The number of text documents in digital form has grown enormously in size, including scientific articles or essays. as a consequence, it is of great practical importance to be able automatically organize and classify documents. there are tools to classify the documents, but they are mostly for english documents. indonesian is different from english, for example, indonesian doesn't have past or future form. therefore we develop an applications that could classify indonesian documents efficiently and effectively. the method used classify documents of scientific articles is method keyword similarity. making these applications using the PHP progamming language and database software in use is MySQL. testing of this application by using abstract data writing or scientific research from student of gunadarma university mayoring information system. data used in this research of 50 data abstraction. from the test result of 50 document classified abstraction, there are four classification of research topics or the writing of scientific are information system, web progamming, desktop progamming and computer networks. test result obtained that only 3 documents that are not in accordance with the manual classification, means there is 6% of that is not appropriate and 94% of the appropriate, of the 50 classified documents.