Methodology

The SJR2 (SCImago Journal Rank 2) indicator is a refined metric developed to capture a journal’s true scientific prestige, going beyond simple citation counts.

While many ranking systems treat every citation equally, SJR2 looks deeper. It evaluates who is citing the journal and how closely related the citing and cited journals are in subject matter. This provides a richer, more accurate picture of a journal’s influence within the scientific community. In essence, not all citations are treated the same. Citations from highly respected and topically similar journals carry more weight than those from unrelated or lower-impact sources.

Traditional citation counts often reward quantity over quality. A journal could appear influential simply because it is frequently cited, even if those citations come from less relevant or low-prestige sources. SJR2 corrects for this by introducing context and quality into the equation:

• Prestige flows through citation networks, meaning a journal inherits influence from those that cite it.
• The thematic closeness (measured using cosine similarity between co-citation profiles) ensures that citations from journals in similar research areas are more meaningful.
• Mechanisms are in place to prevent artificial inflation, such as capping how much prestige a journal can transfer and redistributing influence from journals that don’t actively cite others ("dangling nodes").

All SJR2 calculations are powered by data from Scopus, one of the world’s most comprehensive citation and abstract databases. Scopus is known for its broad international coverage, rigorous journal selection process, and structured indexing of scholarly content.

1. Citable Documents including research articles, review papers, short surveys and conference proceedings. Only documents published in the three years leading up to the current ranking year are considered.
2. Citations: These are the references made to those citable documents by other journals indexed in Scopus. But in SJR2, a citation’s value depends on its source and subject-matter relevance.

1. Calculating Raw Prestige (PSJR2)
   This first step computes each journal’s initial prestige based on citations received and thematic relevance.
2. Normalization by Size
   Since larger journals naturally attract more citations, the raw score is then adjusted to account for how many documents the journal publishes. The final SJR2 score reflects prestige per article, allowing for fair comparisons across journals of different sizes and fields.

The first step in determining a journal’s SJR2 score involves calculating its raw scientific prestige, known as Prestige SJR2 (PSJR2). This phase mirrors the logic behind reputation systems, where prestige is not only earned by being cited, but also by who is doing the citing and how closely aligned the journals are in subject matter.

To ensure fairness from the outset, every journal starts with the same initial prestige score:

 $\mathrm{Initial\; Prestige}=\frac{1}{N}$

Where: N = the total number of journals in the Scopus dataset.

This step ensures that no journal has an unfair advantage simply due to its size, subject area, or historical status. It levels the playing field so that prestige must be earned through meaningful citations.

Once the initial prestige is set, journals start to gain or lose prestige based on how often and by whom they are cited. This step reflects a journal’s influence in the research community. Two major factors determine how prestige is transferred:

• Citation Volume
  The more times a journal is cited by others, the more prestige it receives. Citations are treated as "votes of influence" from other journals.
• Thematic Closeness
  Not all citations are equal. Journals that are closely related in subject matter transfer more prestige. This is measured using cosine similarity, a mathematical method that compares how often two journals are cited alongside each other in the same papers. The idea: if two journals are frequently co-cited, they likely cover similar topics, and thus a citation from one to the other is more meaningful.

This ensures that citations from relevant fields carry more weight than citations from unrelated disciplines.

To protect the integrity of the system and prevent abuse, two safeguards are built into the prestige distribution process:

• 50% Cap on Total Transfer
  A journal can pass on no more than 50% of its own prestige to others. This discourages artificial inflation through strategies like excessive outbound citations or citations stacking.
• 10% Cap per Citation
  No single citation can transfer more than 10% of the citing journal’s total prestige. This prevents any one link from having too much influence and helps avoid manipulation via self-citation or "citation cartels" (where journals cite each other excessively to boost metrics).

These rules ensure that prestige is earned fairly, not engineered through strategic citation behaviours.

Some journals, especially new or niche ones, may not cite others or may not receive any citations themselves. These are known as dangling nodes.

Rather than allowing these journals to act as dead ends in the prestige network, the system redistributes their prestige across all other journals in proportion to how much each journal receives from others.

This redistribution maintains the mathematical balance of the system and ensures that total prestige in the network remains constant, without being "lost" in inactive journals.

Once a journal’s raw prestige score (PSJR2) is calculated, the next step is to ensure fair comparisons between journals of different sizes. Bigger journals tend to receive more citations simply because they publish more. But does that mean they are more prestigious? Not necessarily.

To level the playing field, SJR2 applies a normalization step—adjusting for how many citable documents each journal publishes. The result is the SJR2 score: a size-independent measure that reflects how much prestige each document carries, not just the total a journal accumulates.

Normalizing by Citable Documents: This normalization ensures that journals are judged not by how much they publish, but by how influential each piece of work is.

 $\mathrm{SJR2}=\frac{\mathrm{PSJR2}}{\mathrm{Number\; of\; Citable\; Documents}}$

• SJR2 = 1.0 → Average prestige per document
• SJR2 = 2.0 → Each document has twice the average influence
• SJR2 < 1.0 → Below average, but still a recognized source

This makes the scores comparable across all disciplines and years, regardless of the journal's size or publication frequency.

A key innovation in SJR2 is the use of cosine similarity to measure thematic closeness between journals. This is a smart way to check how related two journals are, based on:

• How often they are co-cited (mentioned together) in the same reference lists
• The assumption that journals cited together are likely to cover similar topics

• More Relevant Citations Count More: A citation from a thematically similar journal carries more weight than one from an unrelated field.
• Less Noise from Unrelated Disciplines: This reduces distortion caused by cross- disciplinary citations that don’t truly reflect subject-area influence.
• Fairer Across Fields: Every discipline has different citation habits. Cosine similarity helps adjust for that, giving equitable representation across sciences, social sciences, and humanities.

SJR2 builds on and improves the original SJR method by introducing thoughtful refinements that make it more robust and fairer:

1. Thematic Relevance is Prioritized:
   Rather than treating all citations equally, SJR2 asks: "Is this citation coming from a closely related journal?" If yes, it transfers more prestige. If no, it contributes less. This makes influence context-aware, not just count-based.
2. Citation Influence is Capped to Prevent Abuse:
   To discourage manipulative practices like excessive self-citation or mutual citation pacts, a journal can’t give away more than 50% of its prestige and no single citation can account for more than 10% of its total prestige. This ensures that prestige is earned organically, not gamed.
3. Smarter Handling of 'Isolated' Journals:
   Some journals don’t cite others much or aren’t often cited themselves. These “dangling nodes” could distort the system. SJR2 solves this by redistributing their prestige proportionally to the rest of the system—based on how much others receive, not just based on output. This keeps the prestige network balanced and fair.
4. Fair for Journals of All Sizes:
   Big journals may publish more and attract more citations but that doesn’t automatically make each article more influential. By dividing prestige by output, SJR2 highlights journals that produce high-quality work, whether they publish 500 articles or 5.

• Size-Independent: Fair to journals regardless of output volume.
• Context-Aware: Weights citations based on thematic relevance.
• Discipline-Agnostic: Adjusts for differing citation behaviours across fields.
• Manipulation-Resistant: Built-in safeguards limit self-citation abuse.
• Reflects Prestige, Not Popularity: Prioritizes influence over raw numbers.

For full information please read our article: https://www.scimagojr.com/files/SJR2.pdf