Using word-cooccurrence to determine health problems from health-care documents

摘要

This research aims to determine health problems, especially symptom occurrence on the health-care documents for the automatic problem-diagnosis system. This research concerns the actual event of the symptom occurence mostly expressed by verb phrase on a simple sentence. Thus, we apply a word cooccurrence with the symptom concept having the first word as a verb to determine the symptom occurrence on a certain simple sentence. However, the research still contains two main problems, the first problem is the ambiguous word cooccurrence. The second problem is to determine the number of words (starting with verb) to identify word co-occurrence with the symptom concept from the verb phrase after the stop word removal. Therefore, Pearson Correlation was proposed as the solution to these problems. The results of this research can provide the high precision of the health-problem determination.