Given a stochastic process $\{X_t: t\in R^+\}$, which takes value in $R$, there is always a natural filtration $(\mathcal F^X_t)$ induced by $X_t$, i.e. $\mathcal F_t^X = \sigma(\{X_s^{-1}(A): s\le t, A \text{ is Borel-measurable}\})$.
Now, consider the inverse situation: Given a filtration $(\Omega, \mathcal F_t)$, (we may consider the special case: $\Omega = R^{R^+}$ first) is it true that there is always a stochastic process $(X_t)$ which is adapted to $\mathcal F_t$, and that $\mathcal F_t = \mathcal F^X_t$ for all $t\in R^+$? And if it is not true, can we add some conditions (hopefully, only some weak conditions) to make it true?