Anaisa asked 5 years ago
Hi!
Sorry to bother again. I am working on a factiva corpus. I uploaded 130 news articles as my corpus. I extracted 50 terms from my corpus and then edited it in openrefine, 41 terms left. Then I index it into the corpus using corpus term indexer and a new table ISItermedited created. Then I want to produce a new network map : field 1 – ISItermedited and field 2 – source name. But in the network map extracted, only around 20 terms are listed. What happened? Is that because other 21 terms are not close enough to the source name ? I don’t understand how it works.
Thanks and have a nice day!
Lionel Staff answered 5 years ago
Dear Anaisa,
You have to fine tune:
1/ the proximity measure: https://docs.cortext.net/analysis-mapping-heterogeneous-networks/mapping-edges-definition/ by default the proximity measure is distributional (https://docs.cortext.net/metrics-definitions/)
2/ the number of nodes for each variable (Number of nodes and Field2 number of nodes in advanced option) Remember, only the linked nodes will be shown (you may want to play with Hide isolated nodes).
L
Anaisa
replied 5 years ago Dear Lionel,
Thanks for this answer! So I set the proximity measure to chi2 ( I thought that for hetegemous map automatically is chi2) and in the advanced option, I chose to see the hided nodes ( which are not linked) and set field 2 to 50 ( field 2 is my terms)/ But altogher, the linked and unlinked nodes are not 40 in total.. I don’t know why.
And why are the terms aren’t linked ? Because it is source name and terms extracted from articles, I suppose there is a link between these two?
Thanks !
Anaisa
LionelIt is due to Chi2, I think.
Play with RAW instead of Chi2 (and Hide isolated nodes on No), and you should have everything (but without any measure on proximity).
Anaisa
replied 5 years ago The terms are still not completed.. I don’t know why. But why are those terms disappeared? Could I explain as they are not proximate enough to source name? And in RAW the map is not really good as very little terms are linked.Thanks!!
If all the terms do not appear with the other proximity measures, it means that they do not reach the threshold which is automatically calculated. You may want also to play with Proximity Threshold in advance option for edges, put no, and define your own Threshold to have more linkages.
Anaisa
replied 5 years ago I set the threshold to 0 and put no for optimitic proximity threshold. There are always just around 30 terms appeared. Then I run a corpus explorer to see the PC-Soucename-ITerm, headline and fulltext, and then I found for example, institut pasteur&le soleil for several articles in this first column of PC-Soucename-ITerm where in the fulltext it doesn’t include this term institut pasteur… I don’t know if it is normal?
Mille merci!!
Anaisa
replied 5 years ago Dear Lionel,
I think I have figured out the problem. Actually I guess there are some problems during the term index. I used openrefine before but now I changed to Openoffice and it seems that it all works!
Thanks for your patience and all the help! And thanks for offering us this awesome platform!
Bonne soirée !
Anaisa
See below !