[SOLVED] Difference in performance for nbr of spells?

All questions about the GEXF (see http://gexf.net before)
Post Reply
User avatar
seinecle
Gephi Community Support
Posts: 546
Joined: 08 Feb 2010 16:55
Location: Lyon, France
Contact:

[SOLVED] Difference in performance for nbr of spells?

Post by seinecle » 22 Jul 2011 15:13

Hi,

I have a script which generates spells for every time period, even when they are consecutive:

VERSION 1
<spell start = "2005.0" end="2005.0">
<spell start = "2005.1" end="2005.1">
<spell start = "2005.2" end="2005.2">
<spell start = "2005.3" end="2005.3">
<spell start = "2005.8" end="2005.8">

Obviously it would save space to have instead:

VERSION 2
<spell start = "2005.0" end="2005.3">
<spell start = "2005.8" end="2005.8">

My question is, is VERSION 2 leading to better computational performances or better memory management than VERSION 1 when I export the gexf in GEPHI? (I'll have 100,000 spells or more). Is it worth writing a script to convert my gexf to VERSION 2?

Thx!

Clement

admin
Gephi Community Manager
Posts: 964
Joined: 09 Dec 2009 14:41

Re: Difference in performance for nbr of spells?

Post by admin » 24 Jul 2011 22:23

Do you have continuous or discrete time?

Here it is not an issue of how Gephi behaves but pure data consistency. The first version says that, for instance, the element DOES NOT EXISTS on the continuous period between 2005.0 and 2005.1.

I'm not sure if this what you want.

User avatar
seinecle
Gephi Community Support
Posts: 546
Joined: 08 Feb 2010 16:55
Location: Lyon, France
Contact:

Re: Difference in performance for nbr of spells?

Post by seinecle » 25 Jul 2011 07:37

Oh... I get it now! Thx.

Indeed I want continuous time, so that's version 2.

I wrote a script working in Excel which converts spells from Version 1 to Version 2, working with dates formatted like in the example. Anybody interested can drop me an email to get it.

Best,

Clement

Post Reply