Yapay zeka teknikleri ile desteklenmiş bulanık bilişsel haritalama yöntemi kullanılarak kurumların dijital dönüşümlerinin değerlendirilmesi = The evaluation of digital transformation in institutions using fuzzy cognitive maps supported by artificial intelligence techniques

Abstract

Dijital dönüşüm süreci, insanlar, makineler, ekipmanlar arasında gelişen teknolojiler kullanarak birbirleriyle etkileşimli sistemler oluşturmayı amaçlamaktadır. Dijital teknolojilerin kullanımının artmasıyla şirketlerin rekabetçi kalabilmek için yeni stratejiler benimsemesini gerektiren değişiklikler ortaya çıkmaktadır. Gelecekte, organizasyonlar ürün tasarımından dağıtıma, müşteriye ve geri bildirime kadar tüm zinciri içeren fiziksel ve sanal sistemlere ihtiyaç duyacaklardır. Bu nedenle, organizasyonların dijital dönüşümü anlamaları ve uygulamaları elzemdir. Literatür incelendiğinde dijital dönüşüm olgunluk modelleri çalışmalarında genellikle mevcut durum analizleri yapılmış ve sayısal yöntemler içeren çalışmaların eksikliği görülmüştür. Dijital dönüşüm gibi karmaşık modellerde süreç yanlızca alanında uzman kişiler tarafından kesinlik içermeyen neden-sonuç ilişkileri ve birçok geri bildirim döngüsü yoluyla birbirine bağlı ilgili kriterler şeklinde temsil edilebilir. Bu sebeple çalışmada bir esnek hesaplama yöntemi olarak karmaşık sistemlerin modellenmesi ve modeldeki ilişkilerin bağlantılar ile sağlandığı bulanık bilişsel haritalar yöntemi seçilmiştir. Bulanık bilişsel haritalar yöntemindeki, uzmanların oybirliği ile modelin kriterleri arasındaki ilişkilerin ortaya çıkarması aşamasındaki zorluğun ortadan kaldırılması ve uzmanların subjektif ifadelerinin sayısallaştırılabilmesi için ise bulanık DEMATEL yöntemi kullanılmıştır. Bu çalışmada ilk olarak dijital dönüşüm ile ilgili gözardı edilecek bir kriter bırakmamaya çalışılarak, literatür ve paydaşlardan alınan bilgilere göre 20 kriter içeren kapsayıcı bir model oluşturulmuş ve bulanık DEMATEL yöntemi kullanılarak kriterler arası ilişkiler belirlenmiştir. Bu ilişkiler kullanılarak daha sonra bulanık bilişsel haritalar yöntemi ile öncelikle statik analiz yapılmıştır. Statik analiz yapılırken öncelikle temel senaryo oluşturulmuştur. Daha sonra sırasıyla her kriter için sabitlenmiş değerler kullanılarak algoritma boyunca bu değerlerin değiştirilmemesi sağlanmıştır. Her kriterin çıktı kriterini etkileme düzeyinin belirlenebilmesi için 19 senaryo oluşturulmuş ve temel senaryodaki çıktı değerinden farklar ortaya çıkarılmıştır. Kriterlerin sonucu en fazla etkilemesi bakımından önem sıralaması tespit edilmiş, zeki teknoloji, zeki organizasyon ve zeki imalat kriterlerinin ilk üç sırada yer aldığı görülmüştür. Çalışmanın dinamik analiz kısmında savunma sanayine üretim yapabilecek altı firma için mevcut durumlar 88 sorudan oluşan anket aracalığı ile elde edilmiş, firma senaryoları oluşturulmuş ve gelecekteki dijital dönüşüm seviyeleri diferansiyel evrim algoritması kullanılarak desteklenmiş bulanık bilişsel haritalar yöntemi ile analiz edilmiştir. Seçilen altı firma mevcut durumda dijital dönüşüm açısından iyi, orta ve kötü seviyelerde bulunan firmalardır. Son olarak aynı altı firma için en önemli üç kriterin mevcut durumları revize edilerek duyarlılık analizi yapılmış ve değişimler incelenmiştir. Duyarlılık analizi yapılmasında amaç, firmalara dijital dönüşümlerine doğru başlangıç noktaları konusunda karar destek sağlamaktır. Bu tez çalışması kapsamında, önerilen modelin ve kullanılan entegre yöntemin dijital dönüşüm süreci konusunda bir karar destek sistemi ve organizasyonlara yol gösterici olacağı ortaya konulmuştur.

The digital transformation process aims to create interactive systems between people, machines and equipment by using developing technologies. With the increasing use of digital technologies, changes are emerging that require companies to adopt new strategies to stay competitive. Organizations will need physical and virtual systems that cover the entire chain, from product design to distribution, customer and feedback in the future. Therefore, it is fundamental that organizations understand and implement digital transformation. When the literature is considered, current situation analyses are common in studies of digital transformation maturity models, and there was a lack of studies involving numerical methods. Digital transformation is a phenomenon that emerges with the aim of producing flexibility, efficiency, technological and quality products or services by providing improvements in traditional production. It is based on increasing the production efficiency, value gains and social welfare of enterprises with the help of advanced technologies. Digital transformation is a process that can be applied to various fields, and this process includes fundamental changes in the way people do business in institutions, in addition to maintaining social activities such as communication, health and entertainment with technological tools. The increasing importance of digital transformation in today's world has revealed the need for the analysis of this process. It is very important that institutions can accurately determine their digital transformation levels and make realistic and concrete plans according to the current situation. However, despite the increasing interest in the issue of digital transformation, it is still an unresolved concept. In the studies modeling the digital transformation process in the literature, survey studies have generally been put forward on the basis of determined criteria. With these survey studies, the current levels of digital transformation of organizations have been revealed. However, the interpretation of the relationships between the criteria in the models has not been addressed in many studies. In addition, there are few studies in the literature, which are based on analytical methods, which criteria are more important in terms of the digital transformation process. The first main purpose of the study is to create an inclusive model of a complex structure such as the digital transformation process. In such a comprehensive model, it is very important to be able to represent the knowledge of stakeholder groups such as the private sector, academia and consultancy companies. In addition, a well-prepared questionnaire should be included in the study, in which the participants can best express the representation of their perceptions. With these features, it is aimed to fill the basis of the study and to add value to the literature with the proposed integrated method. The second goal is to create a fuzzy cognitive maps model that is robust, semi-quantitative, allows feedback in dynamic interactions, can easily allow for adding and removing criteria due to its flexible structure, and can model a variety of qualitative, dynamic and participatory complex systems that represent many real-life problems. With this method, it will be possible to determine the predictions of the organizations in terms of the future digital transformation processes, beyond the studies that only analyze the current situation in the literature. In complex models such as digital transformation, the process can only be represented by experts in the field as imprecise cause-effect relationships and related criteria interconnected through many feedback loops. For this reason, the fuzzy cognitive maps method has been chosen as a soft computation method in the study, in which complex systems are modeled and relations in the model are provided by connections. And, also fuzzy DEMATEL method was used to eliminate the difficulty in revealing the relations between the criteria of the model with the consensus of the experts in the stage of Fuzzy Cognitive Maps. In this study, 20 criteria were chosen to represent the different aspects of a model and the relationships between them were determined using the fuzzy DEMATEL method. Then, using these relations, a static analysis was performed to establish the order of importance of the criteria using the fuzzy cognitive maps approach, and it was discovered that the criteria of intelligent technology, intelligent organization, and intelligent manufacturing ranked first, second, and third, respectively. In the dynamic analysis, the current situations for six companies that can produce for the defense industry were obtained through questionnaires consisting of 88 questions. Company scenarios were created and future levels of digital transformation were analyzed with fuzzy cognitive maps method developed using genetic algorithm. The six selected companies are currently good, medium and bad in terms of digital transformation. Finally, the current status of the three most important criteria for the same six companies was revised and a sensitivity analysis was made and the changes were examined. The sensitivity analysis is used to provide decision support to companies on where they should start their digital transformation. For this purpose, the initial value of the three most important criteria in their current situation for each firm is set to 1, which indicates the best level. Thus, the future predictions of the success of the digital transformation process of change were taken. The output variable of the digital transformation model is reached a steady state value of 0,9689 after eight time periods in the first firm, 0,9707 after seven time periods in the second firm, 0,9683 after eight time periods in the third firm, 0,9695 after seven time periods in the fourth firm, 0,9784 after six time periods in the fifth firm, and 0,9840 after six time periods at the sixth firm. According to the results, if the second firm focuses on the three most important criteria, intelligent technology, intelligent organization and intelligent manufacturing, it will have the opportunity to surpass the third and fourth firms, which are better than itself in the current situation. The results determined in the study give companies the opportunity to see the results of their changes and focus points. In this thesis study, which aims to analyze the digital transformation process, As a result of the examination of digital maturity models in the literature and expert opinions, a comprehensive digital transformation process model covering all existing models has been revealed. By using fuzzy DEMATEL and BBH methods together, it was ensured that the completely objective thoughts of the experts were reflected in the model. It has not been found that the two methods are used in the literature as in this thesis study. Unlike studies in the literature in which the digital transformation process is analyzed, future situation predictions are provided to organizations based on their current situation. With the use of nonlinear Hebbian and genetic algorithms as learning algorithms in the fuzzy cognitive maps method, the steady states that can arise in complex problems in the fuzzy cognitive maps are prevented from reaching the same values. Within the scope of this thesis, it has been revealed that the proposed model and the integrated method used will be a decision support system and a guide for organizations on the digital transformation process. The model created and the methods used for the analysis of the model can be easily applied to all sectors with the desired changes thanks to its flexible structure and can be used as a decision support system.