This is a replication by Justin Tomlinson of "Building Reputation in StackOverflow: An Empirical Investigation" by A. Bosu, C. Corley, D. Heaton, D. Chatterji, J. Carver, and N. Kraft at the University of Alabama.
This project relies on a large series of queries into a postgresql database supplied by the MSR conference from August 2012 of the stack overflow data dump. Due to the large file size, that will not be in my GitHub repository, but it can be found here:
http://2013.msrconf.org/challenge.php#challenge_data
The project also relies on an open-source network exploration and manipulation software called Gephi, which is available here:
http://gephi.org/
Directories:
queries/:
This directory contains all of the script files that can be used with
a postgresql database of the datadump. The output is not necessarily
obvious though, so be sure to read the comments of what each query is
doing.
results/:
This directory contains the results from the queries. These are not
automated, they are written by hand from the results of the queries,
and some additional calculations are added to give more relevant
results, such as certain ratios and other metrics.
Files:
metricsquery.sql
This file is a postgresql script file that will query the data dump and
get the necessary values to calculate the different metrics.
metrics.txt
This file shows the results of the metric queries and the calculated
metrics.
categoryquery.sql
Using the results of the Gephi implementation, each category (made up
of certain tags) is queried for different metrics. Since there are 122
tags, I have omitted all of the ones not listed under "Top Areas"in
the chart from the research paper.
category.txt
This file shows the results of the metrics applied to each of the
Gephi categories. There 14 categories which are listed at the end
of each metric.
expertqueries.sql
This shows the queries used to get information on the "experts" of
stackoverflow.
experts.txt
This shows the results of queries and hand-calculations for the
experts category.
dowqueries.sql
This file contains the queries used to get information about
individual days of the week's activity from the database.
dow.txt
These are the results from the query and hand calculations made
from the results of the query. Each result label is followed by a
character (U,M,T,W,R,F,S). These represent the days of the week
beginning with Sunday and ending with Saturday.
intervalqueries.sql
This file is using the StackOverflow online query to get the results
since the postGRESql database did not use dates as a timestamp, so I
could not get a specific time. Unfortunately, stack overflow only
allows 50000 results to be returned, so the median interval is not
the same as the research paper.
intervalSS.csv
This is the file that was used in Excel to find the difference
between the two times and the median of them all.
intervalAASS.csv
This is the same as intervalSS, but for the accepted answer interval
and not the first answer interval.
interval.txt
This is the results of the queries once they were imported to
Microsoft Excel and the difference and median was taken.
fcqueries.sql
These contain queries used on the 10 fastest contributors to reach
a score of 20000 to get different metrics.
fc.txt
These are the results of the queries on the ten fastest users. The
first ten results listed are the user id (UID) of each user. Then
following results are labeled in the same order of users.
FinalReport.pdf
This pdf is the final report of my findings comparing my results to
the paper's, and drawing conclusion from it.