System for evaluating the reliability and novelty of medical scientific papers

Abstract

• It is proposed a method for classifying medical texts based on their novelty and reliability. • Novelty and reliability measures are calculated based on a Knowledge Graph. • Knowledge Graph created by processing more than 500,000 medical research articles. • Quadrant-based visualization to group medical texts according to their novelty and reliability. As society develops, the number of published research articles raises. Besides, the pressure to publish has been increased because the competitiveness between researchers working on similar topics. Although this increment is desirable, it leads to multiple issues. At a reader level, insurmountable barriers to keep up with the state of the art appear. From a publisher perspective, it is a very demanding task to determine which research articles are worth publishing. Automatizing these tasks appears as a core solution. In the case of readers, a previous evaluation of articles would simplify the filtering process. As for publishers, they could perform preliminary selections or estimate the reviewing effort. This paper presents Medical Evaluator System for Scientific Interoperability ( MESSI ) system to overcome all these issues. It is able to evaluate the novelty and reliability of health-related texts. The novelty calculation is based on previously acquired knowledge after processing more than 500,000 papers. The reliability estimation is based on the reputations of similar articles calculated based on previously defined metrics. Multiple experiments have been addressed to illustrate the viability of the proposal. The obtained results show a good performance that encourage to continue evolving the system.