Likelihood

The likelihood is $L(\theta) = P(D \mid \theta)$. The direction is the part people reverse: the data is fixed and known, and $\theta$ is what varies. It answers "how probable is what I actually saw, if these parameter values were the truth?"

That is not the probability that the parameters are correct. $P(D \mid \theta)$ and $P(\theta \mid D)$ are different quantities, and a likelihood is not a probability distribution over $\theta$ — it does not integrate to one over $\theta$, so you cannot read a confidence interval off it directly. Turning it into a genuine belief about the parameters requires a prior and Bayes' theorem.

Likelihoods are also only comparable between different parameters on the same data, never across different datasets. Maximum likelihood estimation picks whichever $\theta$ makes this quantity largest, and because it is a product over observations, implementations always work with its logarithm.