In accordance with Article L111-7 and Article D111-7 of the French Consumer Code, ALVISS.AI wishes to inform
you through this section of the methods by which content, services or products may be referenced or
dereferenced on its services, as well as the criteria that may influence the ranking in the results pages.
Referencing, dereferencing and ranking methods
Alviss.ai aims at indexing references of publications regarding any topic from any scientific journal
including preprints and scientific blogs of high quality. Sources include Pubmed, Arxiv, Biorxiv, Medrxiv,
Twitter and many others. It filters out predatory journals. You
can ask to be referenced or dereferenced using the
The PaperRank algorithm analyzes the feedback publications have been having on the Alviss platform and the
Web in general including social media and citations in other publications. When the feedback is positive the
rank of a given publication improves. Ranking is applied in the recommendation system of the scientific
watch functionality of the Alviss application and in the search engine.
By default, the search engine ranks articles by their relevance. No commercial agreement influences the
ranking of publications in the search engine.
The Home page shows new articles' information in a condensed format with a horizontal navigation.
Upon clicking on a section or an article you will navigate deeper into the site with more information
displayed to you in a vertical format.
Your domains section
Alviss uses algorithms to recommend the best and recently published publications based on your domains of
interest and the estimated quality of the publications. The recommendation system takes into account your
interaction with the Alviss platform.
For example if you liked (clicked on a button) a publication
Alviss will try to find similar content.
Have you ever wanted to know which publications everybody is talking about right now? They will appear in
the Trending section.
It is possible to sponsor your publications inside the Alviss application. They will be displayed in the
Sponsored section. When no articles have been sponsored the section is hidden.
Relevance reflects the traction of topics contained in an article in today`s debate. This metric
uses PaperRank for its calculation.
Originality represents the rarity of topics studied by an article in the previously published
Methodology increases when an article reports methods that are associated with a higher level of evidence
e.g. randomized trial or meta-analysis.
Feedback is the percentage of positive feedback an article has been having.
Quality is an overall metric representing the area of the radar plot obtained using the metrics Relevance,
Originality, Methodology and Feedback.
H-index is the maximum value of h such that the given author has published at least h
articles that have each been cited at least h times.
Aldex (Alviss index) is the maximum value of a such that the given author has published at
least a articles that have each been cited at least a times weighted by the duration since publication and
positiveness of the feedback.
Influence represents the breadth of an experts' network.
Leadership reflects the number of people under an experts' management as per their published work.
It also increases when the published work is very important e.g. guidelines and recommendations.
Quality is an overall metric representing the area of the radar plot obtained using the metrics H-index,
Aldex, Influence and Leadership.