I don't know Python, but looking at the documentation, it'd go something like this (probably contains a few syntax errors):
thesaurus = dict()
dataFile = open('somefile.dat', 'r')
for line in dataFile:
parts = line.split('=')
word = parts[0]
synonyms = parts[1].split(',')
thesaurus[word] = synonyms
sentence = raw_input("sentence: ")
originals = sentence.split()
replacements = []
for i,word in enumerate(originals):
replacements[i] = random.choice(thesaurus[word])
print replacements
Keep in mind that is probably not good code. You'll want to add checks to see if the word exists in the thesaurus and such. Also, that code assumes the file is in this format:
jump=hop,skip
sound=audio,noise,racket
If you wanted to avoid using up lots of memory, you could store the thesaurus in a database rather than in an in-memory dict.