On the bottom of page 7 of the paper Logarithmic Integrals by Morshed (arXiv link) there is an "interesting generating function": $$ \log(1+x) \log(1-x) = \sum_{n=1}^\infty \left(H_n - H_{2n} - \frac{1}{2}\right) \frac{x^{2n}}{n}$$ but it doesn't seem to be correct. The function on the left has the series $$-x^2 - \frac{5}{12}x^4 - \frac{47}{180}x^6 - \ldots$$ while the function on the right is $$-x^2 - \frac{13}{24}x^4 - \frac{67}{180} - \ldots$$ These do not match. Not knowing how to actually find the power series for $\log(1+x)\log(1-x)$, I decided instead to try to identify the function on the right, using the identity $$\sum_{n=1}^\infty H_n x^n = \frac{-\log(1-x)}{1-x}$$ as a starting point. In the end I was able to obtain $$\sum_{n=1}^\infty \left(H_n - H_{2n} - \frac{1}{2}\right) \frac{x^{2n}}{n} = \log(1+x)\log(1-x) + \tfrac{1}{2}\left(\mathrm{Li}_2(x^2) + \log(1-x^2)\right).$$ But the power series of $\mathrm{Li}_2(x^2) + \log(1-x^2)$ is very simple (it's $\displaystyle \sum_{n=1}^\infty \left(\frac{1}{n^2} - \frac{1}{n}\right) x^{2n}$) and so $$\log(1+x)\log(1-x) = \sum_{n=1}^\infty\left(H_n - H_{2n} - \frac{1}{2\color{red}{n}}\right) \frac{x^{2n}}{n} \tag{*}$$ which means the paper had a typo.
Question. How would you derive (*) without knowing it had anything to do with harmonic numbers?
I'm dissatisfied with my derivation because I started with a "guess" that was "close" (by virtue of it being a typo) and was able to go from there. I'd be especially interested in a "power series proof" (multiplying together the series for $\log(1-x) = -x - x^2/2 - x^3/3 - \ldots$ and $\log(1+x) = x - x^2/2 + x^3/3 - \ldots$ and simplifying the coefficients).