Ozgur Yilmazel

A comparison of Relational Databases and information retrieval libraries on Turkish text retrieval

The present work covers a comparison of the text retrieval performances of relational databases and IR Systems over a TREC-like test collection for Turkish. The effects of language specific preprocessing and different query lengths for different information retrieval systems are investigated. Showed that language specific preprocessing improves retrieval performance for all systems and also Relational Databases are slower with longer queries.

Guided Navigation Using Query Log Mining through Query Expansion

We introduce a method to help people to refine their search queries based on other people’s similar queries. We compared two different methods of collaborative filtering on search logs of a popular e-commerce site to do query expansion. In item-based method, by identifying similarities between different items, and in user-based method, by identifying similarities between different users, we compute recommendations for users. The experimental results show that item-based query expansion method provides better performance than the user-based method. While the user-based method improved the system performance 20%, the item-based method provided success of 73%. Also, item-based method recommended better expansion terms than the user-based method, which is important in helping web users to easily access information needs by formulating qualified queries. This document is a comparison of these two methods, and their effects on system performance.

Document embedding approach for efficient authorship attribution

Authorship attribution has been well studied in terms of text classification with many diverse feature sets. However, finding topic independent features is hard and trained models with hand crafted features in one domain may not work in another domain. In this study we used a semi-supervised neural language model which is known as document embeddings for authorship attribution problem. This method showed significant improvements over bag-of-words representations in a well-known dataset.

Not all software engineers can become good data engineers

The amount of data that businesses collect and analyze has been rapidly increasing, which has triggered an increase in big data teams. With the growth of both the number and size of big data teams, specialized roles are starting to be defined. One such role is the data engineer, who focuses on ensuring that the data is easily available for advanced analytics. Via a case study, this paper explores the role of the data engineer and the key characteristics that enable someone to be a good data engineer. The paper also explores if good software engineers could become good data engineers. Our findings show that the knowledge and skills required to be a data engineer are significantly different from those required to be a software engineer. Hence, not surprisingly, we found that that not all software engineers could become good data engineers.

Mobile Applications at a Mega University: Anadolu University Campus App

This paper explores the use of mobile applications to aid on-campus and off-campus students at a mega university. Anadolu University — with over 1,900,000 students enrolled from over 30 countries — is the worlds second largest university by enrolment(List of largest universities by enrolment, 2014). From its early days, the Universityhas used various means to access its students. During the last decade,with the introduction of mobile technologies and smartphones that are connected everywhere, the expectations of students have changed. Students now expect to be in contact with their educational institutions without any barriers. Anadolu University released its first mobile campus app onaniOS platform in May 2012. Students adopted the App quickly and the mobile app user community requested new functionalities. Since then,the University has released three major and over 25 minor releases of the app on both iOS and Android smartphones. This paper describes the lifecycle of Anadolu Campus App and its evolution over the years. It has been widely acceptedby our students both on campus and off campus, andthe increasing number of users gives an insight into the high rate of adoption of smartphones.

Quality benchmarking relational databases and Lucene in the TREC4 adhoc task environment

The present work covers a comparison of the text retrieval qualities of open source relational databases and Lucene, which is a full text search engine library, over English documents. TREC-4 adhoc task is completed to compare both search effectiveness and search efficiency. Two relational database management systems and four different well-known English stemming algorithms have been tried. It has been found that language specific preprocessing improves retrieval quality for all systems. The results of the English text retrieval experiments by using Lucene are at par with top six results presented at TREC-4 automatic adhoc. Although open source relational databases integrated full text retrieval technology, their relevancy ranking mechanisms are not as good as Lucenes.