Iranian Journal of Information Processing & Management (Dec 2019)
Assessing challenges and rate of matching Content Retrieval Conformity with Images in Scientific Databases
Abstract
The purpose of this research is to match the content retrieval with images in the articles of scientific databases. The research method is a correlation-based procedure and the statistical population of the present study is the scientific databases such as Science Direct, Scopus, Medline and Web of Science. Six papers were extracted from each base along with a separate issue. First, the title, abstract, problem statement and conclusion of the articles were inserted into the software Etreme Picture Finder, separately. Then, the images related to title, abstract, problem statement and conclusions the articles were extracted from the relevant software. The images extracted have been inserted into the Visual Similarity Duplicate Image Finder software. The results show that the highest level of content matching was found with the images in the WebAsiensis database and the least amount of match in the database of the SinjServer. There were also the most similarities between the titles, the problem statement, the abstract and the conclusions with the images of the articles in the bases. Failure to observe the standards of visualization in the images used in scientific articles was a serious challenge. Images that have been removed from image editors (text, color, edges, margins, etc.) have not been recovered, and if recovered, they have had a fairly similar resemblance. The lack of comprehensiveness of the superficial capabilities used was the next challenge. Another problem is the non-observance of or compliance with a standard style sheet in the layout of images in the articles. Some images have been brought to the articles regardless of the text. This item was found to be smaller in the Web site, but was seen in many other sites
