I'm reviewing a conference manuscript (computer science) and i found 2 other recent methods related to the same problem which were published on Arxiv. So, i think it is highly recommended to compare their method to those ones. The Arxiv papers also claimed to beat the state-of-the-art, but on different datasets, so one cannot compare them based on the reported numerical results.
So, should/could i say that it is a weak point that the authors have not included/used those recent relevant works in their introduction or evaluation basis? Or they do not have such an obligation as those papers have not been yet published in any proceedings?