시계열 모델링에 기반한 라이프로그 데이터 관리 = Lifelog Data Management Based on Time Series Modeling

다국어 초록 (Multilingual Abstract)

Wearable computing seems like the next big trend in technology. It creates a new way of human-computer interaction through a small body-borne device that is always on and ready as a helpful assistant. Services that offer attractive new functionalities through the use of various time-series data as lifelogs have been drawing attention. From watches and fitness trackers to sensor-filled clothes and Google Glass, various devices have captured the interest of the public and analysis from industry insiders. While the market value is expected to reach $10 billion by 2015, there is much ambiguity regarding the software applications that will bring such devices to life. Now lifelogging activities are broadly linked to our daily lives seamlessly and transparently processing our emails, photographs, geo-spatial locations and health records archiving them back to the cloud storage. Technical challenges are then the indexing of heterogeneous media including image, audio, video and human knowledge on their syntactic and ontological meaning expansion. These issues also clearly distinguish lifelog data management from legacy data archiving activities in many aspects.

목차 (Table of Contents)

• 1. Introduction １
• 2. Related Works 4
• 2.1 Existing Lifelog Management Architecture 5
• 2.2 Case Study in Lifelog Management 6
• 3.Time Series Modeling: Converting Sensor Streams to Time Series Data 9
• 3.1 Time Series Data Model 10
• 3.2 Lifelog Data Model 14
• 4. Data Translation Algorithm 22
• 5. Conclusions 27
• References 29