Better business definitions for Data integration

When I did my Ph.D., I always ask myself for "how to make better ontology". Today, I found part of the anwers in this web seminar(https://wilshireconferences.webex.com/mw0306lb/mywebex/default.do;jsessionid=wT4PLrhNC1GvTJVhqQjZvcFlGq3DL2vz5h1hKVnz5MpvScwnmvRh!-1186159457?nomenu=true&siteurl=wilshireconferences&service=6&main_url=https%3A%2F%2Fwilshireconferences.webex.com%2Fec0605lb%2Feventcenter%2Fevent%2FeventAction.do%3FtheAction%3Ddetail%26confViewID%3D405374507%26siteurl%3Dwilshireconferences%26%26%26).

* Wordnet is a good starting point of ontology?
Yes or no..... yes, because it can give the brief overview of a word. No, because it is too short.

* how to create definition of DATA (for me, "definition of data" is same as ontology)

  1. Think about the term belongs to where, and specify the unique meaning of the term in your case (i.e. Human is a rational animal --> human belongs to animal, but human is unique, because human is rational)
  2. Don't make a definition short. It is very important to create definition by adding as much as information you have
  3. Don't use synonym in the definition (But, it is good to have synonym list)
  4. Don't refer each other in the definition
  5. Defintion is like a life, because it can be changed over time. So, manage definition over time.
  6. Ontology is a good source to create definition, since we can share it not only human but also machine
  7. Create definition in the begining of projects 
  8. Add class/properties restrictions. Add broad restrction to top classes, add specific restriction to bottom classes. Don't add "disjoint" restriction at the beginning, because disjoint restriction can be complict to other restriction. Add "disjoint" later.
  9. Check ontology using validator (reasoner)