First you take the XPS data. Now ,Heat it up to some temperature with Oxygen environment and take the XPS again. If there is change (means shift )in peaks you can conclude oxygen vacancy was there. Because oxygen vacancy has large electrostatic energy. Due to reduction in the oxygen vacancy there must be change or distortion in the corresponding oxygen related peak. You must have to do good convolution to detect this change.
The chemical composition can be determined using the area under the curve of the elements being analysed. As Buddhadev said as we go on removing electrons from the element shift in BE is observed which increases with increase in oxidation state. Thus if u hav observed decrease in BE of the element bonded to oxygen u can predict oxygen vacancies. Further you must be vary careful with XPS measurement since with all possible care and accuracy oxygen can be detected upto 5% of accuracy with reference to some standard sample.
As Budhadev and Ganbavle said is correct. But the direct and easy way to identify the oxygen vacancy in semiconducting metal oxides are Photo luminescence spectroscopy.