Topic on User talk:JakobVoss

Hi! Sure about this? Cannot see that a thesaurus (Q17152639) shares all the properties of a thesaurus (Q179797), so I wouldn't expect a subclass of (P279) relation here. Btw. I'm also unsure if thesaurus (Q179797) subclass of (P279) dictionary (Q23622) is really right… since the former is not about spelling, while the latter is.

Yes. I just found a good summary in an article by Vreda Pieterse and Derrick G. Kouriein who cite Spärck-Jones (1992) on the issue of what "thesaurus" is:

• "Most commonly it is used to refer to a vocabulary reference work to aid writing" (thesaurus (Q179797))
• "Secondly, in the library and information science (LIS) field, a thesaurus is a technical term that refers to an instrument that is used for vocabulary control in order to achieve consistency in the description of items and to facilitate retrieval" (thesaurus (Q17152639))
• "Lastly, in the AI field the term refers to the data store of words and phrases that is commonly used as a natural language processing resource" (Q??)

Anyway, thesaurus (Q179797) is no superclass of thesaurus (Q17152639) but it's a subclass of dictionary (Q23622) such as many other kinds of dictionaries, only some of them about spelling.

Google does not seem to know "Vreda Pieterse Derrick G. Kouriein". Any more precise link, digital if possible ?

Spärck-Jones, Karen. 1992. Thesaurus. In Stuart C. Shapiro, ed., Encyclopedia of artificial intelligence 1 2. New York, NY: John Wiley & Sons Inc., pp. 1605–13.

That's a good summary. There also seems to be some confusion on whether a thesaurus has to include hierarchical relationships or not, but in general that seems to be the defining factor that distinguishes it from a controlled vocabulary (Q1469824).

I wasn't aware of the usage within AI, but perhaps we need an item for that as well.

Something else: We have classification (Q5962346) (tree) and classification scheme (Q3475629) (tree), both currently subclass of (P279) knowledge organization system (Q6423319) and not properly disambiguated. What do you think about making just one of them subclass of (P279) knowledge organization system (Q6423319), and making the other about the process or the field of classification? Perhaps classification scheme (Q3475629) subclass of (P279) knowledge organization system (Q6423319), but then what relationship to use between classification scheme (Q3475629) and classification (Q5962346)..? In some sense perhaps classification (Q5962346) product (P1056) classification scheme (Q3475629), but that's only one aspect..

I am trying to clean up the whole knowledge organization system (Q6423319) tree but it's complicated. I agree that there should only be one classification as subclass of KOS but it's not easy to tell which because most Wikipedia language editions don't distinguish between classification as product and classification as process, research area etc.

The process of classification is classification (Q13582682). I will merge classification scheme (Q5962346) and classification scheme (Q3475629)!

Jakob, those definitions seems a bit lax to me, in particular the wikipedia articles gives more details about the kind of instrument. Do we really have example of a thesaurus that does not involve relationship beetween terms ?

Anyway, if it does, they fails to capture something : written thesaurus that actually use relatinships beetween terms are totally mixed up with dictionaries and we just don't have an item for it. So just basing ourselves on those definitions does not work imho.