Proposal of time series data retrieval with user feedback

This paper introduces the concept of relevance feedback for retrieving time series data. Data in medical field handles various kinds of time series data, such as EHR (electronic health record) and consumer health data. Analyzing and accessing such time series data gets to be crucial for doctors to improve the quality of medical treatment. When we are going to develop an interactive interface for supporting a doctor to retrieve time series data of patients, it is supposed that similarity judgment on time series data depends on the knowledge and purpose of a doctor. Therefore, we employs relevance feedback-based retrieval system, which modifies similarity calculation based on the feedback information given by a doctor. In order to obtain enough feedback information from a doctor, we are going to employing two kinds of feedback mechanism: relevance feedback and graphical annotation-based feedback. This paper describes the concept of our relevance feedback-based retrieval system and the feedback mechanisms.