I think I have a working solution to this problem. Here's how I would solve it:
You look at a 6 appearing as a success and everything else as a failure. From here you can use the definition of expectation with the binomial probability function to get something like
(0 $\times$ P(exactly zero 6's)) + (1 $\times$ P(exactly one 6)) + ... + (10 $\times$ P(exactly ten 6's))
to solve the probability portions you use the binomial probability function.
Is this solution a valid way to solve the problem? Is there a better way to solve it? Because this method looks way too time consuming.
Assuming independence, you can do it easier:
Denote by $X_i$ the event "6 occured in roll i", and then:
$$E(X)=\sum_{i=1}^{10} E(X_i)=10E(X_1)=10*\frac16=\frac{10}6=1\frac23$$
EDIT: As noted below, you don't even need to assume independence.