filtering concgrams

In order to select which items are "associated", we need some sort of suitable statistical procedures. The members of each concgram are at present merely associated by co-occurring at least a certain number of times as explained in generating them

 

The Filtering settings in the Controller allow you to specify, for example, that you want to see only those which are associated with a MI (mutual information) score of 2.0 or a Log Likelihood score of 3.0.

 

concgram_filtering

Ensure the statistics you need are checked and set to suitable thresholds, and decide whether all the thresholds have to be met (in the case above both MI and Log Likelihood would have to score 3.0 at least) or any of them (in the case above MI at 3.0 or above or Log Likelihood at 3.0 or above). You can also optionally insist on certain words being in your filtered results.

 

When you press the filter button (filter-button), you will see something like this:

 

concgram_filtered

where the items which meet the filter requirements are separated out and selected ready for concordancing; any others are hidden. To the right you see that the head-word CAESAR here relates to AND HE, HER, I , ANTONY etc. above the thresholds set.

Click the Permalink button if you want to copy a link to this page.