Conditional Probability Dice
Simulate bigram language models: roll words based on what came before
Configure Bigram Probabilities
The bigram model depends on conditional probability distribution for bigrams. We need to know, for each word type, given that we have already rolled that word, what is the probability that it will be followed by another word. Use the conditional frequency counts you derived to get the probabilities. You can enter the raw token counts. The normalize button will transform these counts into probabilities between 0 and 1.
Set the probability of rolling each word given the previous word. For each starting word, the probabilities of following words should sum to 1.0.
Sum: 0.00
Sum: 0.00
Sum: 0.00
Sum: 0.00
Sum: 0.00
Build Markov Model
Generate a visual diagram of your bigram model showing states (words) and transitions (probabilities).
Roll Conditional Dice
Current context: [Start of sentence]
Previous
Current Roll