Resolving Conflict of Interests in Recommending Reviewers for Academic Publications Using Link Prediction Techniques

An honest peer-review process is a key for producing high quality scientific research. However, this process depends on two main factors: (1) the expertise of reviewers in the topic of a submitted paper and (2) the relationships between reviewers and authors. To satisfy the first factor, editors and conferences chairs manually select reviewers. Whereas to prevent any conflict of interest (CoI) between reviewers and authors to satisfy the second factor, reviewers and authors are asked to declare any CoI manually. Such a solution is tedious to all actors and error-prone. To solve this problem and satisfy those two factors, we have developed a novel framework that (1) recommend expert reviewers and (2) resolve the CoIproblem. To develop our framework, we have represented the DBLP citation network dataset as a graph database using Neo4J. A Cypher queries used to select expert reviewers. Various link prediction algorithms, especially the Adamic Adar and the Common Neighbors algorithms, have been utilized to resolve any notential conflict of interest.