Anyone interested in working on a "fragrance map."

Very good recommendation. But I think people who wants to work on this kind of map should be acquainted with:
1-chemical structure of perfumes,
2-As many as possible categorizations

Also we can suggest Grant, to add a part for developing this kind of activities, which can result in a perfume expert system.

I have been reading your posts with growing interest, Bigsly. This is a very ambitious project you are suggesting. The H&R map is in fact the best tool you can have for quick information about the character of any fragrance contained. But as you said, it's aoutdated by now. It was developed by H&R, Holzminden, Germany and has been part of their Book: >H&R Duftatlas<. All of the fragrances on that map are described in more detail in that book (last edition 1991 (!). What is special about their fragrance trees is that top middle and base notes are separately tagged in a similar manner as Turin-Sachez tagged their perfumes.

Examples: YSL Opium (f): top note: aldehydic spicy; middle: spicy floral, base: balsamic, sweet, warm.
KL Homme / Lagerfeld, in full detail: top: fresh, herbal (lemon, bergamotte, rosewood, lavender, orange, aldehyd); middle: floral woody ( carnation, jasmine, geranium, fern, cinnamon, rose, patchouli, sandalwood, cedar); base: sweet powdery (vanilla, benzoin, olibanum, musk, amber, civet)

I am afraid it would be wishful thinking waiting for the book and/or the map to be updated. And I even wonder if it could or should be done at all. There are so many obstacles:
1. new fragrant substances (synthetics) make it necessary to find better descriptions for what used to be violets, ambergris, civet, etc.
2. There are new fragrance notes (new fruits, various smells of water, ozone, etc.) one didn't have in colognes before, which means we have new fragrance families.
3. We no longer have two categories of fragrances, but three: m, f, & uniscents. Where do you place the latter? If you combine all of them on one map there will be a mess to sort out. Masculine and feminine fragrances do not have the same structure, and neither do the respective maps. Have you pondered a little on the approach Editions Malle have taken? I kind of like their way to graphically characterize their fragrances, and it is the only new approach I have seen so far.

The great advantage of the H&R maps remains their historical order and (relative) simplicity. So I would be happy if Symrise, as H&R are now called, would just update what they had developed in the seventies.

Have you seen Michael Edwards Fragrance Directory site?

Your initial proposal has the potential to become an excellent and useful tool. It also has the potential to become an arcane unwieldy behemoth. Research H&R - find out what obstacles prevented them from keeping their map updated - were there technical problems in the initial design or was it purely economics?

The biggest problem with good ideas is knowing when to stop 'improving' them.

Don't scoff at simplicity, it is a far more elegant solution than complexity.

My suggestions would be: Decide on a fragrance family structure. Have no more than two maps - Male/Unisex and Female/Unisex. Decide which fragrances to initially include on each map. Use Community input to position the fragrances in the same manner as H&R. Positioning similar fragrances close to each other would eliminate the need for distracting lines and boxes. Fragrance details would go into an attached index rather than clutter up the map.

Once we understand what this map is not, it is helpful. This map is not detailed. This is not information rich. Instead this map is a helpful grouping tool.

To me, the ultimate but unweildy tool would be a database that allows one to sort all fragrances with particular top notes, heart notes and base notes.

Finally, is Habit Rouge really an amber? It is a sweet oriental, but is it an amber?

http://www.basenotes.net/directory

Haarmann & Reimer's Genealogy is impressive, even considering its limited data set. Now known as Symrise, this is one of the companies that actually creates fragrances, so I would not dismiss its classifications so casually. The problem may not be that H&R's classifications are wrong; the problem may be that the user has a poor understanding of industry-specific terms of art such as "oriental," "ambery," and so on.

On the other hand, categories such as "cheapo," "high quality," and "highly blended" are of little usefulness. Our map may as well be organized according to "blue bottles," "red bottles," "tall bottles," and "short bottles." In the same way, comparing base notes only, and ignoring top or middle notes, seems of limited usefulness, like having a pie contest and judging only the crust and not the filing.

If we are serious about this undertaking, then we should consider it as two separate projects: First, what is our data source? Second, how are we visualizing our data?

Rather than creating an entire data source out of whole cloth, the Basenotes Directory could be used as a starting point for the data. The Directory has fragrances (objects), and fragrances have "notes" and "tags" (sets). The Directory is not complete, it is true, but rather than creating a whole new database, maybe that is just incentive for us to add to the Directory's completeness.

The data may be the easy part of the project. A map can only have two axes: What is the X-axis? What is the Y-axis? Alternately, we could use Venn diagrams, with the fragrances inside overlapping sets of notes or tags. Given the total number of notes, and the number of notes per fragrance, and the total number of fragrances, such a diagram may be impossible in two dimensions. Automating the process of creating the chart from the data would be very desirable. Are there any scientists amongst us who can weigh in on how difficult such a visualization would be?

On the other (third) hand, Michael Edwards has already done all of this work already -- creating a taxonomy, creating a data set, and presenting it visually. Fragrances of the World may be a little expensive, but reinventing the wheel is also expensive in terms of time and effort.

[Later that evening . . . .]

To answer my own question, there is a Java application called VennMaster available here: http://www.informatik.uni-ulm.de/ni/...vennm/doc.html

I whipped up a sample data set (which is a simple text file) of thirteen fragrances and ten notes. The resulting diagram was as complicated as the tax code. I imagine that thousands of fragrances with hundreds of notes would be indecipherable.

Either poor understanding, different perceptions, or both. I believe it's more than obvious that our minds register different things from something as complicated as a mix of smells.

Threads that return frequently are
- when someone is looking for a cologne 'similar to...'
- when people believe to have discovered that X is just a copy of...
Positive replies make me scratch my head occasionally because I cannot detect many of the similarities stated by other community members. And yet I believe that others are just as convinced to have smelled similarities as I am smelling the opposite. The best fragrance map in the world cannot solve the problem of subjective fragrance perceptions. While most of us can't, I find it easy to understand why Turin praises Calice Becker (Beyond Paradise, Tommy Girl). And yet his (and Sanchez') ignorance of the special qualities of perfumes made by Villoresi or Santa Maria Novella makes me speechless. The subjectivity of perception is the only explanation I have for such phenomena.
The H&R maps have been designed for printing (book and posters). That is a limitation we wouldn't have nowadays. I am thinking of different views of one comprehensive database, with and without the time axis. If you eliminated time, similar fragrances - presently defined by families and their subs - would all be on the same spot. You could use the former time axis to redistribute them vertically by another meaningful property expressing proximity and distance (sub-subgroups?). Instead of an annex, I would see a window opening for each fragrance clicked upon, and allowing a look at the details. And imagine a light blinking next to any fragrance on your screen that contains sandalwood!
I would be perfectly happy if Symrise just updated what they have, no matter how they treat uniscents, and let their data specialists and some graphics freak work out all the forms and queries for an attractive program to work with. A lot could be done, even on the basis of MS Access.

p.s. It might be interesting to look at the similar fragrances wizard BN once used. Also Osmoz have something like that, I believe, and I wonder how they find alternatives to your favorite fragrances.

This is an excellent point. Newcomers to any field generally try to understand everything in terms of how it is similar to what they already know. So, picking two fragrances that have only one thing in common -- sandalwood, say -- and saying that they are the same or similar, and ignoring the hundreds of other ways in which they are completely different. To put it another way, one could say that Raisin Bran and bread pudding are just the same because they both have raisins in them. Or that toast and cake are the same because they are both made with flour. No map will satisfy them because, look, X and Y are not next to each other when any idiot can see that they are identical.

Isn't it easiest to just point them (as someone did to me recently) at the search facility n the fragrance.org site?

I mean, at least that way they get an expert opinion.

Not to toot my own horn or anything (cause if I could do that . . .), but that was me. But to answer your question, the experts are obviously wrong or don't know what they're talking about or paid shills or something. In any field, there is no one quite so certain as a novice.

Narcus just asked a great question in another thread - what is the status of this ambitious and important project?

I am continually amazed at the number of interesting threads I somehow failed to view when they were first posted. This is one. Thanks for the bump TwoRoads.

Sorry to bump an old thread.

Is there any progress on this?