[Cover] [Contents] [Index] Previous page Next Section

Page 240
§10.3.2 Language Identification With Approximations
In the following an approximation to a language L will be taken to be another language L' such that 0240-001.gif and L' "conforms" to L. We will consider two different formalizations of conformity. We first introduce the notions of relative density of L' in L and of uniform relative density of L' in L. These notions are then used in Definition 10.40 to present the two formalizations of conformity.
10.39 Definition Let L = { x0 < x1 < x2 < . . .} and L' be members of  e .
(a) The relative density of L' in L (written: rd(L', L)) is defined as:
0240-002.gif
(b) The uniform relative density of L' in L (written: urd(L', L)) is defined as:
0240-003.gif.
10.40 Definition Let 0240-004.gif. Let L and 0240-005.gif.
(a) We say that L' is language d-conforming with L just in case we have
(1) 0240-006.gif; and
(2) 0240-007.gif.
(b) We say L' is language uniformly d-conforming with L just in case we have
(1) 0240-008.gif
(2) 0240-009.gif
The paradigms now to be discussed are analogous to the function paradigms introduced in Definitions 10.30 and 10.31. A scientist attempting to learn a language L is provided with a grammar for a language 0240-010.gif, in addition to access to a text for L. In the first paradigm, L' is a language that d-conforms with L; in the second paradigm, L' uniformly d-conforms with L. Thus (just as before), we conceive our scientists as computing mappings from. N × SEQ to N. Once again, a definition similar to Definition 10.17 can be formulated to describe convergence on advance information x and text T. Officially, we proceed as follows.

 
[Cover] [Contents] [Index] Previous page Next Section