-->
Erdogan Dogdu, Ph.D.

Research

Research Interests

Semantic Web and Linked Data

Semantic Web is made up of technologies that will shape the future of the Web (also referred to as Web 3.0) is envisioned to overcome the shortcomings of the current Web. The future Web will continue to be a Web of Documents, but at the same time it will be a data network (also called “Web of Data”). This data network (Semantic Web) will be easily used by not just people, but by machines, computers, and software, and the desired information will be found much faster and easier. In the field of semantic web, we are working on the design and development of (1) ontology-based complex software systems, (2) discovery and analysis of “linked data” (semantic knowledgebases), and (3) semantic processing of text using linked data, such as text categorization. These studies will lead to the development of question answering systems, smart search engines, and smart digital assistants.

Recent Publications

Data Science, Big Data, and Machine Learning

Data in digital form keeps growing with the advances in Internet and Web technologies. Big Data Analytics is the the process of examining large, unconventional (structured, unstructured, semi-structured and mixed) data to find hidden patterns, unknown correlations, other useful information so that the discovered knowledge can be used by organizations, individuals, and governments to make informed decisions. Big Data research area covers a number of different research directions. Our focus is both (1) on the discovery of new methods for the processing, exploration and analysis of large datasets, and (2) on the application of known methods to the existing big data problems in order to discover, learn, and predict from data. Towards this end we use and apply methods from machine learning, data mining, statistical analysis, and recently deep learning techniques. We also utilize and extend many methods in the areas of text mining and analytics, web analytics, information extraction, graph data and analytics, and semantics. Application areas include web, semantic web, Internet of Things (IoT), financial data, and large corpus of text data.

Recent Publications

Funded Research Projects

SEAS: Smart Energy Aware Systems
Increasing energy efficiency and sustainability via smart energy aware systems in building and micro-grid environments. Use case analysis, SEAS information model and ontology design, SEAS information exchange and web services platform design.

Waste Management Information System: An Expert System Using Ontologies

Rules and regulations are difficult to follow using the written documents. Therefore, domain experts and consultants are needed in many fields to help with compliance with the law and regulations. Waste management and related environmental laws are one of those areas that require great attention and detail because fines and penalties have been significantly increased in recent years all over the world. We have looked at the regulations that are overseen by the Waste Management Division of Environment and Forestry Ministry of Turkey. It consists of 16 separate documents with over 100.000 words. It is very detailed and difficult to search and follow. The regulations are very complex and therefore the Ministry publishes flow-charts to explain the rules for involved parties. There are over 700 environmental consulting companies and more than 11 thousand experts registered by the Ministry to help companies and institutions with the regulations.

We are developing a research project with funding from the Ministry of Science, Industry and Technology of Turkey to develop an expert system that will help companies, institutions, and individuals to comply with complex rules and regulations. The system will enable domain experts to define the rules for any application domain and end-users to monitor their compliance with the rules. For this purpose an ontology-based information system will be built with a dynamic graphical user interface for easy creation of rules with graphical components in a Web interface. The system will utilize reasoning capabilities of ontologies to deduce new information and additional rules, and automatically detect compliance or no-compliance situations for the users and alert them appropriately.

As a case study, we will convert all rules in 16 documents by the Waste Management Division into the system with the help of TAYTEK Waste Management Inc. (Ankara), an environmental consulting company located in Ankara with domain experts and is a partner and sponsor for the project. We will measure the effectiveness of the system with test cases in the field.

Virtual Enterprise

Large scale and multi-national companies have access to diverse customers by expending their global, widespread sale and service networks. These companies are capable to fulfill customers’ requirements rapidly so as a result they are able to increase progressively their global market share.  SMEs with restricted capacities are facing with the problem of losing their local and temporary customers. To cope with this problem collaboration among SMEs with various vertical competencies and high tech start-ups in techno-parks of universities for production of high value added final products seems vital for their growth. For this endavour, this project entitled “Development of an Operational Virtual Factory Framework for the Production of High Value-Added Products by SMEs” has been developed and it is planned to be implemented in Defense and Aviation cluster of OSTIM Organized Industrial Zone in Ankara, Turkey. The main objective of this project is to provide fast and secure access to potential customers, and organize potential SMEs rapidly to respond customers’ requirements.

In this project we are developing an ontology- and agent-based expert information system that will capture the domain knowledge as well as SME information thoroughly, and help create VEs quickly and respond to market demands.[/expand]

ATM Menu Personalization and Performance Optimization

Nowadays, ATMs, or Automatic Teller Machines, are one of the most widely used, indispensable machines since they allow people to do banking tasks without the need for a teller, a person doing those tasks. They are accessible everywhere especially where there are not bank offices and can serve non-stop, 7/24, everyday of the year.

In recent years, the number of ATMs has risen sharply, especially in Turkey, and they are programmed to do almost all the tasks a customer can do in the bank office. Therefore, these machines are not just only for money withdrawal and balance inquiry as in the old days, but they are designed to operate almost like an actual bank branch where customers can deposit money, pay bills, transfer money, do foreign exchange transactions, buy/sell stocks, options, etc.

Since the number of operations an ATM can do is increasing everyday, the user interfaces of these machines are getting more complex accordingly. In the early days, one screen was enough for all the operations, nowadays, there are several menus and submenus needed in order to accommodate all the operations that can be performed at the ATM. Due to this complex structure, customers have to navigate through menus to do their operations and this can get time consuming and frustrating for people who are using ATMs and for those waiting in line.

A similar situation can also be observed for IVR (Interactive Voice Response) systems which provide voice menus for the customers calling via phones and trying to use bank systems or other call center systems. Many banks, companies, organizations, etc. are using IVR systems for their customer support. In this case waiting time is considerably increased due to listening and going through a number of back to back menu options one by one.

Both ATMs and IVR systems have general menus for all customers, but customer needs can be different and people usually use only a small portion of menu options they need. Moreover, people spend time unnecessarily when using these systems due to pressing wrong options or not seeing the options they need immediately.

In this project we will design and develop a system that will profile user operations and accordingly design personalized user interfaces for ATMs. Later on, if needed, this system can also be used for IVR like applications.

To implement user profiling, we will use big data from past ATM operations, and accordingly profile and cluster users. The aim is to shorten the waiting and operation time by reducing the number of push button operations as well easing the accessibility of menu options for ATM users. To accomplish this, we will use machine learning and artificial intelligence techniques. We will optimize menu options according to user profiles and develop a software engine that will utilize machine learning techniques. There will be a data acquisition layer of the software where data from ATM usage logs will be provided for processing. This will be a generic layer where different data sources (IVR usage etc.) can be integrated. Due to a layered design, data gathering, data processing, and menu optimization operations can be done separately and applied to different applications (ATM, IVR, etc.). As a first application, in the context of this project, we will work on ATM menu optimization only.

With the help of this system, the customers will be able to do the operations easy and fast, therefore customer satisfaction will be increased. And, this will affect the company positively due to better customer support and positive returns. The product will be implemented with different customer operations in view and therefore be adapted in different applications in future. Therefore, the product will provide a key and strategic advantage for the company in local and global marketplace.[/expand]

Previous Funded Projects