Discovering the Characteristic of Information for Advanced Information Retrieval, focused on Temporal, Spatial and Human-Centric

다국어 초록 (Multilingual Abstract)

With the development of the web technology including cloud computing and big data technology, the vast information is cumulating in the web. From this perspective, the information retrieval services in the past provide a searching service simply based on the huge information. However, many current researches which not only classify the information according to subject or source, but also offer the tailored search result in response to requests from the user, are being vigorously pushed forward. In this climate, recent studies are showing a great interest in the characteristics of information, especially, temporal, spatial and human-centric. These three characteristics of information can be used as an important element, either individually or together, for enhancing the quality of information retrieval service. Particularly, under the situation that the mobile internet access rate overtakes the PC environment, considering the users’ mobile environment is essential for evolving to an advanced future information retrieval service. Moveover, the three characteristics of information are also considered obligatorily as a crucial factor both to understand a user’s various situations and provide an personalized service. Although there have been many studies on this research area, more research is needed on a few topics. In this thesis, I suggest a solution that should be addressed necessarily, and then identify the results through the experiments. To put it concretely, in temporal characteristic area, I propose the temporal extraction method to utillize temporal information, and then I build the historical retrieval model based on period query using the extracted temporal information. Consequentially, this thesis identifies not only the possibility of extracting temporal information, but also the availability of the proposed period query. Next, I design a mobile location based service that considers a users’ mobile environment based on argumented reality in spatial characteristic area. Particularly, I try to solve a drawback of current location based service by suggesting MixedWalk that is a novelty location based service based on argumented reality technique. Finally, I focus on the representation of sensing the environment through sight, hearing, touch, smell and taste, called sensation information, in unstructured data, especially, textual data on the Web. According to some recent studies, a decision making is based on the sensation information from a unconscious mind. Hence, the sensation information affects a decision making of human before a cognitive thinking. Even though sensation is the first process of human experience against the environments, the study of sensation information extraction is neglected due to lack of sensory expression and knowledge comparing with the sentimental analysis or opinion mining. Hence, I identify which sensation feature has a strong influence on human perceptual experience in a specific topic of corpus. Then, I evaluate our method by comparing with several baselines in terms of the accuracy. Through the research of three information characteristics, this thesis wants to contribute the development of both information service and advanced personalized service.

목차 (Table of Contents)

• SUMMARY i
• TABLE OF CONTENTS iii
• LIST OF FIGURES v
• LIST OF TABLES vii
• NOMENCLATURE viii
• CHAPTER 1 INTRODUCTION １
• CHAPTER 2 RELATED WORK ５
• 2.1 Temporal Information ５
• 2.1.1 Information Retrieval using Temporal Information ５
• 2.1.2 Extracting and Indexing of Temporal Information ６
• 2.2 Spatial Information ８
• 2.2.1 Mobile Augmented Reality Service ８
• 2.2.2 User Interaction in Mobile AR １０
• 2.2.3 Moving Query process for Spatial Network Databases １２
• 2.3 Human Centric Information １３
• 2.3.1 Sensation Information and Its Application １３
• 2.3.2 Human Expperience Mining in Social Media １４
• 2.3.3 Word Sense Disambiguation １６
• CHAPTER 3 METHODOLOGY １８
• 3.1 Temporal Information Area １９
• 3.1.1 Extracting Temporal Information Using Machine Learning Techniques １９
• 3.1.2 Historical Information Retrieval Model based on Period Query ２８
• 3.2 Spatial Information Area ３４
• 3.2.1 A Thee-Layered Service Model for Virtual Movement ３５
• 3.2.2 Virtual Movemenet Queries ３８
• 3.2.3 Query Processing ４４
• 3.2.4 Prototype Implementation and Evauation ４８
• 3.3 Human-Centric Information Area ５０
• 3.3.1 Sensation Feature Extraction and Classification ５２
• 3.3.2 Expermiments and Evaluations ５９
• 3.3.3 Mobile users’ context awareness model ６８
• CHAPTER 4 DISCUSSION ８３
• CHAPTER 5 CONCLUSION ８５
• REFERENCES ８７
• 국 문 초 록 ９４