Designing a Web-based Journal Recommender System: An Experimental Study on the DOAJ Dataset

Researchers generally prefer to publish their manuscripts in a journal related to their subject area. However, they often struggle to identify suitable journals for manuscript submission. This paper aims to develop a content-based recommender system to address this problem. This system recommends similar journals based on a known journal title or subject area. Additionally, it will attempt to design a web-based search interface for the Recommendation System (RS). The DOAJ database was used for this study. This dataset contains 54 columns as fields, and 16454 journal records approximately. Out of which, 21 pertinent columns were chosen for this study. The Flask web framework for Python was used to display results on the web. TF-IDF vectorisation with cosine similarity was employed to compute similarity scores between TF-IDF vector representations of journal metadata.

View article page

Download PDF