An Unsupervised Approach to Genuine Health Information Retrieval Based on Scientific Evidence

In contemporary society, more and more people refer to information they find online to meet their information needs. In some domains, such as health information, this phenomenon has been particularly on the rise in recent years. On the one hand, this could have a positive impact in increasing people’s so-called health literacy, which would benefit the health of the individual and the community as a whole. On the other hand, with a significant amount of health misinformation circulating online, people and society could face very serious consequences. In this context, the purpose of this article is to investigate a solution that can help online users find health information that is relevant to their information needs, while at the same time being genuine. To do so, in the process of retrieval of estimated relevant information, the genuineness of the information itself is taken into consideration, which is evaluated by referring to scientific articles that can support the claims made in the online health information considered. With respect to the literature, the proposed solution is fully unsupervised and does not require any human intervention. It is experimentally evaluated on a publicly accessible dataset as part of the TREC 2020 Health Misinformation Track.