Take-home Message
HopRank: A Biased Random Walker
HopRank: How Semantic Structure Influences Teleportation in PageRank
Lisette Espín-Noboa, Florian Lemmerich, Simon Walk, Markus Strohmaier, and Mark A. Musen
Modeling Navigation on BioPortal
TheWebConf 2019 / Semantics Track
Network Structure
Visible or known
Navigation
Where to go and how?
Biased Transitions
Towards khop neighbors
HopPortation
Khop probabilities
| 𝜷0 | 𝜷1 | 𝜷2 | 𝜷3 | 𝜷4 |
Transitions | 0 | 1 | 100 | 0 | 15 |
Smoothing | 1 | 2 | 101 | 1 | 16 |
Normalization | 0.008 | 0.017 | 0.835 | 0.008 | 0.132 |
HopRank is an extension of PageRank and builds upon insights from information foraging and decentralized search.
We assume that teleportation is not fully random but rather distributed non-uniformly across different khop neighborhoods.
On BioPortal, users tend to be biased towards certain khop neighborhoods depending on the type of navigation.
Model Selection on BioPortal. This heatmap highlights the model—with lowest BIC score—that best describes the number of transitions per ontology and navigation type. ��HopRank outperforms the other models 89% of the time, especially when browsing concepts via details (DE), direct click (DC) and expand (EX). ��When transitions are scarce (i.e., the other 11%), BIC penalizes HopRank since it has more parameters than the other models �(except Markov chain: MC).
Lisette Espín-Noboa | Lisette.Espin@gesis.org | @lespin | arXiv: 1903.05704 |
https://bioportal.bioontology.org
Types of Navigation (navitype) | |
DE: details tab | DC: direct click |
DU: direct URL | EX: expand (children) |
EL: external link | ES: external search |
LS: local search | ALL: all types |
click
search
details
expand
RW: Random Walker (alpha) | MC: Markov Chain | PA: Preferential Attachment | Gr: Gravitational.