When we say $\{\alpha_i\}_{i=1}^r$ a $\mathbb Q$-basis of $k$, doesn't this mean that $k=\operatorname{span}_{\mathbb Q}\{\alpha_1,\dots,\alpha_r\}$, and therefore we can't get things like $\frac1{\alpha_i}$ which are in $\mathbb Q(\alpha_1,\dots,\alpha_r)$?
Or can we somehow get things like $\frac{f(\alpha_1,\dots,\alpha_r)}{g(\alpha_1,\dots,\alpha_r)}$ ($f,g\in\mathbb Q[X_1,\dots,X_r]$) by linear combinations of the basis? Using minimal polynomials or something?
I appreciate any assistance.
Well, if $k$ is an extension field of $\Bbb Q$, then $k$ is also a $\Bbb Q$-vector space.
In your example, $\alpha_1,\ldots,\alpha_n$ form a $\Bbb Q$-basis of $k$ and so each element of $k$ can be written as $$\sum_i c_i\alpha_i$$ where the $c_i$'s are rational numbers.
This includes that the inverses of the $\alpha_i$'s belong to $k$ (as $k$ is a field) and can also be represented this way.
Think about $\Bbb Q(\sqrt 2) = \{a+b\sqrt 2\mid a,b\in\Bbb Q\}$ with basis $\{1,\sqrt 2\}$. Here $$(a+b\sqrt 2)^{-1} = \frac{a}{a^2-2b^2} - \frac{b}{a^2-2b^2}\sqrt 2.$$