User Manual
Contents
1. User Manual to InfluenceModels SimPath version Information e Paper SimPath An Efficient Algorithm for Influence Maximization under Linear Threshold Model Amit Goyal Wei Lu Laks V S Lakshmanan ICDM 2011 e Contact author for this package Wei Lu welu cs ubc ca e Should you have questions about this paper you can also reach Amit Goyal at goyal cs ubc ca Terms of Use Redistributions and use of them in source or binary forms with or without modification are permitted provideded that the following conditions are met e For academic purposes proper acknowledgements must be made and the packages are used as freeware for which the authors do not hold any responsibility e For non academic commercial or other profit involving purposes a special case by case permission of the authors must be taken in advance For the complete terms of release please read http www cs ubc ca welu downloads html Running the Package To compile under Linux simply type S make To run the executable just type InfluenceModels c lt config file txt gt means other options can be used Parameters in Configuration Files or Command Line In the configuration file one can specify various parameters like input file for friendship graph propagation log propagation model etc These options can also be specified on the command line If a parameter is present in both command line and configuration file2. ath and SPS CELF only the cut off threshold for the backtrack method in SimPath and SPS CELF top1 for SimPath only the number of items specified for the Look Ahead Optimization in SimPath tol ldag for LDAG only the parameter for controlling the size of LDAG The default value is 0 003125 1 320
3. the command line has the preference The Phase Parameter One of the most important parameters is phase Depending on its value different modules algorithms of the code are invoked SimPath and SPS CELF both use phase 17 For SPS CELF you shall additionally set the parameter celfPlus to be 1 If it s 0 SimPath will be executed LDAG by Wei Chen el al in ICDM 2010 uses phase 15 Baseline algorithms including High Degree and PageRank use phase 20 For High Degree set the parameter propModel to be HighDeg while for PageRank set propModel to PageRank The basic Greedy Algorithm with Monte Carlo simulations use phase 10 Other Parameters You will also need to use some or all of the following parameters Please be advised that the program might encounter segmentation faults if the needed configuration parameters are missing from either the configuration file or the command line input probGraphFile the path of the input graph file budget the number of seeds to select The default value is 50 mcruns the number of Monte Carlo simulations The default value is 10000 outdir the path of the directory storing output files Consider to use separated folders for different kinds of experiments propModel the influence propagation model that will be used by the program IC represents Independent Cascade LT is Linear Thresholds and LTNew means SimPath or SPS CELF cutoff For SimP
Download Pdf Manuals
Related Search