PR packages include press release writing, distribution via segmented media lists and news agencies, media monitoring and reporting processes.
An AI-supported solution is coming from the academicians of our country to transfer the Ottoman archives and resources, which is one of the biggest problems in social sciences, to today's Turkish language. The project developed by İshak Dölek, a doctoral student at Istanbul University-Cerrahpaşa with his advisor, has been accepted to the TÜBİTAK 1512 support program. The project, which also received support from KOSGEB, was transformed into an initiative with the completed OCR application.
*** This release is originally published in Turkish.
Google Translate Application translates the content you see on this page.***
Our country's academics are transferring Ottoman sources to today's Turkish with artificial intelligence. Osmanlica.com, a domestic initiative that started as a doctoral thesis project by Dr İshak DÖLEK under the supervision of Associate Professor Atakan KURT from Istanbul University-Cerrahpaşa Computer Engineering Department, achieved a 96 percent success in the Ottoman OCR process, which is the first step in transferring Ottoman resources to today's Turkish language.
"Osmanlica.com: Artificial Intelligence-Assisted Ottoman-Turkish End-to-End Transmission" started as a doctoral thesis in order to transfer all kinds of resources in Ottoman archives and libraries into today's Turkish in three steps: Ottoman OCR, Ottoman-Turkish Alphabet Translation and Ottoman-Turkish Language Translation. The project was later accepted to the TÜBİTAK 1512 venture support program through the Entertech Technopolis on the Avcilar Campus. Entrepreneurs became a company under the name of Mina ARGE Bilişim and developed the OCR project, which was the first step of the project. Having successfully completed the OCR project, Mina ARGE is still developing the Ottoman-Turkish Alphabet Translation project, which is a continuation of this project, with the support of KOSGEB and TUBITAK. The company, which has already reached 75% accuracy in its alphabet translation project, continues its R&D activities with a group of computer, language, literature and historians to achieve an accuracy rate of 95% in this application.
Assoc. Dr. Atakan KURT stated: “There are hundreds of thousands of Ottoman books, newspapers, magazines and documents in state archives, libraries and private collections at home and abroad. It is practically impossible to translate so many documents into Turkish manually by human efforts. There have been great advances in IT and AI lately. In this way, problems that could not be solved before can be solved in the light of new developments. We are adapting these new technologies to transfer the documents in Ottoman archives and libraries into today's Turkish. We have achieved significant success in the OCR and transliteration stages. This problem, which has been largely solved in Western countries, has not yet been solved in our country. Our aim is to translate books, magazines, newspapers and archive documents in Ottoman Turkish into today's Turkish so that they can be read and understood by ordinary people and especially the new generation; and to develop the necessary software to shed light on history by unearthing documents that have perhaps never been examined or read until now. In our opinion, this project is a candidate to be Turkey's vision project in social sciences. With this project, hundreds of thousands of books, magazines, newspapers and millions of archive documents will be able to be translated into Turkish quickly.”
Stating that they are carrying out more than one study to translate Ottoman documents into today's Turkish, Dr. İshak DÖLEK, on the other hand, said that "in addition to the Ottoman OCR process, work continues on other projects such as Ottoman-Turkish alphabet translation, language translation from Ottoman to contemporary Turkish, and rika OCR, that is, the conversion of Ottoman handwriting from picture to text with OCR." “For example, we achieved an accuracy rate of 75% in the alphabet translation, which converts an Ottoman text in the Arabic-based Ottoman alphabet into a Latin-based Turkish alphabet. Our alphabet translation application is still the only application that provides service on the Internet.”
Dr. İshak DÖLEK continued his statement: “We achieved 96 percent success in the OCR application, which was tested with an original Ottoman dataset consisting of 21 pages, 3 thousand words and 23 thousand letters written in Ottoman nesih line. In other words, 96 out of every 100 letters were correctly recognized by the application, which is the highest accuracy rate ever achieved. In these tests, we compared our application with five different OCR applications, four of which were from abroad and one from Turkey, and we published the results in the form of a paper at an international conference and an article in a journal. We are happy for the contribution that our practice, in which we continue our R&D activities, will make in social sciences. More than 20 thousand users have tried our OCR application on Osmanlica.com and more than 100 thousand users have tried our alphabet translation application so far. These numbers continue to increase every day. This situation motivates us even more.”
Contact: Tülay Genç | [email protected] | +31 30 799 6022
