Methodology & Scoring Principles

Our platform analyzes researchers' professional output: what they publish, who they collaborate with, their funding sources, and humanitarian impact indicators. We rely exclusively on verifiable public records (OpenAlex, Crossref, ORCID, and institutional databases). We do not pass subjective moral judgments on individuals; all metrics reflect observable academic collaborations and research outputs.

Summary of Key Indicators

• Co-authorship with a countryCounts works co-authored with people at institutions in a country you choose, and works that name a funder based there. Shown as counts with DOIs, not as a score of the person. The same rule applies to every country.
• Humanitarian Impact Score (0% – 100%)Evaluates alignment with UN Sustainable Development Goals (SDGs), Open Access sharing, Global South partnerships, and public health/humanitarian crisis topics.
• Topic Similarity (SPECTER2 AI Model)Calculates cosine proximity between scientific abstract embeddings to identify researchers working on closely related topics and methodologies.
• Dual-Use & Defense Risk FlagsIdentifies ties to defense ministries, military research agencies, and topics mapped to EU Regulation 2021/821 Annex I & Wassenaar Munitions lists.

1. Co-authorship with a country

Version country-ties/1. For a country C and a researcher, from OpenAlex works:

• Co-authored work: at least one other author lists an institution in C, and the researcher's own affiliation on that paper is not in C.
• Own affiliation year: the researcher lists an institution in C on the paper (for example a visiting post). Shown as years, not counted as co-authorship.
• Funded work: the paper names a funder whose OpenAlex record is in C. If funder records cannot be read, funding is shown as unknown, not as zero.
• Window: the last 5 years including the current year, and all years. Undated works count only in all years.
• Based in C: the most recent affiliation is in C. Working in a country is not a tie, so these researchers are labelled and get no tie count in sorting.
• Sorting: by co-authored ÷ (works + 10) in the 5-year window. The 10 extra works keep a researcher with 1 of 2 papers below one with 20 of 40.
• Works: all of the researcher's works in that window, from OpenAlex yearly counts. Our index was collected through collaborations with one country, so its own paper counts would overstate the share. Until the OpenAlex total is loaded for a researcher, Browse marks the row and sorts it after the others. If our index has more papers for a researcher than OpenAlex now does, OpenAlex has moved some of them to other authors; Browse then shows no count and links to the profile. Profiles always use all works from OpenAlex.

Affiliations are as printed on each paper. A co-authorship does not show the nature of the work or anyone's position on it.

2. How the Humanitarian Score is Calculated

The Humanitarian Score reflects the public benefit, openness, and developmental impact of a researcher's scientific output. It combines four objective indicators with transparent weighting:

• SDG Alignment (35% weight): Proportion of works mapped to the 17 UN Sustainable Development Goals (such as Goal 3: Good Health and Well-being, Goal 13: Climate Action, Goal 6: Clean Water, Goal 2: Zero Hunger).
• Open Access Ratio (25% weight): Proportion of publications freely available to researchers and the public worldwide without paywalls (Gold, Green, or institutional repository copies).
• Global South / Developing Country Collaboration (20% weight): Ratio of joint publications involving researchers from lower-income and developing nations (based on World Bank and UN classifications), fostering equitable global knowledge sharing.
• Humanitarian & Crisis Response Focus (20% weight): Bibliometric detection of topics addressing public health emergencies, epidemiology, vaccines, refugee health, disaster relief, poverty reduction, and clean energy.

3. Dual-Use & Military Application Risk

We screen funding sources and institutional co-authors against standard international control lists:

• Defense and Intelligence Bodies: Agencies such as DARPA, ONR, DSTL, military academies, and defense ministries from any nation. Rules apply uniformly across all countries.
• Export-Controlled Topics: Title and topic classification based on EU Regulation (EU) 2021/821 Annex I (dual-use materials) and the Wassenaar Arrangement Munitions List (guidance systems, energetic materials, cyber-surveillance).
• Contextual Caveat: Military research bodies often fund basic open science. Findings do not prove that research was applied for military ends; they flag relationships that prospective students or collaborators may wish to explore.

4. Asynchronous Computation & Live Ingestion

Metrics are precomputed across the indexed repository of researchers and automatically refreshed:

• Catalog Ingestion: The browse index caches metrics and runs asynchronous background persistence tasks to keep scores up to date.
• Live OpenAlex Profiles: Opening an individual researcher's profile triggers a live OpenAlex query that calculates their most current Open Access, SDG, and institutional indicators, asynchronously syncing the latest data.

5. Four Kinds of Statements

• FactTaken directly from a verified public record (publication metadata, funder registry, or ORCID).
• InferenceAlgorithmic computation derived from public records (e.g., semantic similarity or keyword detection).
• FeedbackAggregated, verified reviews from former students and collaborators (displayed only with ≥ 5 reviews).
• Not knownInformation that could not be verified from public databases (e.g., unpublished funding or unrecorded years).

6. What We Do Not Collect

• Personal private emails, phone numbers, or private social media accounts.
• Photos, age, birthplaces, or family information.
• Inferences regarding political, religious, or personal beliefs.
• Arbitrary personal rating stars or subjective character judgments.

7. Corrections & Responses

If any data point or affiliation is inaccurate, researchers and collaborators can submit documentation for review. Verified researcher responses are displayed directly adjacent to profile findings.