<![CDATA[Twitter is a service application that is popular because it can be used to interact and communicate in everyday life. A lot of various new types of automation software increases to disseminate information immediately. Twitter does not strictly check the automation tweet, therefore there is no prevention of the use of bot on a regular basis. Low restriction of the use of automation services on Twitter led to the emergence of market Spam-as-a-Service consisting of counterfeiting program, abridgement ad-based on service and sales account. Each of these services allows spammers to do the spam deployment process by using automation tweet services. So it is necessary to do a research on the classification of the tweet to know the type of category is included in the category of spam or not spam. Spam classification process begins with the preprocessing consists of several stages, namely; cleansing, case folding, tokenization, filtering and stemming. The next step are process of term weighting, until the process of classification using Improved K-Nearest Neighbor method. The results obtained on the basis of implementation and testing research of the classification of Spam on Twitter produces an average Precision of 0.8946, Recall of 0.9405, F-Measure of 0.9155 and results accuracy of 89.57%. Where is the number of documents, a comparison or balance the proportion of training data and the determination of k-values that are used too well or whether the process of classification of the document.]]>
Copyrights © 2018
