GitXplorerGitXplorer
j

paper-evolving-graphs1

public
2 stars
0 forks
0 issues

Commits

List of commits on branch master.
Unverified
15f2f7277f11c87cb3e472a2c5d564721a60de0e

Update Makefile for arXiv submission

jjiahao committed 9 years ago
Unverified
8f104ee96009fea3219772b41adbf28b52aa43fb

various minor fixes

committed 9 years ago
Unverified
cda2df0a57820b2d5c58d15fffd725a3e93ca4cd

Describe Julia machine better

jjiahao committed 9 years ago
Unverified
73698fff011f490d54a6cf2440bb4eb6643b1e06

clarify causal edges in the experiment

committed 9 years ago
Unverified
ff82d0dad653c771ba2fd8a9ab13c896bf0c9f03

Fix typo: ocirc -> odot

jjiahao committed 9 years ago
Unverified
28a0ae040bc13215aa6309cd69fe64363f01873c

Final round of edits

jjiahao committed 9 years ago

README

The README file for this repository.

The right way to search evolving graphs

Jiahao Chen (@jiahao), MIT CSAIL

Weijian Zhang (@weijianzhang), School of Mathematics, University of Manchester

Abstract: Evolving graphs arise in many different contexts where the interrelations between data elements change over time. We present a breadth first search (BFS) algorithm for evolving graphs that can track (active) nodes correctly. Using simple examples, we show that naïve matrix-matrix multiplication on time-dependent adjacency matrices miscounts the number of temporal paths. By mapping an evolving graph to an adjacency matrix of the equivalent static graph, we prove the properties of the BFS algorithm using the properties of the adjacency matrix. Finally, demonstrate how the BFS over evolving graphs can be applied to mining citation networks.

How to compile this manuscript

You will need a LaTeX distribution such as TeXLive that has the latexmk utility.

Simply clone this git repository and run

    make

The result is an output directory containing the compiled PDF as output/paper.pdf.

Related code

EvolvingGraphs.jl is a Julia package containing an efficient implementation for the breadth-first search on an evolving graph.