I wonder if the characteristics of RG members will change after a full transformation? The description given by Ali Akhaddar tells about indicators regarding publications and not questions and answers. And in calculating research interest they will not take into account non-members of the RG. And what about auto-citation, can one use such instrument to increase the weight?
Research Interest is how we measure scientists' interest in your research.
Metrics are a powerful tool to evaluate science and move it forward. When used well, metrics can help researchers and institutions understand how their research is being received and applied around the world. But deciding what constitutes a good metric is a complex issue. While no single metric can give you the full picture of the impact your research is having, we believe our new Research Interest score can help you complete the picture.
At ResearchGate, we're committed to giving you insights on how people read, recommend and cite your work. Millions of verified scientists interact every day with each other's work through our platform – creating a unique opportunity to offer a broad understanding of interest in their work. With the Research Interest score, we now have a quicker and more comprehensive way to do just that.
How Research Interest is calculated
Through our research and input from scientists, we decided to focus on how ResearchGate members are showing interest in individual research items and how that interest is evolving over time. We believe that this new score can give authors a faster, more comprehensive picture than citations or reads alone.
We built the Research Interest score to be intuitive, so that researchers can quickly understand and use it in as many situations as possible. This score is focused on research items and scientists' interactions with them, using concepts that are familiar to our members. To provide an overview of a researcher's body of work, we've also added a Total Research Interest score, which simply adds up the Research Interest scores from all of an author’s research items.
What the Research Interest score includes
When researchers read, recommend or cite a research item, its Research Interest goes up. Based on our data and feedback from scientists, we chose to focus on these interactions to reflect the lifecycle of a scientist's increasing interest in a piece of research. First, a researcher accesses a research item. If it sounds of interest, they will read the full-text. If they like what they read, they might recommend it. And if the work is really relevant, they might cite it in their own research.
This is how we decided on a system for weighting the different forms of interaction:
A read has a weighting of 1.
A full-text read has a weighting of 3.
A recommendation has a weighting of 5.
A citation has a weighting of 10.
What the Research Interest score doesn't include
To make Research Interest meaningful to our members, we decided to exclude certain types of data:
Reads by people who are not ResearchGate members
Measuring interest from the verified scientists on ResearchGate allows us to provide the 'who' behind the metrics, which we feel is a unique value for authors in understanding how their work is received.
Multiple reads and recommendations by a researcher in a single week
A researcher interacting multiple times with the same research within a short period of time doesn't represent an increase in interest, but leaves the score more open to abuse.
Interactions from bots, crawlers and other automated systems
Our automated bot detection system is constantly monitoring abnormalities so that we can react quickly to any fraudulent activity. You can also send feedback to our support team if you suspect any unusual activity in your stats.
I wonder if the characteristics of RG members will change after a full transformation? The description given by Ali Akhaddar tells about indicators regarding publications and not questions and answers. And in calculating research interest they will not take into account non-members of the RG. And what about auto-citation, can one use such instrument to increase the weight?
New Total Research Interest on RG is a good move. But some of the shortcomings mentioned by by Dr.Valentyn are to be seriously considered. Interactions with peer researchers could also be reflected in RG.
It looks like the "scores" of researchgate are fairly closely related to "popularity". Nobel laureates and other outstanding scientist do not have necessarily extremely high scores (as is also true, e.g., for the h-index). Presumably, for a selection committee in science such scores and indices are, indeed, of minor importance.
Accordingly, one should consider them with caution and, perhaps, humour.
This is good, but it is not without shortcomings! How do you differentiate quality from quantity in respect to research interest measure... Taking into consideration two researchers A and B; A has 2 papers published in the Nature and New England Journal of Medicine, B has 45 papers among which 43 were published in predatory journals after previous rejections in good journals and two others in 'unknown' journals. In the long run, B has 400 research interest score and A has 100.
Here, do we want to say B attracts more weight than A?.
Although, this is a welcome development sieving seeds out of sand stones but there is need for a better metric system.
In my own opinion and little understanding about TRI. It is not mainly about the quality of the journal you published with but rather the interest it generated among the researchers on this platform alone. If ones paper has worthy impact it will attract the members of research community and will want to download, view it or do other things concerning the publication regardless of the publisher. There are good and useful research outputs are are published in low impact journals that have great impact with high citation counts. It is about quality research not only publication in high impact journal is enough to make a paper of good quality.
I fully agree with your considerations. Indeed, to evaluate the "relevance" of a publication, one has to study the quality and impact of that publication,
and not of the journal, in which it has been published. Actually, the positive impact is not necessarily given by the number of citations. Perhaps, for instance, colleagues may like to refer to that publication, because they want to point out errors in it. An example is the paper proposing, erroneously, "cold fusion" (about thirty years ago), which received almost two thousand citations.
Indeed, the article provides a profound analysis of the RG score. Thanks for the link. The paper gives a clear warning for (rather naive) selection and evaluation committees, when taking into account the score. Luckily, so far, I never encountered such naive committees, mixing up scientific quality or academic reputation with the RG score (or other metrics, as well). Roughly speaking, that analysis shows that the RG score measures partly scientific achievements, but also, to a fairly large extent, the activities of persons on RG.