, NoSQL databases are becoming more and more popular. We distinguish several different types of such databases and column-oriented databases are very important in this context, for sure. The purpose of this paper is to see how column-oriented databases can be used for data warehousing purposes and what the benefits of such an approach are. HBase as a data management system is used to store the data warehouse in a column-oriented format. Furthermore, we discuss how star schema can be modelled in HBase. Moreover, we test the performances that such a solution can provide and we compare them to relational database management system Microsoft SQL Server.

There is a connection between reactive agents and active databases that has been already explored earlier ; so far it is known that both fields rely on reactivity and that reactive agents can be implemented in active databases. When talking about active databases triggers are unavoidable. Although one way to implement triggers is to use the Structured Query Language (CREATE TRIGGER statement in SQL), another (more interesting) one is based on the idea of Trigger By Example (TBE) approach. TBE is (on the other hand) based on Query By Example (QBE) that is another language used to work with databases (QBE is a graphically oriented language a bit easier to learn and use than SQL). In this paper a new approach (based on the already mentioned Query By Example and Trigger By Example approaches) to implement reactive agents in active databases is presented, the so-called Agent By Example (ABE) approach.

Faculty of organization and informatics in Croatia represents an interesting choice for vast number of students. In order to enroll students had to pass the entrance exam that consisted of several groups of questions covering different courses. Based on students’ high school knowledge (and grades) as well as entrance test results (and some additional activities) we used to determine the list of students that were eligible to enroll. Later on (during their study) those students passed some exams as well. This paper describes the solution that was built in order to analyze the success of students from different schools and regions. For that purpose a data warehouse was built that integrated data from several different sources and front-end business intelligence tool was used to build reports that were used to clarify the whole situation.

There is a connection between reactive agents and active databases that has been explored in previous papers ; in brief we know that both fields rely on reactivity and that reactive agents can be implemented in active databases. When discussing active databases, triggers are unavoidable. Although one of the ways to implement triggers is to use Structured Query Language (CREATE TRIGGER statement in SQL), another (more interesting) one is based on the idea of Trigger By Example (TBE) approach. TBE is based on Query By Example (QBE) which is another recognized language used to work with databases (QBE is a graphically oriented language and therefore a little bit easier to learn and use than SQL). In this paper a new approach to implement reactive agents in active databases is presented, the so-called Agent By Example (ABE) approach. This idea is based on the already mentioned Query By Example and Trigger By Example approaches that are extended and modified as is presented and described in the paper.

Google company often surprises us with their new services ; one such service is “Google circles”. The main concept of this service is the possibility to organize friends and (for example) co-workers in circles in order to share information more efficiently or just to stay in touch with one another. The main idea of this paper is to analyze how such circles are formed by using business intelligence capabilities ; it is strongly believed (assumed) that people in a circle have (share) similar habits. This paper analyzes how business intelligence (i.e. analytical processing) can be used to analyze sales data based on the circle membership.

Business intelligence (BI) and data warehouses are used in many different business areas, but mostly in those that earn more money and that are more profitable (insurance companies, telecommunication companies, etc.). Although BI is not a new technology, and is (in fact) quite mature, in some business areas, however, it is not used at all. In this paper we try to identify and analyze the main reasons why BI is not used in the textile industry in Croatia. For that purpose, we have built two questionnaires (one for managers and one for IT specialists in the textile industry) and the results we have obtained are quite interesting. Further on, we show that a BI system (and a belonging data warehouse) can be successfully implemented after all, even though managers and IT experts do not share the enthusiasm about such a project.

This paper demonstrates how Business Intelligence and Data Warehouses could be used for educational purposes in order to analyze entrance exams (University of Zagreb, Faculty of Organization and Informatics, Croatia). In order to enroll, students have to pass an entrance exam that consists of several groups of questions. Before the solution described in the paper was developed, we intuitively knew something about the students and the entrance exam results, but information we got by analyzing the data in the developed data warehouse changed our point of view (some results confirmed our expectations and some were surprising and require further analysis).

Data Warehouse and Business Intelligence have been used in the recent years in many areas for analyzing different phenomena and number of interesting articles can be found on this subject. In this article we are going to show how Data Warehouse has been used in educational field for analyzing university exams. Although, the intuition was helpful, exact data yielded through the process changed our point of view. Some results confirmed our expectations, some reviled limitations and some will require further analysis and adjustments to be made.

The need to understand the behavior of bidders increases with the popularization of online auctions. Every bidder is seeking the way to buy items as cheap as possible, but then again all participants in the process should be satisfied. But there are bidders who participate in auctions and wait until the very last moment to submit their bids, not allowing others to react (properly) to their bids. So sometimes one doesn’t get some items because somebody offered less than 0.5% more money for the same item. For that reason we decided to simulate online auction(s) in order to see what would happen if, after the regular (English) auction was over, agents had an opportunity to offer within another (sealed-bid) auction. This would ensure that seller agents receive more money, but it would also allow (early) bidders to react to "late submissions" in order to buy items after all.

Active databases are capable to react to certain events by performing some actions. Reactivity is, on the other hand, very important property immanent to agents. This paper explores the connection between the two mentioned fields and demonstrates how active rules can be used to implement reactive agents in multiagent systems. Further on, this paper shows how one such agent can be used in order to solve physical conflicts that may arise in multiagent systems when allocating scarce resources.