Our semantic search should be used to explore a previously unknown research field. It works best when the user provides us with a sentence or question about a COVID-19 related research field, i.e. "How does COVID-19 affect childrens mental health?".

The search algorithm is able to analyze and generalize a given query and provide the user with papers that match the query's topic. It is capable of recognizing connections between words, e.g. temperature, weather, humidity. These connections are used to show papers of the overall topics that do not necessarily contain any word that the user provided.

We use natural language processing techniques to find correlations between a given query and the content of a paper. For semantic analysis, we trained a BERT model on our dataset that learned to find similarities between papers and a given query.

You can find out more details about the project at the frequently asked questions page.

The keyword search uses Elasticsearch to find papers with matching keywords in their title, abstract or authors. Despite being very efficient when the user provides a good set of keywords, the search is not able to generalize the search query.

You can find out more details about the project at the frequently asked questions page.