If all you want to do is keep track of which authors contributed to which papers, then all you need is a simple intersecting entity (see more) like this:
Here the columns of the
SUBMISSION table are just the two primary key columns from
AUTHOR. These two columns will jointly form the primary key of the intersecting table and will also each individually be foreign keys to their respective tables.
NOTE: Don't add any other columns to the
SUBMISSION table, unless those columns can reasonably be expected to vary row by row for the same paper and for the same author. For example, it would be a bad idea to add a date to the submission table unless a single paper is submitted at various times by multiple authors. If you need to track information about the event of an author or group of authors submitting a paper, then you need to normalize out the extra information to avoid creating update anomalies. Consider the following:
In the above ERD, notice how the extra information that is not author-dependent remains at the
SUBMISSION level, while the new intersecting entity
CONTRIBUTION is added to record the connections between authors and a particular instance of the submission of a paper.
This four table model is more robust because it allows for papers to be resubmitted multiple times with changes in authorship each time. If this kind of real world nuance is important to your system then you would be better off with the four table approach than with a three table approach.
Here is the four table model with more column details at OP's request: