Evaluation of Erzurum Ahkâm Defter (m.1742–1749) numbered 1 with the text mining technique

Depending on the technological developments, a large amount of data has begun to be formed. The data in question can be analyzed using data mining and text mining methods, and useful information can also be extracted from these data. While data mining is applied on structured data, text mining is applied on semi-structured or unstructured data. When we look at the archives or databases, it can be seen that the amount of data stored in them is very high. In this context, Ottoman archival documents can also be considered as an important data source. Most of this data is in the form of unstructured or semi-structured textual data. The process of manually arranging and analyzing large amounts of textual data, and therefore obtaining useful information, is a rather difficult and time-consuming process. In this study, the textual judgment data contained in the Erzurum Ahkâm Defter numbered 1 were automatically classified according to their subjects by applying text mining methods. Decision Tree, Support Vector Machine, Gradient Boosted Trees, Random Forest and K Nearest Neighbor algorithms were used in the classification process. The classification achievements of the algorithms were evaluated according to the document frequency, binary weighting, term frequency, inverse document frequency and inverse class frequency weighting methods used. The most successful result was obtained from the Gradient Boosted trees algorithm according to the inverse document frequency weighting method with an accuracy rate of %0,812. Suggestions were made for the development of the classification model created and for the text mining applications to be applied on the Ottoman archive documents.