A Study on Web Search and Analysis based on Typicality
Kosetsu Tsukuda · Kyoto University Research Information Repository (Kyoto University) · 2014
With the increase in the amount of information on the Web, the number of people who access web-based information increases.A general and common goal of a web search is to find information about an unknown topic.Although there are various ways to support users when they search for a topic, one solution is to realize a search based on typicality.Cognitive psychology suggests that showing typical instances in a category is useful to understand the outline of the category.After understanding the outline of the category, it is helpful to achieve greater understanding of the category by showing atypical examples and unexpected examples.Therefore, in this thesis, we focus on searches for and analysis of data based on typicality and unexpectedness.We consider the two types of typicality of information, such as an object, an object set, and a relation.The two types are "typicality based on data (TD)" and "typicality based on social recognizability (TSR)."We also consider typicality based on central tendency and frequency of instantiation, which were proposed in cognitive psychology, for TD and TSR.This thesis includes the following three research topics: