Difference in metrics results

[pavlina....@gmail.com]

Hi all,

I was wondering if you have any idea of why there could be a difference between the results for e.g., closeness centrality between Gephi and ORA? Specifically, it seems that the nodes that have the lowest centrality in ORA, have a centrality of 1 in Gephi.  All other nodes have similar closeness centrality results from both tools. So I find this strange.

Would be happy if someone has an idea of why that is.

Thx and BR

Pavlina 

[Kathleen Carley]

chance are you have weighted data or multiple components

you need to be very careful - see this report

Wei Wei, Jürgen Pfeffer, Jeffrey Reminga and Kathleen M. Carley, 2011, “Handling Weighted, Asymmetric, Self-Looped and Disconnected Networks in ORA,” Carnegie Mellon University, School of Computer Science, Institute for Software Research, Technical Report, CMU-ISR-11-113.

--
Computational Analysis of Social and Organizational Systems (CASOS)
The Institute for Software Research, Department of Computer Science
Carnegie Mellon University, Pittsburgh, PA, USA
http://casos.cs.cmu.edu/
---
You received this message because you are subscribed to the Google Groups "ORA Google Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ORA-google-gro...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/ORA-google-group/f2efbafd-eeec-4e1f-83a2-16fe4a9a77ebn%40googlegroups.com.

[pavlina....@gmail.com]

Thank you!

[Judeth Choi]

Hi there!

I’m using ORA-MAC. I have a meta-network from Twitter data. I want to convert it to a dynamic meta- network, but when I do I just end up with all of the data in 1 meta-network—not broken down by date/time (I want to break it down by day).

Did I miss a step?

…Or is there a workaround given that I already have an xml file (i.e. I can’t use the import wizard to re-import with DNA).

Thank you,

judy

[Jeff Reminga]

Hi Judy,

There is currently no routine to convert a meta-network into multiple meta-networks to create a dynamic meta-network.

For most general-purpose datasets such a conversion would not be possible, but with Twitter data imported by ORA the routine is possible but has not as of yet been added to the interface.

We will post a notice when we add the routine.

Thanks,

Jeff

--
Computational Analysis of Social and Organizational Systems (CASOS)
The Institute for Software Research, Department of Computer Science
Carnegie Mellon University, Pittsburgh, PA, USA
http://casos.cs.cmu.edu/
---
You received this message because you are subscribed to the Google Groups "ORA Google Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ORA-google-gro...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/ORA-google-group/12DD39CE-46FB-4CA7-B4A3-5A9FEF8E3BBC%40andrew.cmu.edu.

[Kathleen Carley]

In the mean time you can segment the json by e.g. week an import each separately and then combine those as a dynamic meta-network,

To view this discussion on the web visit https://groups.google.com/d/msgid/ORA-google-group/CACr%2B-gQ7HJ_QrL0SPSH3Nm%3Dff9DKWRrxpphdj7FxctwF2zXK8g%40mail.gmail.com.

[Judeth Choi]

Thank you!

I filtered some of the data and combined meta-networks in ORA, so I have the xml file(s). It is probably easiest for me to add an attribute categorizing data by week and then split it into groups, no?

Also, are there any rough estimates of the processing power needed to run normal measures on x amount of data (GBs?, # of tweets?). (I can also consult with researchers off of this forum if easier.)

With gratitude,

judy

To view this discussion on the web visit https://groups.google.com/d/msgid/ORA-google-group/CAFtd-nbazW%3DuKeAydbpEpJOHY1ND5rJMEJjKhMWGCgh%2B02%3DeNQ%40mail.gmail.com.

[Jeff Reminga]

Hi Judy,

Are you able to re-import from JSON files? If so, then you can easily import the data into a dynamic meta-network with a time-level of aggregation of your choosing.

Splitting data after import is much more difficult, although combining data is easy. And so we recommend importing data with a small time aggregation.

As for the time and space requirements of measures, the main interface menu time Analysis \ Measures Manager contains a list of all the measures and a "Computation Time" column that labels measures as Fast, Normal, or Slow. Fast measures require very little additional memory; Normal and Slow measures sometimes need to duplicate the dataset in memory.

If you are running on large datasets, it is typical to disable Slow measures, as the slowest of them (such as Betweenness Centrality) will take a substantial amount of time.

This can be done while running a report in this manner:

There are other tools in ORA to trim a dataset, such as by component, k-core, or grouping algorithm, such as those available from the main menu:

Data Management \ Meta-Network Groups

Thanks,

Jeff

To view this discussion on the web visit https://groups.google.com/d/msgid/ORA-google-group/FEDE4C73-1C8D-4C95-9AB9-019CC9191330%40andrew.cmu.edu.