Browsing by Author "Tasar, Ceren Ocal"

Now showing 1 - 5 of 5

Citation - WoS: 32
Citation - Scopus: 30
Data analytics for quality management in Industry 4.0 from a MSME perspective
(Springer, 2025) Gorkem Sariyer; Sachin Kumar Kumar Mangla; Yigit Kazancoglu; Ceren Ocal Tasar; Sunil Luthra; Sariyer, Gorkem; Tasar, Ceren Ocal; Luthra, Sunil; Mangla, Sachin Kumar; Kazancoglu, Yigit; Ocal Tasar, Ceren
Advances in smart technologies (Industry 4.0) assist managers of Micro Small and Medium Enterprises (MSME) to control quality in manufacturing using sophisticated data-driven techniques. This study presents a 3-stage model that classifies products depending on defects (defects or non-defects) and defect type according to their levels. This article seeks to detect potential errors to ensure superior quality through machine learning and data mining. The proposed model is tested in a medium enterprise—a kitchenware company in Turkey. Using the main features of data set product customer country production line production volume sample quantity and defect code a Multilayer Perceptron algorithm for product quality level classification was developed with 96% accuracy. Once a defect is detected an estimation is made of how many re-works are required. Thus considering the attributes of product production line production volume sample quantity and product quality level a Multilayer Perceptron algorithm for re-work quantity prediction model was developed with 98% performance. From the findings re-work quantity has the highest relation with product quality level where re-work quantities were higher for major defects compared to minor/moderate defects. Finally this work explores the root causes of defects considering production line and product quality level through association rule mining. The top mined rule achieves a confidence level of 80% where assembly and material were identified as main root causes. © 2025 Elsevier B.V. All rights reserved.
Development of semantic web application architecture for natural language based querying
(CEUR-WS ceurws@sunsite.informatik.rwth-aachen.de, 2018) Ceren Ocal Tasar; Murat Komesli; Murat Osman Unalir; Tasar, Ceren Ocal; Komesli, Murat; Unalir, Murat Osman; B. Tekinerdogan , A.H. Dogru
With the developing technology and increased demand on practical usage of knowledge users require to interact with systems by using natural language. Researchers discover the necessity of natural language queries should be converted to machine understandable format to meet the requirement of users. The necessity of the conversion triggered the researches on studies on question answering systems that is one of the systems mostly interact with natural language queries. For question answering systems that utilizes structural knowledge sources understanding user intention accurately consists of the processes of natural language processing representing the user intention and forming structured query to generate the answer. A technique and an architecture to convert natural language query to ontology query language by utilizing from both linguistic and semantic technologies are proposed in this study. Automatic conversion of an unstructured natural language query to a structured ontology query language SPARQL and process of producing answer over linked data are explained in the proposed technique and architecture. A different approach other than the related studies is followed in terms of defining a “Query Semantization” layer in the proposed architecture and developing a natural language aware ontology for the technique proposed. Contribution to the literature is expected by combining natural language processing technique and semantic web technologies with a different approach. © 2019 Elsevier B.V. All rights reserved.
Citation - WoS: 14
Citation - Scopus: 17
Predicting cost of defects for segmented products and customers using ensemble learning
(Elsevier Ltd, 2022) Gorkem Sariyer; Sachin Kumar Kumar Mangla; Yigit Kazancoglu; Lei Xu; Ceren Ocal Tasar; Tasar, Ceren Ocal; Kumar Mangla, Sachin; Kazancoglu, Yigit; Sariyer, Gorkem; Xu, Lei; Mangla, Sachin Kumar; Ocal Tasar, Ceren
Due to technological advances Big Data Analytics (BDA) has become increasingly important over the last few years. This has led companies to evolve BDA capabilities (BDAC) to manage operations and make better decisions. In this study we propose a model Clustering Based Classifier Ensemble Method for Cost of Defect Prediction (CBCEM-CoD) incorporating clustering classification prediction and learning techniques of BDA for quality management in the manufacturing industry. CBCEM-CoD (1) is fact-driven as it is based on a fundamental problem of the manufacturing industry (2) integrates different BDA techniques in a specific way when an output of one technique is used as an input of another and (3) extracts insights from real-world big data and directly offers many implications for practice. In the first stage of the CBCEM-CoD k-means and agglomerative clustering techniques are used comparatively for segmenting customers and products. CoD values of each product and customer segment are predicted using ensemble learning techniques in the second stage. The model is tested using a case data set from the kitchenware industry. As a result 53 and 720 different types of customers and products in the train data set are segmented in optimal numbers of 4 and 20 clusters. Around 89% accuracy is obtained for CoD predictions in the test data set. These results have substantial business value since they inform managers how to prioritize their focus on specific products and customer types to reduce the cost of a defect. We also highlight the importance of developing BDAC in dynamically changing environments to create a competitive advantage. © 2022 Elsevier B.V. All rights reserved.
Citation - WoS: 3
Citation - Scopus: 4
Systematic mapping study on question answering frameworks over linked data
(Institution of Engineering and Technology journals@theiet.org, 2018) Ceren Ocal Tasar; Murat Komesli; Murat Osman Unalir; Tasar, Ceren Ocal; Komesli, Murat; Unalir, Murat Osman
Employing linked data technologies and semantic endpoints for question answering systems are expanding approaches among the researchers. Therefore systems that combine syntactic and semantic analysis and enrich input questions by sentence-level recognition are examined. A systematic mapping study is conducted to identify and analyse the studies from major databases journals and proceedings of conferences or workshops published between 2010 and 2017. With a set of 14 research questions inclusion and exclusion criteria are specified. 53 studies are selected as primary studies from an initial set of 845 papers. This study provides a mapping while focusing on the methods and identifying the gaps between required and existing approaches. Popular approaches which have gained the most attention among researchers are given as a conclusion. Moreover a comparison between the authors' study and related work in the literature is given to point out the differences and the contributions of their study. As the result of the comparison it is concluded that the study is a novel and original topic on question answering frameworks. © 2019 Elsevier B.V. All rights reserved.
Citation - WoS: 12
Citation - Scopus: 15
Use of data mining techniques to classify length of stay of emergency department patients
(INDEX COPERNICUS INT, 2019) Gorkem Sariyer; Ceren Ocal Tasar; Gizem Ersoy Cepe; Sariyer, Görkem; Tasar, Ceren Ocal; Cepe, Gizem Ersoy; Öcal Taşar, Ceren
Emergency departments (EDs) are the largest departments of hospitals which encounter high variety of cases as well as high level of patient volumes. Thus an efficient classification of those patients at the time of their registration is very important for the operations planning and management. Using secondary data from the ED of an urban hospital we examine the significance of factors while classifying patients according to their length of stay. Random Forest Classification and Regression Tree Logistic Regression (LR) and Multilayer Perceptron (MLP) were adopted in the data set of July 2016 and these algorithms were tested in data set of August 2016. Besides adopting and testing the algorithms on the whole data set patients in these sets were grouped into 21 based on the similarities in their diagnoses and the algorithms were also performed in these subgroups. Performances of the classifiers were evaluated based on the sensitivity specificity and accuracy. It was observed that sensitivity specificity and accuracy values of the classifiers were similar where LR and MLP had somehow higher values. In addition the average performance of the classifying patients within the subgroups outperformed the classifying based on the whole data set for each of the classifiers.