Don't get Partition to work

[brode]

Hello everybody,

I'm very satisfied until now with gephi, i like to visualize relationsships between people in economic context.
But I have one big problem.

I load this here for example:

Code: Select all

<?xml version="1.0" encoding="UTF-8"?>
<gexf xmlns:viz="http:///www.gexf.net/1.1draft/viz" version="1.1" xmlns="http://www.gexf.net/1.1draft">
    <meta lastmodifieddate="2010-03-03+23:44">
        <creator>Gephi 0.7</creator>
        <description>Graphendatei</description>
        </meta>
    <graph defaultedgetype="undirected" idtype="string" type="static">
        <attributes class="node">
            <attribute id="0" title="country" type="string"/>
            <attribute id="1" title="gmf" type="boolean"/>
        </attributes>
        <attributes class="edge">
            <attribute id="0" title="typ" type="integer"/>
        </attributes>
        <nodes count="3">
            <node id="0" label="Hans Mueller">
                <attvalues>
                    <attvalue for="0" value="BRD"/>
                    <attvalue for="1" value="true"/>
                </attvalues>
            </node>
            <node id="1" label="Peter Pandisky">
                <attvalues>
                    <attvalue for="0" value="BRD"/>
                    <attvalue for="1" value="true"/>
                </attvalues>
            </node>
            <node id="2" label="Hans Wurst">
                <attvalues>
                    <attvalue for="0" value="DDR"/>
                    <attvalue for="1" value="false"/>
                </attvalues>
            </node>
        </nodes>
        <edges count="2">
            <edge id="0" label="Kante zwischen " source="0" target="2" weight="13">
                <attvalues>
                    <attvalue for="0" value="3"/>
                </attvalues>
            </edge>
            <edge id="1" label="Kante zwischen " source="0" target="1" weight="3">
                <attvalues>
                    <attvalue for="0" value="2"/>
                </attvalues>
            </edge>
        </edges>
    </graph>
</gexf>

The problem now is I don't get any variables when I click on refresh/nodes/partition. I want to partition variable country for example. But I see them in the Data Laboratory. They are imported, is there any special datatype I have to use??

Maybe you can give me some hints in my general problem. I want to visualize people and different Kind of relationsships between them. So I need also different dimensions for the edges. (for example (1) organisational like working in the same department, (2) emotional, two people like to work together (which has nothing to do with the organisational structure of course, (3) knowledge transfer ) and so on. I want to visualize different aspects of the relationsships, to get an overview. So the edge between to people need Attributes (1) (2) (3). Any suggestions to solve this problem? The distance between the nodes should be the organisational structure for example. But the thickness of the edges should mean something like (2), maybe the colour of the edges something like (3).

Regards,
Kai

[mbastian]

The problem now is I don't get any variables when I click on refresh/nodes/partition. I want to partition variable country for example. But I see them in the Data Laboratory. They are imported, is there any special datatype I have to use??

Hi, there is a bit tricky mechanism to make the difference between partitions and id/results columns. The reasons why it is not showing up is because the number of partitions is superior than 2/3 of the non-null values. Because you have too many differents parts compared to the elements it is not considered as a partition. The mechanism is probably too restrictive and the worst is that it cannot be configured/bypassed for now. We should improve that.

Maybe you can give me some hints in my general problem. I want to visualize people and different Kind of relationsships between them. So I need also different dimensions for the edges. (for example (1) organisational like working in the same department, (2) emotional, two people like to work together (which has nothing to do with the organisational structure of course, (3) knowledge transfer ) and so on. I want to visualize different aspects of the relationsships, to get an overview. So the edge between to people need Attributes (1) (2) (3). Any suggestions to solve this problem? The distance between the nodes should be the organisational structure for example. But the thickness of the edges should mean something like (2), maybe the colour of the edges something like (3).

If I get it correctly you have several relationship types, non exclusive. Because parallel edges are not supported in Gephi you need to find a way to set your data as attributes. I would create first a 'type' column to make Partition work with "department", "emotional", "knowledge", "department & knowledge", "emotional & knowledge", ... By doing this each edge would have a single type and allow Partition to color you edges nicely. Then you could create BOOLEAN columns foe each of your types and set 'true' or 'false' for each edge. That allow to use filters and show/hide particular relationships.
If you put the edges weight, the ForceAtlas layout would use it but maybe you need a different layout for doing that.

What I learned from your use case:
- Bypass the partition limit somehow
- Would be nice to be able to deduce automatically the 'type' column from the three boolean columns
- Would be nice to be able to set the edges weight column, to be able to try different