Complex Edge question

[ina.karova]

Hello,

I am new here, and I scrolled along the website and the forum to check and I think there has not been a [Gephi] problem as mine.
So, here is my question:

I am making an experiment - I ask people to fill-in a report schedule with their personality traits related to their interaction with their parents, close friends, beloved one, and a stranger. Each trait is then categorised by the person as important, not important, completely unimportant, and noted as positive, negative or neutral. As a researcher I want to find out which features cluster most often with other features, i.e. "quiet with parents" clusters with "noisy with friends" or "nice with stranger" clusters with "rude with close friends".

So, I make a Node file for Gephi consisting of the following columns: ID (unique!!!), Label (name of the personality trait i.e. quiet), Report schedule (it has its own ID), Age, Gender, Importance, Relationship, Positivity.

I have troubles with the Edge file: I do not know how to define the Source vs Target so that I get a cluster of traits most commonly clustering with each other. I feel the Source should be the ID of the trait and then the Target - the Report ID. Each trait ID is a unique number 1-549. Each report ID is 1-44 number. Still, Gephi gets confused with it - if Gephi sees that the trait participates in reports N88,59,13, Gephi clusters it with Trait IDs 88, 59, 13, and not with other traits having Report ID 88,59,13.

NOTE: The Report IDs can be met multiple times - i.e. one trait (but with different attributes) can be reported a few times in a single report.

Could you please help me figure out how to arrange the Edge file so that I get clusters of traits most commonly clustering together across various reports? I will be very thankful for clues on this, because I have tried to do it for a week now; and even though it is just a pilot study, in future I will have to deal with a far larger database for my dissertation and I wish to be sure that the definition of Source vs Target is correct ... and so is my research

[ina.karova]

Just to let you know I resolved it.
A friend of mine wrote an algorithm that mathes the IDs of the traits based on the Report ID to which they belong. It is a good solution. The disadvantage is that if all ID attributes are the same, the only one that always differs is the Report ID, so some of the IDs, after filtering based on their attributes, still do not get connected and form individual clusters.
I hope my problem and its solution helps other people using Gephi.