Knowledge Based Chatbot with Context Recognition

More than 200 questions enter every day during the acceptance period of new students in EEPIS, where some of the questions use the same answer. A collection of incoming questions can indeed be answered entirely but the work is less effective and efficient because it makes the answerer answer the same question and it is wasting time. This condition make us to create a special chatbot for the admission of new students in EEPIS. This research processes information using the Text Mining method by improving it ability to handle words using synonym dictionary and stemmed stopword dictionary. In the Text Mining process other results will be generated in the form of unknown words which form the basis of dictionary learning synonyms. After the information has been processed, it become base knowledge for chatbot and the best answer search calculation will be performed by using context recognition for filtering related information. Then it will be calculated using binary cosine similiarity and the best calculation results will be used as answers. The results of this study showed an accuracy of 87.09% with testing 176 questions, 25 answers and creating new synonym dictionary in a case study of new student admissions in EEPIS.