OCR is great for continuous homogenous rich text but sometimes when you have tables it is dicey and does not perform so well. Causes for this are mostly the text resolution, contrast and fonts embedded in the PDF.
Here is an alternative approach you could try:
Step 1. Open this PDF of your in word manually, yes word supports PDF formats but on occasions cannot render the correct formatting. If you are not interested in the formatting anyways, then this should be a good start point.
You can use word activity to open PDF in UiPath and read its content and save it as a variable.
This method depends on the type of PDF, if the PDF is saved with image as the content, this method will not work because the text then is present within an image and not rich text. So word will only show an image and read content (variable) from word will be empty.
Step 2. Then you perform either standard string manipulation or Regex expressions to extract the text you are interested in.