A free fortnightly newsletter on Taxonomy, Thesauri & Ontology and Semantic Publishing
How AI-based search tools help scientists conduct systematic literature review?
Today, when scientists attempt to conduct a systematic literature review, the number of papers on a topic overwhelms them. However, there is a bevy of new AI-based search tools offering targeted navigation of the knowledge landscape. In fact, developers are looking to leverage these tools to automate how hypotheses are generated and validated, writes Andy Extance, freelance science journalist for Nature —an international journal of science.
Most of the AI-based search tools serve a specific niche. These tools provide scientists with a new view of the scientific literature than conventional tools such as PubMed and Google Scholar. In fact, many tools are helping researchers validate existing scientific hypotheses and identify hidden connections between findings. In addition, some of the tools have the capability to suggest new hypotheses for guiding experiments.
These tools are controlled by algorithms that perform two functions. The algorithms extract scientific content and provide advanced services, such as filtering, ranking, and grouping search results. To tag entities, developers can use supervised machine learning such as a paper’s authors and references. This practice will be useful in training sets to teach algorithms identify and extract the tagged entities.
To provide progressive services, algorithms often build knowledge graphs that map relationships between the extracted entities and display them to users. For example, AI could suggest a relation between a drug and a protein if they are mentioned in the same sentence. This will be encoded in the knowledge graph as an explicit relationship in a database, thereby making it machine readable.
Click here to explore the relative merits and demerits of AI-based research tools.