correct registration in the hospital can avoid wasting medical resources and shorten the time of diagnoses. The purpose of this research is to build a hospital registration system to help patients register in the hospital. This system can assist patients, who are not familiar with medical topics, to determine which department they should register by the spontaneous descriptions. Patients' spontaneous descriptions of syndromes were transcribed into text formation. These descriptive texts were analyzed by the keywords punctuation process and syndromes' descriptions analysis process to get the important features, including the syndrome words, degree words, affected region of body, frequency, time and place. Then, an information retrieval based department making process was employed to calculate the relation score between each department and these important features. Finally, the system suggested the most possible departments they should register to the patients. In the experiment, 50 descriptions collected from 50 patients are as the input to the system. The preliminary result shows the top one correct rate is 88%. The proposed system also shows the higher performance than the baseline system.