How does e-FAQ decide if a FAQ answer match is good enough to automatically generate a response?
e-FAQ ranks matches based upon a linguistic matching calculation. e-FAQ gives each question/answer pair a score based upon how many of the query’s words (or their synonyms) were matched. e-FAQ then performs a statistical calculation that is loosely modeled after a bell curve grading scheme. Those question/answer pairs that indicate a significantly higher matching value are given a higher score.