Indian Journal of Science and Technology, Vol 9(28), DOI: 10.17485/ijst/2016/v9i28/97818, July 2016 ISSN (Print) : 0974-6846 ISSN (Online) : 0974-5645 * Author for correspondence 1. Introduction Cognitive domains in Bloom’s Taxonomy (BT) are widely used in educational field to analyze student understanding and knowledge (Anderson, Krathwohl & Bloom, 2001) 1 . Cognitive level consists of six (6) levels, which start from a lower level of learning to a higher level. e lower level is known as knowledge level, which purpose to recall data from previous lesson. Second, comprehension level requires students to have an ability to explain the knowledge learns. ird, application level, student should able to apply the knowledge learned into an action. Fourth, the analysis level address student to investigate information they studied. Fiſth, synthesis level, student needs to link all the information and integrate it into something new. e final level is the evaluation level where the student may achieve when they able to stand for any opinion. To produce a good written exam questions, it must consists all of six cognitive levels. e lectures might have to consume time in classifying the question since they also need to follow the guidelines of constructing exam questions which are JSU and common problem like error typing when construct the questions might happen. e time allocates in classifying the exam question might extend and causes replication of exam questions. us, it could delay the development of exam questions since the process is done manually. Even though, the lectures need to follow the BT level, however, there is a high probability of wrongly classify the cognitive level of exam questions. Furthermore, the exam questions constructs are having a lot of bloom’s keywords which somehow two (2) keywords can belong to more than one level of cognitive or one (1) keywords belonging to two (2) cognitive levels. e problems with keywords belonging cause difficulty for educators to identify the correct level for each exam questions. Each exam questions generate must follow the bloom’s taxonomy guidelines. us, as constructors, they need to classify the exam question generates, according to Abstract This project is about analyzing the effects of classifying the written exam question into cognitive level of Bloom’s taxonomy. Correctly analyze and classify the written exam questions into correct cognitive level can generate a good set of exam questions. As known by many educators, classifying exam question into its cognitive level is a tedious task and required full attention by educators. Moreover, there are situations where one keyword of cognitive level belongs to more than one level which could be an issue of difficult to determine the correct cognitive level of questions. To solve the problem of classifying exam question faces by educators, the techniques information retrieval of text mining were implements in this project. Before that, question bank are required to perform text preprocessing to generate the clean data. The activities done under text preprocessed are such as data transformation, tokenization of question and stop word removal. The effects of classifying the clustered data being analyzed to study the possible hidden pattern of classifying based on Bloom’s Taxonomy. Keywords: Classification, Clustering, Text-Mining Studying the Effects of Performing Text Mining to Improve Classification of Clustered Questions based on Bloom Taxonomy Nur Suhailayani Suhaimi 1* , Norazam Arbin 2 and Nur Najihah Zulkifli 1 1 Faculty of Computer and Mathematical Sciences, Universiti Teknologi MARA (Melaka), Jasin - 73000, Melaka, Malaysia; suhailayani@tmsk.uitm.edu.my, najihah938@gmail.com 2 Faculty of Computer and Mathematical Sciences, Universiti Teknologi MARA (Johor), Segamat - 85000, Johor, Malaysia; noraz574@johor.uitm.edu.my