How to import facebook users data into NodeXL??& facebook Crawler problem..

Sep 30, 2010 at 9:40 AM

Since I am conducting a research about the effect of the social network influence to the business behaviour. May I ask how to import all friend's data from facebook user's network? I am using NodeXL version 1.0.1.132. Thank you very much.

Coordinator
Sep 30, 2010 at 12:37 PM

Hello!

Dr. Bernie Hogan at the Oxford Internet Institute has created a tool for extracting a social network data set from a user's Facebook account.  Details about this tool are available at:

http://www.connectedaction.net/2010/04/25/bernie-hogans-facebook-social-network-data-provider-and-visualization-toolkit/

The tool requires each user to login to their Facebook account in order for the tool to extract the "ego-network" which is the downloaded in the "GraphML" file format.  NodeXL imports GraphML and can then visualize and analyze the ego-network.

If you collect multiple networks (with each user's permission, of course) they can then be merged and contrasted.

I hope this addresses your question.

 

Regards,

Marc

 

Oct 7, 2010 at 7:40 AM
Edited Oct 7, 2010 at 7:41 AM

Thanks for your help!

I got another problem. We are deciding to import multi-level name list from the facebook network using Namegenweb. That means my friend's friend, so here is an example of the second level of the social network (because there is no way to upload the image, plz check it by the link )http://www.douban.com/photos/photo/658609159/ .Since the Namegenweb application can only importing name list in the first level of the social network, it cannot importing the second level or multi-level, So is there any method that we can importing multi-level name list?

Other then that, we are interested to know that whether the NodeXL is capable for further coding? That means is there have the function for us to program another algorithm for further analyzing process.  Thanks for your kindly reply.

Best regards

pherson



Oct 7, 2010 at 7:09 PM

Pherson:

I'm not sure I understand your first question.  Are you asking about Bernie Hogan's work, Namegenweb, or NodeXL?

On the second question, the source code for NodeXL is available on the Source Code tab on the NodeXL site (http://nodexl.codeplex.com/SourceControl/list/changesets).  You are free to modify the source code and rebuild it for your particular needs.  If you are interested only in developing a "plug-in" for NodeXL that imports data from some custom data source into NodeXL, then that is much simpler: please see "For Programmers: About NodeXL Plug-Ins" at http://nodexl.codeplex.com/Thread/View.aspx?ThreadId=71182.

-- Tony

Oct 13, 2010 at 7:34 AM

tcap479

actually I have downloaded my friend list by using  Namegenweb.

Now, I want to do multi-level analysis, which means that if it is possible I can get my friends' friends list. For instance, I have a friend named Gary, whether I can get Gary's friend called Kate who is not in my friend list? Hope you can understand what I am saying : )

But I just can get the list shows all Gary and my friends at this time. All of those friends we are know for each other.  It is just one-level.

 

anyway, I found a software called Egotistics http://individual.utoronto.ca/berniehogan/software.html  It says "This dataset can then be used for multi-level analysis, or just used in your favorite stats package. It imports graphML, raw text files and pajek files."

I downloaded the Egotistics and run it. But I don't know why I cannot import txt file which I downloaded from Namegenweb. but the txt file works in NodeXl.

 

thanks

pherson

 

Oct 13, 2010 at 2:44 PM

Hi Pherson,

There are several issues here that you will need to contend with, and I am not optimistic about a solution. The first is data management through Egotistics, and the second, more pernicious issue involves statistical dependences.

1. Egotistics: The files should be in a single folder and all named as  .graphml Beyond that are internal issues with parsing the files, and I'm not convinced that the Egotistics parser is robust. What it spits out are an alter table and an ego table. These are the data structures that you would need to work with HLM, although, in order to work with other multilevel packages (such as GLLAMM on Stata) you would need to first do a one to many join on the two data sets, so that they are all together in a single table. What Egotistics does is calculate the network statistics that would not be available in these other packages (such as centrality, centralization etc...), and makes them in a table that -readies- a series of single network files for processing as a single table. To note, Egotistics is no longer under active development, as the principal developer has moved on to a consulting job.

2.  Network dependencies in multi-level analysis: Multilevel analysis is a regression-based technique and requires some notion of independence between the cases. This can be accomodated in a sloppy, but often sufficient way by including structural metrics in a model, such as degree and clustering, for each of the level one nodes. However, if  you have the same level one nodes in two different networks (such as having two cases who both have many of the same friends), this will violate this independence assumption in a severe way and give over-estimates of many network phenomena, simply because the alters will be the same in several ego networks. Thus, I would not use this technique with your current research design (which is a respondent driven sample by the looks of things). What I would do is to create a network of the connections between all the people in the sample, and then run it through an exponential random graph model, such as a Statnet for R, or pnet. This will report on the extent to which the network has greater triangles / edge-triangles, alternating k-triangles and so forth. But this really depends on your research question (which was not specified above, if I read correctly).

3. Finally, the image that you uploaded - where did you get that? It looks STRIKINGLY like my own network. I can see several tell-tale sins that look topologically like my own. But I don't recall rendering it in that way. Can you please confirm where  you got the data for that image?

BERNiE

Oct 20, 2010 at 10:51 AM
Edited Oct 20, 2010 at 11:27 AM

Thank you very much!

I just got the image in the pdf file of a book named "Analyzing Social Media Networks with NodeXL".

Few days ago I found a software called Facebook Crawler. I tried it to download my friends' list. 

There are two version of Facebook Crawler. I tried version1. I can log in but I faced the problem which showed that 
" it connected to facebook and generated the: friends_contact_information.csv, but all what that file had was: "

Name,Networks,Sex,Birthday,Hometown,Relationship Status,Mobile,Other,Postal Code,Current Home Town,Address,AIM,Google Talk,Windows Live,Skype,Yahoo,Website,Email #1,Email #2,Email #3,Email #4,Email #5, 

I tried version2 as well, but it was failed to log in , and it showed:

Warning: stream_socket_client() [function.stream-socket-client]: unable to connect to ssl://login.facebook.com:443 (Unable to find the socket transport "ssl" - did you forget to enable it when you configured PHP?) in C:\AppServ\www\version2\php5Extractor.php on line 34

Notice: Connection Error : Unable to find the socket transport "ssl" - did you forget to enable it when you configured PHP? in C:\AppServ\www\version2\php5Extractor.php on line 36
login failed.

facebook changed the port or else?  Does anyone has any idea about it?

thanks!!

pherson

Coordinator
Oct 20, 2010 at 1:53 PM

Hello!

I am not familiar with Facebook Crawler and probably will not be uch help getting it to work.  Does it generate data that is different from the Namegen tool from Dr. Bernie Hogan?

See: http://www.connectedaction.net/2010/04/25/bernie-hogans-facebook-social-network-data-provider-and-visualization-toolkit/

We have had good results using this tool and will be able to support your use of it better than other unrelated tools.

Regards,

 

Marc