10–11:	A threshold of 0.5 is applied to the similarity values for each input header to find the candidate output headers and the candidates are sorted by their descending similarity values.
12–14:	For each input header, our algorithm finds the output header with the highest similarity value.
15–17:	If the output header is not used for any previous match, the match between the current input header and output header is taken while storing its similarity value.
18–20:	If the output header has already been used for a previous input header (prevMatch in the algorithm), their similarity values are compared.
21–24:	If the current input header has a higher similarity value, its match is accepted while discarding the match of previous input header. Then a match for this previous input header is found in a recursive way.
12–14:	If the current input header has a lower or equal similarity value, it is matched with the next candidate output header following the same rules (continuation of the for loop in the algorithm).