High-Level Project Summary
Today, the main problems of scientometric indexes are not taking into account the weight of the reference (after all, citing by a student and a Nobel laureate are quite different things) and the lack of proper differentiation of "productivity" by specific subjects (some scientists have several scientific interests, but the citations for each of them can differ significantly). Our project is created to take into account such nuances. Accounting for links here is more complex and similar to the PageRank system. Indexes take into account both the "weight" of citations (number of references to an article referring to an article by the specified author) and sorting works by different topics.
Link to Final Project
Link to Project "Demo"
Detailed Project Description
This project is planned as part of the єНаука (eScience) platform - a service for the interaction of business and science, the search for scientific supervisors, co-authors, grants, scientific events, and centers for sharing equipment.
Actually, it is important for us to stop blind measuring with indices. Instead, the rating should only reflect how well-suited a particular scientist is to expertise in the particular requested subject matter.
We were inspired by the PageRank system. Our algorithm takes into account not so much the number of citations as their "quality" (weight). Each article thus has a ranking determined by the "weight" of its citations (how cited are the articles that link to that article). It often happens that the topic of the article is too narrow and "not hyped", but still makes an important contribution globally (Just as the picture will not be assembled without the last piece of the puzzle).
Here, for example, is the case where an article has only two citations, but the linked articles are quite cited:
https://scholar.google.com/scholar?oi=bibs&hl=en&cites=6755552079292230834
In addition, when searching for an expert, it is important to take into account not so much the general index of a scientist, but his success and significance in a specific narrow field. That's why we've also included an option to split indexes by tags.
We also tried the "applied output" parameter, but it is adequate only for applied physicists, and in the case of determining the connection "fundamental->applied" a more complex algorithm is required.
After finalizing our algorithm after the hackathon, we will make it an important part of the єНаука (eScience) platform and hope that others will use it as well.
Space Agency Data
Inspired by: https://science.nasa.gov/open-science-overview
Hackathon Journey
Members of our team are also participants in the Innovation Management course (4InnoPipe), which started this September. Our team chose to solve the problems of the scientific community as its topic - both in Ukraine (our homeland) and around the world.
References
Data for testing are taken from: https://scholar.google.com
Tags
#Science, #RaitingSystem, #graph, #search