1. Generally, we can consider from theory and experiments.
2. Novelty of theory: firstly, make sure the key contribution, and search for related work on that part, get a rough understanding on the novelty.
3. Sufficient experiments: if paper are constructed from their own ideas and several well established ideas, make sure they have comparison using only their ideas with other methods, because sometimes simple ideas accumulation will generate good results. It's not enough to prove the efficiency and accuracy of their own ideas. Such paper lacks novelty, just using these kind of trick, so be careful.