What his answer was getting at was that you need to separate the numerical part of the name and put that into a number field. You could then sort on that field. So "101 salem cert tax iv" could be split so that you have just the digits 101: regex would be something like ^\d+. Then you can put that value into a separate number field.

You could then sort on that number field instead of on the file name.

The workflow could be as simple as

1. Search Repository to find the documents you are looking for

2. for each of the documents the above search finds

3. pattern match %(ForEachEntry_CurrentEntry_Name) using ^\d+ in the pattern input

4. Assign the patttern match token to a number field that you have created.