discuss using 43 Things' tags to match goals

If 43 Things uses tags to match goals, I offer the following thoughts about what the “match” list should look like.

Let’s assume that 101,234 goals and 30,824 tags exist, and 1 goal-tag-list (12 tags) matches perfectly, 1 matches all tags but not in order, 2 match on 5 tags, 2 match on 3 tags, and 4 match on 2 tags. The “match ” list might look like this:

• 101,234 goals and 30,824 tags were scanned.
• Match % # tags matched – Goal URL
• 100 …..... 12 …...... URL
• 099 …..... 12 …...... URL
• 084 …..... 05 …...... URL
• 082 …..... 05 …...... URL
• 071 …..... 03 …...... URL
• 068 …..... 03 …...... URL
• 038 …..... 02 …...... URL
• 038 …..... 02 …...... URL
• 036 …..... 02 …...... URL
• 036 …..... 02 …...... URL

(Note: The URLs are hyperlinks to the goals. Remove the leasding zeros. I had to use them to align the list.)

If 43 Things uses tags to match goals, alphabetizing the tag list will be useful for a “first pass” comparison, but the original tag list order/sequence must be preserved because it should represent the relative importance of each tag. The first tag being the most important and the last being the least important.

If Joe’s tag list says “a,b,c,d” it will perfectly match Frank’s tag list of “a,b,c,d”, and almost perfectly match Jim’s tag list of “c,b,a,d”.

Tag sequence will be important!

Links: http://apcala.com/peteh/weblog/221.html

If two people submit different goal titles which are – in fact – exactly the same goal, matching at the goal-description-level will not succeed, but will at the tag-level.

For example, Joe’s goal is to “buy a car” and he lists “Buy, 1998 Honda CRV, low miles” in the tags. Frank’s goal is to “sell his car” and he lists “Sell, 1998 Honda CRV, low miles” in his tags. The goals do not match.

Both sets of tags are identical except for the first tags of Buy and Sell. But since these tags should be recognized by 43Things as a purchase transaction, 43Things lists them as a perfect match.

Let’s try another example in my area of interest. Joe’s next goal is to “find a better job” and lists tags of “New job, Internet, design a job matching system, ZIP=anywhere, work from home, flexible hours, $50-$75/hour”. Five goal setters (job banks) list “offering a new job opportunity” and each list all of Joe’s primary job attributes – “Internet, design a job matching system” – but each only lists a different one of Joe’s remaining attributes. ... 43Things should consider each a match to a degree, but how does 43Things rank the matches? I’m sure someone with advanced math skills has a neat answer for weighting the tags based upon location of tags in the tag-list.