DMCA
Discovery and Raking of Significant Trails
BibTeX
@MISC{Papadogkonas_discoveryand,
author = {Dikaios Papadogkonas and George Roussos and Mark Levene},
title = {Discovery and Raking of Significant Trails},
year = {}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
Trail-based representation has been identified as a useful approach in exploiting context histories. Indeed, trails provide within a single framework an effective way to record, structure and represent interactions between users and physical or digital resources. Yet, this approach faces several challenges before it becomes practical at large scale since trail-based navigation requires considerable storage and computational resources when carried out naively. In this position paper, we propose an architecture for a query engine for trails and associated discovery and ranking techniques that can address such performance considerations and thus support processing of large numbers of recorded trails. In particular, we propose a stochastic model for the representation of trails and trail aggregates, and suitable data structures for efficient storage, filtering and retrieval which result in significantly improved performance. Furthermore, we propose algorithms and associated metrics that we use to rank trails and identify significant ones. Such significant trails can be subsequently used to guide search and navigation. We conclude by illustrating the use of these techniques in the context of three case studies.
Keyphrases
significant trail computational resource digital resource associated metric large number context history trail aggregate several challenge stochastic model trail-based representation case study suitable data structure efficient storage performance consideration trail-based navigation useful approach support processing position paper effective way large scale considerable storage single framework represent interaction significant one query engine
