Revealing Trends Based on Defined Queries in Biological Publications Using Cosine Similarity

Extracting valuable information in terms of number and content of published papers in any field of research will simplify decision making for future researches and investments. A novel and simple text mining approach, called TrendFinder, has been developed in this paper to reveal the content-based trends of expert-defined queries in selected biological published papers during the last five decades. So, in order to evaluate the results, three different data sets were collected and four vectors of selected keywords were considered as the four queries. "Title", "Published date" and the "Abstract" were downloaded for three series of journals namely, "Conservation Biology", "Ecology", and "The American Naturalist" as data sets, including total number of 19,010 papers. In order to show the trend between each query and the Abstract of each paper, Cosine similarity method was used by TrendFinder. Afterwards, three diagrams were demonstrated content-based trends of the four defined queries on the three provided data sets.