In this paper, we propose a system that serves as solution for the problem of automated assignment of reviewers to papers. With a steady increase in the number of research domains and huge submissions at journals and conferences, peer review happens to be the pivotal element to maintain quality standards for academic publications. Scientific and vigorous process for reviewer assignment is very crucial. Assigning appropriate reviewers poses a great challenge as it needs to consider many important aspects of like- relevance between reviewers and submissions, expertise, authority, diversity, recency and scientific impact. Existing approaches are based on matching the set of reviewers with submitted papers and assignment maximizes the similarity by satisfying the constraints such as load, coverage and conflict of interest. Traditional approaches be unsuccessful in i) identifying the multiple multi-disciplinary subject domains of paper and reviewer ii) assign set of reviewers so as to cover all the subject domains of paper achieving higher topic coverage. The proposed system addresses both of these issues. The proposed is named as UPRPAS (Unsupervised Proactive Reviewer Paper Assignment System) uses Latent Dirichlet Allocation (LDA) based algorithm to build the topic model-based on the extracted contents of submissions and expertise of reviewers for calculating the similarity, and then find the best match and assignment. The basic idea is to inevitably build representations of semantically relevant aspects of both papers and reviewers in order to facilitate the construction of a relevance matrix. The performance of the proposed systems is evaluated using conference datasets and is compared with baseline algorithms. Experimental results show that paper and reviewer profiles are built more accurately with higher collective matching degree and topic coverage. The systems accurately perform the assignment of reviewers to papers. The work also contributes a reviewer matching dataset and evaluation that will be useful for further research in this field.

Bag of Words (BoW), Coherence of Topic Model, Latent Dirichlet Allocation-LDA, Perplexity, Reviewer Assignment Problem-RAP