ThingsNavi: finding most-related things via multi-dimensional modeling of human-thing interactions

With the fast emerging Internet of Things (IoT), effectively and efficiently searching and selecting the most related things of a user's interest is becoming a crucial challenge. In the IoT era, human interactions with things are taking place at a new level in ubiquitous computing. These interactions initiated by humans are not completely random and carry rich contextual information. In this paper, we propose a things searching approach based on a hypergraph, called ThingsNavi, where given a target thing, other related things can be found by fully exploiting human-thing interactions in terms of multi-dimensional, contextual information (e.g., spatial information, temporal information, user identity). In particular, we construct a unified hypergraph to represent the rich structural and contextual information in human-thing interactions. We formulate the correlated things search as a ranking problem on top of this hypergraph, in which the information of target things can be propagated through the structure of the hypergraph. We evaluate our approach by using real-world datasets and the experimental results demonstrate its effectiveness.