| Summary: | 碩士 === 逢甲大學 === 工業工程與系統管理學研究所 === 96 === Human daily experience constantly incorporates larger amounts of important information. The social aspects of human work, leisure, and life can be helped or hindered by this tremendous volume of information. Social computing promises more harmonious leisure and more efficient work by providing better information retrieval. People put information that they have experienced into episodic memory; future searches for memories will depend on this episodic structure. The human memory process has receiving, conversion, storage or retrieval phases, all of which can encounter problems.
Information technology (IT) systems can support human episodic memory. However, retrieval of these memories is often complicated by that fact that these memories are often similar, complex and trivial. This study will support the development of augmented memory system architecture. In order to achieve effective user support, how to effectively retrieve the necessary memory will be a key issue for this system.
In this study, we proposed two solutions for this core issue. (1) Use context-aware environmental context data to make context-records (Context & Content). The difference between records of context-record method and records of traditional method is that context-records of an event’s context elements (people, places, and objects) can assist future recall. In order to verify whether these proposals can support human needs, our first set of experiments compare the traditional-records method with the context-records method, for the retrieval of required memories.
(2) Use typing and clicking to users to tag information according to expertise. A human will recall memories in an area of expertise differently than memories in an unfamiliar area. The different level of knowledge structure can be divided into expert-level and novice-level. In order to understand the role of expertise in absorbing information, retrieving information, the difference in tagging behavior, this study conducted a second set of experiments to provide further support to memory system design.
Preliminary results from the first set of experiments showed that the context-records method had significantly (p-value < 0.05) better efficiency than the traditional-records method. The second set of experiments suggests that the same tagging methods will be appropriate for expert knowledge and for novice knowledge.
|
|---|
