Re: [igraph] Reverse engineering affiliation networks

[Ahmed Abdeen Hamed]

Sorry about the nonsense sentence in my previous email. I meant to say that Case#2 represents a scenario that will not make my inquiry possible as indicated.

-Ahmed

On Fri, Jun 5, 2015 at 12:14 PM, Ahmed Abdeen Hamed <address@hidden> wrote:

Thank you for this response!

My purpose of finding the original link to the event was to hierarchically cluster maximal cliques. I see the case two different case scenarios that will not make this possible. If I am interested in all events attended by all members such that

In Case #2:  (A,B) --> E1 and (A,C) --> E2

What is the computational cost for this step?

-Ahmed

On Thu, Jun 4, 2015 at 5:51 PM, Tamas Nepusz <address@hidden> wrote:

Hi,

This is not possible in general as it is easy to construct two networks which
result in the same affiliation network representation -- even if you use
weights to indicate the number of events attended jointly. For example, with
three people (A, B and C):

Case #1: A, B and C attended event E1.
Case #2: A and B attended event E1, A and C attended event E2, B and C attended
event E3.

In both cases, the affiliation network will be a triangle.

T.

On 06/04, Ahmed Abdeen Hamed wrote:
> Hello friends,
>
> Say your original network was constructed from people and events. Now, an
> affiliation network is derived based on the notion of people who are going
> to the same event. If you have computed cliques from this affiliation
> network, can you link this clique back to the original event being
> attended? Any examples would be appreciated.
>
> Sincerely,
>
> -Ahmed

> _______________________________________________
> igraph-help mailing list
> address@hidden
> https://lists.nongnu.org/mailman/listinfo/igraph-help


--
T.