Re: [WNET] Sense keys in HyperDic 2.0

McBride, Brian wrote:
> Are you thinking in terms of adegree of precision that Wordnet does not
> provide?

Most certainly ...

I liked your phlogiston example.

However, I think that given the imprecision - stability is relative.

If we have a sense ID x which is related to certain words and phrases in 
version N, and in version N+1 there is a desire to have a new sense 
which is a minor change to the sense indicated by x, I think it will be 
an *engineering* decision as to whether that change is sufficiently big 
to require a new sense ID y, or whether sense stability is not violated 
by *changing* the sense and reusing sense ID x.

Continuing with your physics case  - the concept atom gets revised 
relatively frequently - and in many ways the older conceptualization of 
'atom' ceases to be available - but whether this requires a new sense ID 
or not is open to debate in my view. (Physicists have got by without 
needing to invent a new word, although I guess they might qualify 'atom' 
if they wish to talk about an atom given a particularly physical theory)

So, I see sense stability rather like API stability ... as you rev some 
software you want the deprecated API calls to still work, but there are 
corner cases and you accept that, and things do get obsoleted.

Jeremy

Received on Thursday, 22 July 2004 10:53:26 UTC