1. Right, it depends on exactly that, the direction of the edges. At the
moment, my query returns influencer: influencee, so that clearly means if
the edges run towards the first column in the table, which is influencer.
Am I right?
2. I know we can't do PR from counts. I didn't mean that at all. You are
right, that factor does mean a lot (Aristotle influenced Marx), and there
can be even more complicated analysis done, but at this stage, I am just
getting the feel for the data. Thanks to you, new ideas like PR have popped
up which I aim to utilize through Gephi. Moreover, how since you mentioned
Freebase, I'd like to ask you if you know how can I use the influence
node<http://www.freebase.com/influence/influence_node> to
do the same query that I am doing here on DBPedia.
I gave up because I couldn't get it to work and moreover it returns JSON so
I thought I will have problems doing a UNION of DBPedia with Freebase
results. If you can give me some insight into this, I'd be able to perform
my analysis with more data. Duplication won't be an issue due to the SQL
query handling that, it will still give me some more data which is good.
Thanks Tom.
On Sat, Dec 21, 2013 at 8:09 PM, Tom Morris <tfmor...@gmail.com> wrote:
> On Sat, Dec 21, 2013 at 2:58 PM, Ali Gajani <aligaj...@gmail.com> wrote:
>
>> Many thanks for your input Tom. An in-degree is the number of incoming
>> edges towards that node. I think that captures *influencer: influencee
>> (Aristotle : Alexander, Aristotle : Myself)*, which means, in this
>> scenario, Aristotle (the node), has an in-degree of 2. I thought in-degree
>> was a measure of influence rather than an outdegree. Remember, this is
>> going to be plotted as a *directed* graph in Gephi. I'll be curious to
>> know how I'll actually distinguish in-degrees and out-degrees in Gephi
>> practicaly, but anyway.
>>
>
> It really depends on whether your relation is influencer -> influenced ->
> influencee or influenced <- hadInfluencee <- influencee (ie which way the
> directed edges in the graph run).
>
>
>> Moreover, I didn't quite get about how I could do a Page-Rank style style
>> aggregation on this specific scenario. Could you please provide some
>> examples using actual person names so I can digest it well in my head.
>> Thanks for getting my head working though, but I still believe the
>> Wikipedia data gives you a decent impression of influence to an extent,
>> albeit not the most accurate, but it kind of appears to be right in one way
>> or the other.
>>
>
> You can't do PageRank from just the counts. You need the full network of
> links. As an example, if Marx had the most direct influencees, but
> Aristotle influenced Marx, shouldn't that count for something? Perhaps
> more? BTW, Freebase actually thinks Nietzsche is first by simple count,
> not Marx, but the underlying data is so biased and incomplete for both
> Wikipedia & Freebase, that I'm not sure it's worth pursuing a more
> sophisticated weighting.
>
> Tom
>
> p.s. If you're using Gephi, it has a PageRank implementation
> http://wiki.gephi.org/index.php/PageRank
>
>
>>
>>
>> On Sat, Dec 21, 2013 at 7:49 PM, Tom Morris <tfmor...@gmail.com> wrote:
>>
>>> On Sat, Dec 21, 2013 at 2:24 PM, Ali Gajani <aligaj...@gmail.com> wrote:
>>>
>>>> ... I want to make sure I can use this dataset to count indegrees (high
>>>> influencers) properly. It is impossible to survey all the rows to ensure
>>>> the knowledge is true, but I am asking anyway.
>>>>
>>>
>>> Presumably you mean out-degree if you're talking about influencers. A
>>> simple count doesn't sound like it'll capture the real influence. Even if
>>> you assume that Wikipedia has a comprehensive and unbiased coverage of
>>> influential people (almost certainly not true), shouldn't influencing
>>> someone influential count more? That would imply you need to do a page-rank
>>> style aggregation of link weights.
>>>
>>> Tom
>>>
>>
>>
>>
>> --
>>
>>
>> Ali Gajani
>> Founder at Mr. Geek
>> www.mrgeek.me
>> www.aligajani.com
>>
>>
>
--
Ali Gajani
Founder at Mr. Geek
www.mrgeek.me
www.aligajani.com
------------------------------------------------------------------------------
Rapidly troubleshoot problems before they affect your business. Most IT
organizations don't have a clear picture of how application performance
affects their revenue. With AppDynamics, you get 100% visibility into your
Java,.NET, & PHP application. Start your 15-day FREE TRIAL of AppDynamics Pro!
http://pubads.g.doubleclick.net/gampad/clk?id=84349831&iu=/4140/ostg.clktrk
_______________________________________________
Dbpedia-discussion mailing list
Dbpedia-discussion@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/dbpedia-discussion
