The quoted text indicates only that they parse out "structured data elements." The test for Copyright infringement is a four prong test.

1. Access to the original work 2. Copying of the original work 3. Substantial similarity between the oriinal and the copy 4. Damages

If you were to compare select elements from the database and compare each element to the work from which is was scraped, would you find that there is a substantial similarity between the data elemment and the work in its entirety from which the element was obtained? If the answer is NO, then the tjhird priong of the test fails, and there is no infringement.