Home | You are at

Frequency of Updates to Articles


Written by Tim Brody, last updated on August 31 2000 09:09:17.

How often are papers updated after their initial submission?

Data

Using the "date of submission" meta data in the abstract files, a data set can be built containing a list of dated updates to individual papers.

A directory listing was taken of the entire Los Alamos Archive. This was filtered into only abstracts (.abs files). A script, papers2updates parsed this directory listing into another data set containing one submission ("update") per line, with the date of submission as the first entry on the line (as obtained from reading the abstracts).

19951219 abs astro-ph/9512121 199512 ftp/astro-ph/papers/9512/9512121.abs
19980616 abs astro-ph/9512121 199512 ftp/astro-ph/papers/9512/9512121.abs
  • Date of submission
  • Type of submission
  • Page reference
  • Date of first submission (create date)
  • File location

To generate a frequency of updates this was parsed by a script called "updates2freq", which added together all updates that occurred on the same time difference from the initial submission. For example a paper "a" was updated 2 days and again 5 days after initial submission. A paper "b" was updated 1 day and again 5 days after initial submission. This would generate a frequency graph of:

1 2 5
1 1 2

This was performed for the entire data set to produce a frequency table that could be plotted.

0 422
1 3832
2 1738
3 1328
4 1014
5 863
6 882
...

This graph is based on the time differences between the initial submission and subsequent submissions.

The following graphs show the same data, with initial periods of submissions removed.





Home