||when in comparing two records there are values in corresponding
fields and they are not the same, i. e., the fields disagree, yet there is a relationship between the values, which may
indicate the original intent of the values are the same, then they may be said to be in p. a. As a particular relationship may carry with it a characteristic likelihood
that the intent is the same, one may want to assume it proportional to the likelihood of a link and adjust its weight accordingly.
Normally there is a new field populated to carry the standard value of the identifier that disagrees. Presumably it is possible
to assign to each variant a degree of dependence that the standard has to it.
(Cf. agree, disagree, similarity function). (Cf. Newcombe, 1988, pp. 19-20)
[Cf. RLG Chapter 2, §1, ¶4 & RLG Chapter 5, §3, ¶6] |