Despite my "a picture paints a thousand words", I'm still having trouble understanding your data.

For example, in the following two lines, only the presence of a part no for the last manufacturer distinguishes them. Doesn't that make the second a duplicate of the first with missing information? Ie. redundant?

NGK_STK.  NGK_P/N  NGK ALT  ACCEL  AC_DELCO  AUTOLITE  BECK/ARNLEY
1027      AP9FS    N/A      N/A    84TS      32        176-5178
1027      AP9FS    N/A      N/A    84TS      32        N/A
[download]

There appears to be a significant column of information missing from the above table?

I could imagine that the above data represents the recommendations by the different manufacturers for the plugs in their range that would be applicable to two difference vehicles. Say the first is the normally aspirated version of some mark, and the second is the turbo-charged version. And whilst most of the manufacturers recommend the same plug for both, the BECK/ARNLEY plug is unsuitable for the latter variant. And, they have no suitable alternative in their range.

My point is, that whilst plugs from different manufacturers may be interchangable for a given vehicle, each manufacturers plugs have different ranges of operating parameters, which means that plugs from two different manufacturers are not interchangable for all applications.

The upshot is, the reason you are having so much trouble coming up with a normalisation schema, is because the key field--the vehicle--is missing from your table. Any attempt at normalisation based upon grouping of part numbers without taking the vehicle into consideration is at best doing your customers a dis-service. It could pursuade them to purchase plugs that are unsuitable for their particular vehicle, that might have limited life due to (say) overheating. Or worse, that could damage their engines by (say) holing their pistons by burning too hot.

At worst, it could be dangerous.

For example, in the following two lines, only the presence of a part no for the last manufacturer distinguishes them. Doesn't that make the second a duplicate of the first with missing information? Ie. redundant?

Grrr. That's what I was trying to avoid when presenting this data. Remember when I said I'd be trimming off a number of the columns? That's what happened.

A lot of the data is redundant - that's just how it's organized - but not because it's for different vehicles or whatever; it's because whoever put it together didn't use an easier way to say "vendor X's plugs 123,124, and 125 match vendor A's plug 999". Instead, they end up saying

A    B    C    D    X
999  001  002  003  123
999  001  002  003  124
999  001  002  003  125
[download]

It's dumb, and I know it's one of the things I have to factor out - but I figured that I'd be taking care of lots of redundancy anyway, and this would just get taken care of as part of the parsing process.

My point is, that whilst plugs from different manufacturers may be interchangable for a given vehicle, each manufacturers plugs have different ranges of operating parameters, which means that plugs from two different manufacturers are not interchangable for all applications.

I assure you that this is not the way it works; plugs with different operating parameters get a different part number. The parameter ranges vary widely, but they're standardized - for exactly the reasons you state - and they're comprehensible as data.

I was at an auto parts store two days ago, and had them look up a plug for me - an NGK B7HS-10 - and they pulled up a screen that showed all the equivalents, which exactly matched the list that I had. They never asked me what vehicle I had, and it wouldn't have helped them if they did: it was a Yamaha 15HP outboard motor, and I no auto parts store is going to list that as a vehicle. :)


--
"Language shapes the way we think, and determines what we can think about."
-- B. L. Whorf

I assure you that this is not the way it works;

Sorry, but this is a subject I know a little about so you'll excuse me if I don't take your word for this. Just because your local autoparts shop has made some attempt at deriving a table of cross-manufacturer equivalences, doesn't mean that they knew what they were doing when the did so. Having been burned (expensively) with just such a "Yeah! That'll do the job" assurance...I'm twice shy!

plugs with different operating parameters get a different part number.

Yes, within any given manufacturers range. But, as the presence of two or more 'equivalent' parts within a single manufacturers range testifies, the ranges can overlap. So, for example, you might have:

Cylinder Temp: low----A--B-------C-------D-----------high
ManA
  plug1        ********** 
  plug2             **************
ManB
  plug1          *********************
  plug2                   *********************
[download]

For a vehicle requiring C, ManA.plug2 and ManB.plug2 are interchangable.

But using that equivalence to suggest fitting ManA.plug2s to a vehicle D, specified as taking ManB.plug2 could be costly.

This is why all the manufacturers ask for the application in order to recommend the plugs in their range.

You might want to read this, paying particular attention to the last paragraphs on pages 2 and 3.

---

Examine what is said, not who speaks -- Silence betokens consent -- Love the truth but pardon error.

"Science is about questioning the status quo. Questioning authority".

In the absence of evidence, opinion is indistinguishable from prejudice.

"Too many [] have been sedated by an oppressive environment of political correctness and risk aversion."