Metadata for efficient storage and retrieval of life log media

Abstract

Every situation and event in daily lives can bear the values as useful information for later times. life log records information occurring in one individual's daily life, which can disappear without the user's recognition, in the storage device. And, whenever necessary, re-searching can be attained. The types of information that can be logged in life log system are the ones of surroundings (voice, image, etc) that are entered via every kind of sensors attachable to human bodies. These multimedia data have disadvantages such as requiring high-capacity memory space in saving and also acquiring inaccurate search results together with long searching times. Therefore, this paper defines the metadata for saving and searching the life log media in order to access these kinds of problems efficiently. This can be performed effectively from the aspects of saving and searching by recording in metadata such as EXIF(Exchangeable image file format) at logging stage of the information entered from each sensor. After defining these metadata, this paper generated virtual data fit for the schema and performed searching experiments. The searching results varied depending on the data storage spaces used. The data storage spaces used in the paper are PostgreSQL(RDBMS), eXist(Native XML DBMS), Apache Lucene, and the fastest searching result was shown in the Apache Lucene. ？2008 IEEE.