In term extraction, the automatic term extractor module computes a number for each word sequence it encounters. This is a complex number that is calculated from the length, the frequency, and many other properties of each word sequence.
This number is called the candidate score.
If the score for a word sequence is greater than a pre-defined threshold (that is, reference value), the automatic term extractor considers it a term candidate, and adds it to the candidate list.
In memoQ, the candidate list editor sorts the candidates in a descending order by the score, so the more relevant candidates are supposed to occur at the beginning of the list.