<![CDATA[The peer review process is one of the crucial stages in scientific publications to ensure the quality, validity, and originality of a scientific article. One of the main challenges is finding reviewers who are competent and relevant to the research topic. This research aims to develop a Content-Based Filtering-based reviewer recommendation system to make it easier to select the right reviewer. This method analyzes the content similarity between the article to be reviewed and the reviewer's expertise based on historical review data in the form of title, abstract and track name using TF-IDF Vectorizer and Jaccard Similarity techniques. The results indicate that the Jaccard approach achieves a precision of 8.33 %, recall of 6.75 %, F1‑score of 7.45 %, and accuracy of 20 %, whereas Cosine Similarity yields only 5 % precision, 4 % recall, 4.4 % F1‑score, and 15 % accuracy. These low metrics reveal limitations in the reviewer profile dataset and underscore the need to enrich features, improve data completeness, or adopt a hybrid methodology. The primary contribution of this research is an empirical demonstration that TF‑IDF combined with Jaccard outperforms Cosine in the context of reviewer recommendation, while offering practical recommendations for enhancing recommendation systems within the scientific publication ecosystem.]]>
Copyrights © 2027
