jukaradayi / link_stream_benchmark

Benchmark

BENCHMARK PIPELINE

Documentation

A documentation can be found in docs/build/html/index.html. Additionnal documentation for the package used can be found in the packages root/docs.

Pipeline

The pipeline can be configured using the benchmark.conf file. The configuration will be pasted in the output folder (for reproducibility). You can then call

bash benchmar_pipeline.sh

In benchmark.conf, you can define which dataset you want to process, and the parameters needed for each dataset.

The tools needed to preprocess each dataset should be set in utils (#TODO define as submodules ??)

Mawi

In utils/mawi_stream, find the tcpdump command to read the pcap initial files, and an awk script that can be used to only consider edges when the exchange was reciproc

Taxi

in utils/taxi_stream, find a python package to process the raw taxi data

Model

in utils/model_generation, find a python package to generate graph and timeseries from different models, defined by the modelGeneration.yaml file.