Inferring User Search Intention Based on Situation Analysis of the Physical World
ABSTRACT Ubiquitous search of the physical world has recently received significant attention. In this paper, we investigate situation
analysis based user intention recognition, which can be useful to retrieve information that may perfectly satisfy a user’s
needs. We first introduce a hierarchical user intention model based on CRFs (Conditional Random Fields). With the model, a
BP (Belief Propagation) based inference method is proposed to recognize user search intention based on situation analysis.
A variety of sensing mechanisms are adopted to collect context information of the physical world for robust and reliable situation
analysis. We developed a prototype in a real home setting and experiments were performed to examine the effectiveness of the
KeywordsSituation analysis-search intention-CRFs-ubiquitous search