Doğal Dil İşleme Tekniklerini Kullanarak Türkçe Öykülerden Anahtar Sözcük Çıkarımı

Texts, which are written sources, are among the primary tools people use to transfer information. Extracting information from texts is of great importance in terms of accessing information. Checking thousands or millions of text data one by one by people to extract the necessary information leads to a lot of time and cost. Natural language processing techniques are used to eliminate these problems. Keyword extraction, widely used in natural language processing, is one of the basic methods that help summarize the content by determining a text's most essential and representative words. Keyword extraction from the text is possible in semantic context-based and frequency-based. Semantic context allows an inference based on the meanings of the words in the sentence. Frequency-based occurrence is obtained based on the frequency of repetition of the words in the text. Understanding the Turkish language's structural characteristics is essential to compute word frequency. Root words that do not take any suffixes and stems that gain a new meaning by taking derivational suffixes have a meaning on their own. However, words with inflectional affixes do not acquire a new meaning and cannot convey meaning independently. In this study, pre-processing and stemming operations were applied to Turkish short stories using a dictionary, and the frequencies of words were calculated to extract keywords. The results demonstrate that the proposed method provides a simple and effective approach for extracting keywords from narrative texts and applies to such text types. In addition, a new method was developed that includes previous studies in the stemming process. This method also provided more accurate results than other studies.