If I have two tables, each with verbose text fields of variable word count from few to many, between which I want confirm the existence and extent of a partial word match in one table with records in the other, how can one possibly approach this in an elegant manner? Ideally, each record in the parent table would show, via a portal, a list of partially matching (since it is unlikely to have an exact match as the result) records in the child table, sorted, of course, by decreasing frequency of matching words. The higher the word/partial word match count, the higher the match rating. Ideally, one would want to exclude common words such as definite/indefinite articles, pronouns, conjunctions, etc., presumably by performing a substitute function first. The number of records and the number of word options is just too large to manually define keywords. I could imagine substituting, then going word by word through an iterative process but this is clunky and would not recognise variations in words, for example 'gold' versus 'golden'. Is there perhaps a mathematical approach to this? I guess this borders on fuzzy logic or machine learning.