Specific keyword extraction from unstructured curriculum vitae using deep learning methods

In today's world conditions, technology is developing day by day and the number of data on the internet is increasing considerably. With the increase in the diversity in the informatics sector, which plays an active role with these developments, new positions and new working areas are emerging. Having many data sources on the internet does not indicate that they are meaningful. Due to the rate of increase in data, it becomes increasingly difficult to distinguish between necessary and unnecessary information. The important thing is to be able to extract meaningful data from meaningless data. Data that provide semantic integrity and work is very valuable in every field today. In order to make people's job easier and to seize various opportunities in the computer age, very intensive studies are carried out in this field. When people make job applications in various fields, they send their personal resume information that introduces them to companies. In this way, the company learns the necessary information for the person to be recruited. Companies with many resumes in their hands and pools want to extract keywords with various methods to classify the people to be recruited and to use the data they have in the most efficient way for them, as in other sectors. With the help of keywords, the relevant category required for a text or a resume can be learned, and the shortest summary information about the subject can be obtained. In this way, classification can be made as desired and a semantic integrity can be provided. Keyword definition is not made for each file in each dataset. It can take a lot of time manually to extract keywords correctly, the rate of mistakes is high and manual intervention is very difficult. Therefore, there are many different approaches and studies for keyword extraction. Statistical methods, linguistic methods, machine learning algorithms, deep learning methods and recently increasing artificial neural network methods have been used in keyword extractions. In the early studies, keyword suggestions were made for the most common words on the resume. In recent studies on artificial neural networks, it is aimed to deepen the learning for keyword extraction, to increase the accuracy rate more and to perform faster processing. In this thesis, approaches have been made on keyword extraction on resumes of the informatics sector. The data pool created from the resumes and the data in the data pool were used together with their explanations in order to create semantic integrity. Deeper learning is aimed with artificial neural networks, and it is used in keyword words studied on the target sector in order to give more accurate results. Thanks to keyword extraction; It is aimed to achieve semantic integrity and classify individuals more accurately than their background. It is aimed to achieve more successful classification with keyword extraction for large data sets.