Last link contains a chart clearly showing how this kind of algorithm works. (I understand and memorized this chart.)
$k$ is to hide $b$. $m'_0,m'_1$ is to reveal exactly one message, because Bob knows only one of $k_0,k_1$.
Bob make a key-lock pair, where $k$ is a key, $k^e$ is a lock(box with lock) to protect $Hash(b)$. Alice make two key-lock pairs, $k_i$ for $Hash(i)$, and locked $m_i$ with $k_i$.