Let's say I have a defaultdict in the following form:
theta = defaultdict(float)
The key consists of a tuple of strings i.e. (label, word), and the associated value is the probability that the given word fits the given label (part of speech tagging).
For example, the word 'stand' could be a noun or a verb. So I could do something like:
theta[('NOUN', 'stand')] = 0.4
theta[('VERB', 'stand')] = 0.6
theta[('ADJ', 'stand')] = 0.0
and so on for the remaining parts of speech labels.
What I need to do is have the dictionary return a value of 1 by default if it is invoked with a word that it does not contain and the associated label is 'NOUN', and return 0 for all other associated labels. For example:
value = theta[('NOUN', 'wordthatdoesntexist')] # this should be 1
value = theta[('VERB', 'wordthatdoesntexist')] # this should be 0
How can I do this? Can I do it in the initialization step, using lambda? Or is there some other way?
A defaultdict can't do that; the default factory doesn't have access to the key. You'd have to write your own dict subclass, using the __missing__ hook dicts look for when you try to access a missing key:
class SomeAppropriateName(dict):
def __init__(self, *args, **kwargs):
super().__init__(*args, **kwargs)
def __missing__(self, key):
val = 1.0 if key[0] == 'NOUN' else 0.0
# Uncomment the following line if you want to add the value to the dict
# self[key] = val
return val