Data classification has become a widespread field of knowledge, especially in light of the tremendous development of information technology and expert systems. Among the most important fields of knowledge that are concerned with data classification is the classification of text data. In our research, we present a model for an expert system whose task is to classify university theses and theses based on their Arabic titles and using the Nave Bayes algorithm. In the research, a database was used, which was compiled using the (Scrapping) technique, which is to collect data from repositories spread on the Internet. A database of (6100) titles for a master''s thesis and a doctoral thesis was built. These titles were divided into (seven) categories representing the general specialization of the thesis - thesis. Among the most important challenges faced by the research is the nature of the Arabic language and its degree of complexity. The results of the research proved that the model was able to classify theses and theses at a rate of (86%), which is a very promising percentage. The proposed model was also evaluated by distributing a questionnaire to a number of specialists in computer science and information systems. The questionnaire included measuring the impact of two independent variables, namely, the accuracy and importance of the expert system. The results proved that there is a statistically significant effect of the independent variables on the dependent variable (the desired benefit of the expert system). The results of the research open new horizons in the process of classifying textual data, especially the Arabic language, and open the way towards implementing other algorithms and comparing the results.