2021_pbds_reproducibility

We implemented PBDS in our GProM system which is a middleware. Provenance sketch capture and use are available as extended SQL commands in GProM. We include in this repository the extended version of our paper (./techReport), raw results and scripts for plotting them, SQL code generated by GProM for capture and use queries, the GProM source code (for the branch that support PBDS), and the Python code the drives some of the end-to-end experiments. The datasets we are using are publicly available. If the paper gets accepted we will create a docker image with all datasets (and prepared databases for the open source systems we are using). We will make DB-X available on one of our servers for reproducibility.

GProM source code

Is included in folder ./gprom. Note that GProM is available as open source at https://www.github.com/IITDBGroup/GProM. We just include the source code for convenience.

Raw results and plot scripts

We include the raw experimental results and scripts for plotting here:

path: results

Python code for end-to-end

This code is included in

path: experiments/end-to-end/

SQL code for Experiments

The SQL code of queries generated by GProM for capturing and using scripts can be found in the following folders. We organize this by figures in the paper.

postgres tpch 1g and 10g (Fig.6 a,b,f,g in main paper)

path: experiments/postgres/tpch

postgres stackOverflow (Fig.6 d and e in main paper)

path: experiments/postgres/stackOverflow

monetdb tpch (Fig.6 h,i,j,k in main paper)

path: experiments/monetdb

db-x tpch 10g (fig.6 c in main paper)

path: experiments/dbx

stackOverflow end-to-end mix (fig.7 a in main paper)

path: experiments/end-to-end/stackOverflow/mix

stackOverflow end-to-end sel (0.7%, 2% and 5% in main paper)

path: experiments/end-to-end/stackOverflow/q2_3_sels

ps vs mv syn-Q1 (fig.8 a,b and c in main paper)

path: experiments/ps_vs_mv/syn-q1

ps vs mv syn-Q2 (fig.8 d in main paper)

path: experiments/ps_vs_mv/syn-q2

ps vs mv stackOverflow (fig.8 e in main paper)

path: experiments/ps_vs_mv/stackOverflow

ps vs mv end-to-end (fig.8 f,g,h,i and j in main paper)

path: experiments/ps_vs_mv/end_to_end

technical report - crimes

path: experiments/postgres/crimes

technical report - movies

path: experiments/postgres/movies

technical report - optimization

path: experiments/optimization

technical report - end-to-end crimes

path: experiments/end-to-end/crimes

technical report - end-to-end stackOverflow SDV

path: experiments/end-to-end/stackOverflow/q3_sdv