An already suggested solution for extending the Kyoto ontology is using a kind of interview that guides the user through the ontology. An Italian project that has already been working on this is the Senso Comune project: http://www.senso-comune.it/portale/ and http://www.loa-cnr.it/Papers/lexicon_oltramari-vetere.pdf.
Their basic idea is that non-expert users can extend a lexical resource for Italian without being bothered with the ontology. To enrich the ontology, the project uses a QA system called TMEO (Tutoring Methodology for the Enrichment of Ontologies) that is _embed_ded in Dolce. On the Senso Commune wiki (only in Italian, unfortunately) one can find examples of the questions that the user needs to answer:

TMEO v.0.9 (Italian)
Nota: Lemma + glossa = concetto (lessicalizzato)
1. Puoi toccare un/una [lemma] nel senso di [glossa]? (SI/NO) / Se (SI) Allora
(can you touch a [lemma])
2. Puoi contare un/una [lemma] nel senso di [glossa]? (SI/NO)/ Se (SI) Allora
(Can you count a/an [lemma])
3. Diresti che [lemma] nel senso di [glossa] è capace di decidere o muoversi? (SI/NO) Se (NO) Allora
(Does [lemma]have the ability to take decisions or to move?)
4. [lemma] nel senso di [glossa] è la materia di cui è composto un oggetto? (SI/NO) Se (NO) Allora
(Is [lemma] a material of which you can compose an item?)
5. [lemma] nel senso di [glossa] è piuttosto un oggetto costruito dall'uomo? (SI/NO)
(Is [lemma] an item that has been made by men?)
STOP

Altrimenti

6. In un/una [lemma]nel senso di [glossa] diresti che avvengono dei cambiamenti? (SI/NO)
Se (NO) Allora
7. Si tratta piuttosto di uno stato fisico? In un/una [lemma], cioè, nel senso di [glossa], puoi
osservare uno o più corpi fisici senza registrare mutamenti? (SI/NO)
Se (NO) Allora
8. O diresti invece che [lemma] nel senso di [glossa] è una misura di tempo? (SI/NO)
Se (NO) Allora
STOP

(The answers to the questions correspond to)

3. SI --> [concetto] IS-A AGENTE (Es. persona, cane, robot,...);
4. SI --> [concetto] IS-A SOSTANZA (Es., oro, acqua, carne,..); )
5. SI --> [concetto] IS-A ARTEFATTO (Es. tavolo, automobile,..); NO --> [concetto] IS-A QUALITA' FISICA (rosso, alto, duro,...)
6. SI --> [concetto] IS-A PROCESSO (caduta, corsa, migrazione);
7. SI --> [concetto] IS-A STATO (depressione, stato fondamentale (Fisica), sedere (verbo))
8. SI --> [concetto] IS-A QUALITA' TEMPORALE (corto, lungo, noioso (di evento); NO --> [concetto] IS-A ASTRATTO (numero, insieme, ecc.)


This list of questions corresponds to eight classes, but the ontology that is used in Senso Comune has about thirty. Furthermore, it seems that the user needs to answer the questions top-down through the ontology. This means that the user possibly needs to answer many more questions than these eight, which is not so user-friendly and should not be necessary either.
A solution for this, is to combine above TMEO procedure with questions that can shorten the time the users need to determine where a concept should be in the ontology.

Some possibilities and challenges:

1. First of all, determine if a concept should be a type in the ontology by testing rigidity/non-rigidity. (Rudify/Rolify)
2. Interview the user by presenting frequent relations from the term data_base_.
Possible relations that apply to the concept, like
- N prep N: ‘populations of species’
‘populations of vertebrates’
‘population of plants’
‘populations of algae’
-> prep ‘of’’ expresses a meronymy relation.
3. Derive patterns like N or N, N and N from the KAF documents: one of the nouns is already in the wordnet and related to the ontology. If so, the second noun is probably close related to the first noun, and therefore can be placed around the ontological type of the first noun.
apes (X) and humans (Y)
fern meadows (X) and grasslands (Y)
For every unknown X one could derive a list of Y, and determine where the Y’s are in the ontology. At that point a TMEO-like procedure can be started that takes the most frequent mapping to the ontology for Y as a starting point. If the user cannot decide or the outcome is wrong, a full TMEO procedure can be started.
4. Validation questions that ask for e.g. synonyms or hypernyms.
5. Specify the extra constraints for a new type in the ontology. (how?)

6. A more challenging possibility is not to ask the user, but ask Google. One could think of making an ‘Ontofy’ that incorporates some or even all of above possibilities.