Abstract
Searching for information within a database using a wide range of search terms usually leads to an excess of search results. We propose a method of information seeking that uses exploratory filtering instead. Document information retrieval was conducted to support the planning of the new research theme. The obtained document sets were filtered using a dictionary of known information. After filtering, the remaining information was arranged in a list of infrequent search results. For example, literature was searched for using a disease name, and the normal targets of a specific search were filtered out. Subsequently, a list of novel action mechanisms was obtained. We propose a method of semiautomatic filtering using PubMed API by removing the general language and synonym detection. It is important to be able to extract infrequent but significant details from a surfeit of information.
