Computer sciences department at the University of Technology grants MSC to the student Mohammed AbdulJalel for his thesis "Text Classification Using Evolving Intelligent System"
In this thesis, these problems solved by using the proposed classification system based of fuzzy logic and genetic algorithm. The proposed system passes through four phases, that are data collection, preprocessing, feature extraction and classification phase. In data collection phase, this system is based on the Sandy Hurricane event, where data is collected for the time period from 10.27.2012 to 11.7.2012. Set of 1002 tweets take from initial data as testing data and set of 1000 tweets as training data. The pre-processing phase that used to enhance the texts to get the best possible results, where the text contains symbols, hashtag, numbers and unwanted words. In the pre-processing phase, six steps are used such as hashtag manipulate, remove additions, tokenization, remove stop words, stemming and lemmatization. In feature extraction phase, eleven features are extracted from each tweet to give a more accurate result. In classification phase, proposed system used fuzzy logic and genetic algorithm. The proposed system in classification phase passes through three steps that are fuzzification, inference and defuzzification. The fuzzification phase is used to convert the real inputs into fuzzy sets and compute degree of membership for each value in feature vector. Genetic algorithm used through fuzzification step to generate new membership degree. The inference phase is based on fuzzy rules, where the fuzzy rules are a collection of linguistic values for each.
The discussion committee includes Asst.Prof.Dr.Suhad Mallalah, Asst.Prof.Dr.Hasaneen Amer from University of Technology/ computer sciences department and Asst.Prof.Dr.Haitham Abdullatef from Al-Nahrain University/computer sciences as members and Asst.Prof.Dr.Yusra Hussein and Lect.Dr.Nuha Jamel as supervisors.