With the constant stress from work load and daily lifepeople may show symptoms of melancholia. However, mostpeople are reluctant to describe it or may not know that theyalready have it. In this paper a novel system is proposed todiscover clues from patient’s interaction with psychologist orfrom self-recorded voice or text messages. A user friendlyinterface is provided for patients to input text messages or recorda voice file by mobile phones or other input devices. A speech-totextconversion software is used to convert voice mails to simpletext files in advance. Based on the text files, a data mining modelis used to discover frequent keywords mentioned in the text orspeech files. The association rules can be used to helppsychologists diagnose patients’ degree of melancholia.Experimental results show that the proposed system caneffectively discover melancholia keywords.
展开▼
