If you do a COLLOCATES search and you don't specify anything for the context (click here for an example), then a minimum Mutual Information score limit is imposed on the results. Without these limits, the most common collocates will be words like the, of, by -- high frequency noise words that you probably don't want and which don't provide much insight into the meaning of the word. But by forcing the Mutual Information score to something like [3.0], these high frequency noise words are eliminated. You can reset this limit higher or lower, if you would like.
|