Creating expert decision systems with qualitative datasets obtained from education management software by using opinion mining algorithms

The size of produced data has significantly increased with the use of internet-connected smart devices in all areas of life. This data in the Internet environment are generally nonclassified form. This dataset is called unstructured data. It is not possible to obtain information without using any technique on the unstructured data. Therefore, data mining techniques have been developed in order to process and obtain meaningful information from these dataset. Data mining techniques are intended to make meaningful information on the data. Especially with the increased use of social media and information communication system, processing and classification of big data has a major problem. Despite the rapid development of multimedia technology, the most abundant form of data on the internet is text data. Therefore, text mining that is a branch of data mining has been developed. In order to use text mining techniques in an efficient manner, the structure and grammar of natural language should be well known. A distinct text mining techniques should be developed for each natural language due to differences in language structure and grammar system. In this thesis, data mining, text mining, opinion mining techniques has been examined. Firstly, a literature review has been done on these methods. In the implementation phase of the thesis an expert decision support system has been developed. This system can be automatically detected that the user will have the idea. In order to test developed expert system, it have been utilized the opinions of users in the Co-operative and Debater Learning Object Warehouse (CDLOW) system. In the developed system, more intelligent classifier has been used. Performance of these classifiers has been analyzed. Used intelligent classifiers are Naïve Bayes and K Nearest Neighborhood (k-NN).