Academic Intelligence (ACADINT): Intelligence Discipline Guide

Research output is the earliest public indicator of national and corporate capability. A capability appears in a paper years before it appears in a product, and long before it appears in a threat report.

What Academic Intelligence is as a discipline

Academic intelligence is the exploitation of scholarly output, funding records and institutional affiliation data to map who can do what, with whom, and with whose money. It treats publications, preprints, grants, conference programmes, dissertations and lab web presence as a structured record of expertise. The discipline builds capability maps rather than reading lists: it identifies which institutions hold rare technical skills, which individuals are the actual centre of a subfield rather than the most senior author, and how research groups connect across borders through co-authorship, joint appointments and funding.

Sub-methods include co-authorship and citation network analysis, affiliation disambiguation, funding acknowledgement mining, bibliometric trend detection, and conference and patent-to-paper bridging. It sits at the front of the intelligence cycle as an indications and warning source, and at the back as a validation source: independent published work is often the only way to confirm or refute a technical claim made by a vendor or an adversary.

Why it matters

Only academic intelligence answers where a specific technical capability physically resides and who holds it personally. It answers whether a capability is broadly distributed or concentrated in a handful of labs, when a field crossed from theory to demonstration, which foreign entity is co-funding sensitive work, and whether an individual's declared affiliation matches their actual output. It also provides expert identification for due diligence, litigation support and technical validation, and gives early warning on dual-use research well upstream of procurement.

What analysts actually look for

These are the concrete, observable signals that carry weight in this area of work:

Co-authorship networks revealing collaboration between institutions that have no public commercial or governmental relationship
Funding acknowledgement lines naming grant numbers, ministries, defence agencies or corporate sponsors behind ostensibly academic work
Undeclared or shifting dual affiliations, where an author lists different institutions across contemporaneous papers
Sudden publication volume growth by a single institution in a narrow technical area, indicating deliberate capability investment
Equipment and methods sections that disclose specific instrumentation, materials and facility capability held by a lab
Dissertation and thesis topics identifying the pipeline of specialists being trained and their supervisors
Retractions, corrections and integrity findings that undermine a claimed capability or an expert's reliability
Conference programme and keynote patterns showing which groups are recognised as leading by their own field

Where the data comes from

Authoritative and openly available collection points. Always confirm licensing and terms before operational or commercial use:

OpenAlex — Free open catalogue of works, authors, institutions and funders with an API suitable for network and trend analysis
Semantic Scholar — Free corpus and API with citation contexts, influential citation counts and author disambiguation
PubMed and PubMed Central — Biomedical and life science literature with full text for many articles, essential for bio capability mapping
arXiv and other preprint servers — Earliest public disclosure of results in physics, computing and mathematics, often months ahead of journals
ORCID and ROR — Persistent identifiers for researchers and organisations, the backbone of reliable author and affiliation disambiguation
Crossref — Metadata for scholarly publications including funder registry data and update and retraction notices
NIH RePORTER and EU CORDIS — Awarded grant records with abstracts, amounts, principal investigators and participating organisations
Retraction Watch database — Documented retractions and reasons, used to weight the reliability of claimed results and researchers

A working method

A repeatable sequence beats ad-hoc searching. This is a practical starting workflow:

Define the capability, not the keyword — Describe the technical capability in terms of methods and measurable outcomes, then derive search terms, since field vocabulary shifts and keywords alone miss adjacent work.
Assemble the corpus — Pull works from at least two independent indexes to reduce coverage bias, deduplicating on DOI and normalising author and institution identifiers.
Disambiguate entities — Resolve authors via ORCID and institutions via ROR before any counting, because raw name strings systematically inflate and split output.
Map the network — Build co-authorship and citation graphs to identify central individuals, bridging authors between clusters, and cross-border collaboration edges.
Mine funding and methods — Extract acknowledgement and methods text to attribute sponsorship and to inventory the instrumentation and materials a lab demonstrably holds.
Corroborate off-corpus — Check patents, conference talks, procurement notices and lab web archives to confirm that publication activity reflects real operating capability.
Trend and warn — Track publication and citation velocity over time to flag fields moving from theory to demonstration, and brief the change rather than the snapshot.

How this connects across the intelligence taxonomy

Intelligence work does not respect neat boundaries. The mission domain you are working, the disciplines you practise, and the data points you pivot on are one connected system. These are the direct relationships for this entry — every link is also a tag, so you can follow any thread across the whole library.

Applied in these mission domains

Operates on these data points

Company / Organization — A legal entity — corporation, LLC, NGO, or business.
Code Repository — A source-code repository — leaks secrets, reveals developers, and anchors supply-chain risk.
Patent — An intellectual property filing granting invention rights.
Keyword / Narrative — A search term, topic, hashtag, or narrative tracked across media and platforms.
Person / Name — A named individual — the subject of identity resolution and profiling.
Facility / Site — A physical installation — plant, base, port, data centre — with a fixed footprint and function.

Related disciplines

Patent Intelligence — Patents, Filings, and Innovation Signals
Technical Intelligence — Technology Capability, Design, and Exploitation

Inside the platform: where Academic Intelligence lives

The Quantus platform is 204 pages behind a 147-item sidebar organised into six working groups: Command (24 items), Dashboards (15), Threat Theaters (14), Intelligence Domains (15), Investigate (34), and Administration (45). This entry is not a page in isolation — it is a thread running through several of them.

The modules that matter most here:

discipline.php?d=ACADINT — Discipline hub
source-catalog.php?disc=ACADINT — Source catalogue filtered to this discipline
search.php — Company / Organization profile
correlate.php — Correlation graph
cases.php — Case management

Each dashboard is local-first: it renders from the platform’s own database rather than depending on a live third-party call, so it still works when an upstream API is unreachable or rate-limited. Heavy aggregates are cached with a hard query time cap and degrade to the last good value instead of hanging the page.

Automation, playbooks and AI skills

Analysis that only happens when someone remembers to run it is not a capability. The platform ships a 30-step automation pipeline (cron.php) that collects, ingests, resolves, enriches, correlates and scores on a schedule — 25 seeders, 11 resolvers and 7 enrichment runners, all idempotent and cursor-based so a run can be interrupted and resumed without duplicating or losing work.

AI skills that apply

The 16 one-click operations in ai-skills.php are deterministic jobs, not free-text generation. The ones that matter here:

Threat Hunt
Auto-Collect Feeds
Enrichment → Local
Summarise (Copilot)
Generate Report

Alerting closes the loop: rules in alerts.php fire on new indicators matching a saved query, so a first sighting in this area raises a notification rather than waiting to be noticed at the next review.

Feeds, data sources and the API

The collection layer runs a feed registry of free, machine-readable sources — bulk blocklists and trackers (Maltrail, IPsum, FireHOL, the full abuse.ch corpora, phishing databases, Emerging Threats, Spamhaus, DigitalSide, ThreatView), authoritative government feeds (CISA KEV, OFAC, UN and EU sanctions lists), and reference datasets (RIR allocations, ip-to-ASN and geolocation tables, MITRE ATT&CK, EPSS). collect.php pulls them server-side on a schedule; feeds.php and source-catalog.php show what is registered, what it covers and when it last ran.

Anything the platform holds is reachable programmatically. The REST API in api.php exposes 11 endpoints — status, stats, search, lookup, recent, export, bulk_check, top_threats, by_category, categories, check — and export.php streams 18 formats in bounded chunks, so a million-row export neither exhausts memory nor times out:

STIX 2.1, MISP, OpenIOC 1.1, CEF (ArcSight), LEEF 2.0 (QRadar), Zeek/Bro intel, Snort/Suricata rules, Palo Alto EDL, BIND RPZ, hosts blackhole, iptables, CSV, JSON, NDJSON/JSONL, XML.

That covers the CTI standards (STIX 2.1, MISP, OpenIOC), SIEM ingestion (CEF, LEEF, Zeek), detection engines (Snort/Suricata), and direct enforcement (Palo Alto EDL, BIND RPZ, hosts, iptables) — so intelligence developed here can be actioned in the tools you already run, without a manual reformatting step. A TAXII 2.1 server and a MISP/RSS feed are also served for pull-based sharing.

Use cases

Three ways this entry earns its keep in day-to-day work:

Triage under time pressure. An artifact or report lands and you need a defensible read in minutes, not days. Define the capability, not the keyword is the first move; the platform pre-computes the enrichment so the analyst spends the time on judgement rather than lookups.
Building the picture. A single indicator is rarely the story. Disambiguate entities turns one artifact into a network — shared infrastructure, repeated selectors, the same operator behind different names — via the correlation graph and the cross-entity link engine.
Producing something actionable. Analysis that ends in a document nobody can use is wasted. Trend and warn feeds the case file, the detection rule, the block list or the referral — with sourcing attached so the recipient can verify it.

Case management (cases.php), watchlists, saved searches and scheduled reports mean the work persists between sessions and survives an analyst leaving the team.

How each sector uses Academic Intelligence

The same entry is worked very differently depending on who you are, what authority you hold, and what you are ultimately producing. A military analyst is supporting a commander’s decision; a journalist is meeting a publication standard; an NGO caseworker is protecting a person. The underlying artifacts are shared — the constraints, outputs and thresholds are not.

🎖 Military and defence

Defence science and technology staff use academic intelligence to answer whether an adversary can field a capability and when. Publication and conference activity gives years of warning ahead of procurement, and co-authorship networks show whether a capability sits in one laboratory or across a national programme. Products feed technology watch, capability assessment and research security decisions about which foreign collaborations to permit within defence-funded programmes. Constraints matter: analysis must target institutions, funders and demonstrated capability rather than the nationality or ethnicity of researchers, and published claims require technical validation before they enter a capability estimate that will drive investment.

🕵 National intelligence

National intelligence uses scholarly output as an early indications and warning source and as an independent validation layer against denial and deception. Requirements typically ask where a specific technical skill physically resides, who personally holds it, and which foreign entities co-fund sensitive work. Because collection is open, findings can be shared with allies, regulators and universities without exposing sources, which makes this discipline unusually useful for protective outreach. Handling still matters: profiles of named academics are personal data, and products should be caveated so a bibliometric signal is not read as evidence of wrongdoing by an individual.

👮 Law enforcement

Law enforcement encounters academic intelligence in export control, research security and economic espionage cases. Publication records establish what a researcher demonstrably worked on, funding acknowledgements establish who paid, and affiliation histories establish where controlled knowledge may have travelled. Evidentially, bibliographic metadata is weak on its own: it shows association, not transfer. A charging decision needs the underlying grant agreements, employment contracts, disclosure forms, communications and travel records obtained under production order or warrant. Investigators should also guard against inferring intent from collaboration patterns entirely normal in international science, a recurring source of failed prosecutions.

🔍 Private investigation and corporate security

Corporate teams use academic intelligence for technology scouting, expert identification, competitor research programme mapping and pre-hire verification of technical claims. It supports litigation by finding qualified experts and by locating prior art or independent replication of a disputed technical assertion. It also supports insider risk work when an employee's external publication activity conflicts with confidentiality obligations. What a private actor may not do is profile individuals on the basis of nationality, obtain non-public institutional records through pretext, or use bibliometric patterns as a screening criterion for employment decisions, which creates discrimination exposure in most jurisdictions.

📰 Journalism and OSINT media

Journalists use scholarly records to verify technical claims, identify independent experts and trace funding behind advocacy science. Verification standards should include checking whether a paper is peer reviewed or a preprint, whether it has been retracted or corrected, whether the journal is legitimate, and whether the finding has been replicated. Conflict of interest and funding disclosures deserve as much attention as results. When reporting on individual researchers, offer a right of reply and take care with early career academics for whom exposure carries disproportionate professional risk. Do not treat citation counts as a proxy for correctness.

🌍 NGO, humanitarian and human rights

Research and advocacy organisations use academic intelligence to identify independent expertise, map who funds contested research, and document dual-use or ethically problematic work for accountability. Do no harm applies directly: naming individual researchers in repressive jurisdictions can expose them to state action, and collaboration mapping can be misread as an accusation. Documentation for accountability should focus on institutions, funders and published claims, with individual attribution used only where necessary and justified. Duty of care extends to local academic partners, who may face consequences for association with an investigation long after publication.

🎓 University and research

For universities themselves this is research security and science-of-science work. Methodology centres on entity disambiguation, coverage bias between indexes, and the difference between output volume and capability. Any study profiling identifiable researchers needs ethics committee approval, and data-sharing agreements with index providers usually restrict redistribution of full records. Reproducibility requires publishing query strings, index version, snapshot date and disambiguation rules, since author and institution resolution decisions drive results more than the analysis that follows. Cite datasets with DOI and version, and deposit derived aggregated data rather than raw personal records.

Playbook: working Academic Intelligence end to end

A repeatable sequence, from the moment the requirement lands to the moment a product is delivered and the case is closed out. Each phase states what you are trying to establish, not merely what to click — the point is a defensible chain of reasoning, not a checklist.

Phase 1 — Define the capability, not the keyword

Write the technical capability as methods, materials, instruments and measurable outcomes before choosing search terms. Field vocabulary shifts across communities and languages, and keyword-first searching systematically misses adjacent work that solves the same problem differently. A good output is a capability statement plus a term set including synonyms, deprecated terminology, instrument names and relevant classification codes. Stop when a domain expert reviewing the term set cannot name an obvious missing route into the literature.

Phase 2 — Choose indexes and document coverage

Select at least two independent indexes with different ingestion policies, because coverage bias in a single source shapes every downstream count. Record each index's version, snapshot date and known gaps such as language, discipline or preprint coverage. A good output is a coverage note a reader can use to judge which conclusions are robust. Stop adding indexes when marginal new records fall below a few percent and are not concentrated in a relevant language or subfield.

Phase 3 — Assemble and deduplicate the corpus

Retrieve records, deduplicate on DOI where present and on normalised title, author and year where it is not, and preserve the source index for every record. Retain retracted and corrected items with their status flags rather than dropping them silently. A good output is a corpus table with stable identifiers and provenance per record. Stop when duplicate resolution is stable, and document the residual duplicate rate rather than claiming it is zero.

Phase 4 — Disambiguate authors and institutions

Resolve authors through ORCID and index-native identifiers, and institutions through ROR, before any counting. Raw name strings both split one person into several and merge several into one, especially for common names transliterated from non-Latin scripts, and this error dominates results. A good output is an entity table with resolution confidence per record and a documented rule set. Stop when a manual sample of a hundred records shows resolution error below your stated tolerance.

Phase 5 — Map the collaboration network

Build co-authorship and citation graphs with dated edges. Identify structurally central individuals, bridging authors who connect otherwise separate clusters, and cross-border edges into the subject institution. Bridges matter more than hubs for capability transfer questions. A good output is a network with communities labelled by topic and an explicit note on how last-author and consortium conventions in the field affect interpretation. Stop when adding a further citation generation stops changing community structure.

Phase 6 — Mine funding and acknowledgements

Extract funder names, grant numbers and acknowledgement text, then reconcile against grant databases to attribute sponsorship accurately. Acknowledgements also name facilities, instruments and collaborators that never appear in the author list. A good output is a funder-to-output table with grant identifiers where available and unresolved funders flagged. Stop when the major funders of the cluster are identified, noting that non-disclosure of funding is itself a reportable finding.

Phase 7 — Inventory demonstrated capability

Read methods sections and supplementary material to inventory what a laboratory demonstrably operates: instruments, materials, computational resources, sample access and safety containment level. This converts publication activity into physical capability, which is what the customer actually asked about. A good output is a per-institution capability inventory with the paper citation supporting each item. Stop when the inventory covers the equipment that gates the capability in question.

Phase 8 — Bridge to patents and products

Match key authors and institutions to patent applications, spin-out companies, standards contributions and procurement notices. The paper-to-patent bridge shows when work has moved from investigation towards application, the transition most customers care about. A good output is a timeline showing publication, filing and commercialisation events for the cluster. Stop when the bridge is established or shown to be absent, since absence of filings is itself informative about intent.

Phase 9 — Assess trajectory, not snapshot

Measure publication and citation velocity, entry of new groups, shifts from theory to demonstration papers, and changes in the proportion of applied venues. Normalise against field-wide growth so you do not report the expansion of science as a capability signal. A good output is a trend statement with the normalisation method stated. Stop when the trend is stable enough to brief, and re-run on a fixed cadence rather than continuously.

Phase 10 — Validate off-corpus

Confirm that publication activity reflects real operating capability using archived laboratory web pages, conference programmes, dissertation records, equipment procurement notices and imagery of facilities where relevant. Groups sometimes publish above their true capability, and classified or commercial programmes publish far below it. A good output states, for each key claim, whether independent corroboration exists. Stop when the highest-consequence claims are corroborated or explicitly marked as uncorroborated.

Phase 11 — Apply the ethics and proportionality check

Before writing, review whether individual-level findings are necessary for the question, whether nationality or ethnicity has crept in as a proxy, and whether naming a researcher creates personal risk out of proportion to the analytical gain. Convert individual findings to institutional findings where possible. A good output is a documented decision on individual attribution. Stop and escalate if the product would function as an accusation against a named person without evidence of misconduct.

Phase 12 — Report with methods and limits attached

Deliver the judgement with the corpus definition, index versions, snapshot dates, disambiguation approach and known coverage gaps attached, so a reader can assess and reproduce it. State plainly which conclusions are robust to coverage bias and which are not. A good output can be rerun by a different analyst six months later and produce comparable numbers. Stop before presenting counts without denominators, which is the fastest way to mislead a non-specialist customer.

The platform ships this as a step-checked workflow in playbooks.php, so progress is recorded against a case rather than held in someone’s head.

Source register: what to collect from, and how

Sources are listed with their access model so you can plan around cost and licensing before you build a dependency on them. Open means no account required; registration means a free account or API key; licensed means paid or institutional access. Always confirm current terms — licensing changes, and a source that was free for research may not be free for commercial or evidential use.

Source	Access	What it gives you	How it is used here
OpenAlex	Open	Open catalogue of works, authors, institutions, sources, funders and concepts with a full API and periodic bulk snapshots	Primary corpus assembly and network analysis, with ROR institution identifiers already attached to most records
Semantic Scholar	Open	Large multidisciplinary corpus with citation contexts, influence flags, embeddings and author disambiguation, exposed through an API	Second independent index for coverage checks and citation network work, and for finding methodologically related work
Crossref	Open	DOI registration metadata including funder registry data, licences, retraction linkage and deposited reference lists	Canonical deduplication key and the reliable route to funder attribution where publishers deposit it
PubMed and PubMed Central	Open	Biomedical and life sciences citations with MeSH indexing, plus full text for deposited open access articles	Methods and materials mining for biological capability assessment, including containment level identification
arXiv	Open	Preprint server for physics, mathematics, computer science, quantitative biology and related fields with daily postings	Earliest public signal in fast-moving fields, typically months to years ahead of journal publication
ORCID	Open	Persistent researcher identifiers with self-asserted affiliation, employment, education and works records	Author disambiguation and detection of undisclosed dual affiliations across institutions and countries
Research Organization Registry	Open	Open persistent identifiers for research organisations with relationships between parent, child and successor entities	Institution disambiguation, resolving campus, hospital and institute variants to one parent entity
NIH RePORTER	Open	US federal biomedical grant awards with abstracts, investigators, institutions, funding amounts and linked publications	Attributes funding to named projects and reveals programme scale before results are published
CORDIS	Open	EU framework programme projects with participants, budgets, coordinators and reported results and deliverables	Maps European consortium membership and budgets, exposing cross-border collaboration structures directly and by name
NSF Award Search	Open	US National Science Foundation awards with abstracts, investigators, institutions, amounts and programme codes	Identifies funded work in physical sciences and engineering ahead of publication
Retraction Watch database	Open	Structured records of retractions, expressions of concern and corrections with stated reasons and dates	Integrity screening of the corpus and detection of misconduct clusters within a group or journal
OpenAIRE	Open	European aggregation of publications, datasets, software and projects linked to funding streams	Links outputs to European funding where publisher metadata is incomplete, and surfaces datasets and software
DBLP computer science bibliography	Open	Curated bibliography of computing publications with strong author disambiguation and full conference coverage	Authoritative for computing fields where conferences, not journals, carry the significant results
Networked Digital Library of Theses and Dissertations	Open	Aggregated global index of doctoral and masters theses with institutional and subject metadata	Reveals training pipelines and unpublished experimental detail produced by a laboratory's students
Directory of Open Access Journals	Open	Curated index of vetted open access journals with editorial policy, peer review and licensing information	Venue legitimacy screening before treating a publication record as evidence of capability
Web of Science or Scopus	Licensed	Curated commercial citation indexes with normalised affiliations, subject categories and field-normalised metrics	Benchmarking and field normalisation where comparability with published bibliometric studies is required

Prefer sources that publish a methodology and a revision history. A dataset that changes silently is a liability in any product that has to survive challenge.

Tooling

Tools commonly used against Academic Intelligence. None of these replace judgement, and each carries its own failure modes — know what a tool infers versus what it observes.

VOSviewer — Builds and visualises co-authorship, co-citation and term maps from bibliographic exports; fast and readable, but clustering parameters strongly affect the story it tells.
Bibliometrix and biblioshiny in R — Scripted bibliometric analysis with reproducible workflows; excellent for repeatable studies, though it does little disambiguation itself and needs clean input.
CiteSpace — Detects emerging research fronts and citation bursts over time; useful for trajectory work, with an interface and parameter set that reward experience.
OpenAlex API and pyalex — Programmatic corpus assembly with institution and funder identifiers attached; coverage is broad but metadata quality varies by publisher deposit practice.
OpenRefine — Reconciles messy affiliation and funder strings against ROR and the funder registry; effective at scale but reconciliation decisions still need manual review.
Gephi or igraph — Network metrics and layout for collaboration graphs; powerful, though centrality measures need care where field authorship conventions differ.
GROBID — Extracts structured metadata, references and acknowledgement sections from PDFs; strong on well-formatted articles, weaker on scanned and non-English documents.
Zotero group libraries — Collects and shares references with notes and retrieval dates; suited to team corpora, not to large-scale quantitative work.
Wayback Machine and archiving tooling — Preserves laboratory pages, staff lists and equipment descriptions that are frequently removed; coverage of dynamic and non-English sites is uneven.

AI skills and automation in detail

These are deterministic jobs with defined inputs and outputs, not open-ended prompting. Each is idempotent and cursor-based: interrupt one and it resumes where it stopped rather than duplicating work or losing progress.

Threat Hunt — Runs saved hypotheses against the corpus and surfaces what matches, with the query preserved as a versioned artifact.
Auto-Collect Feeds — Pulls the registered feed set server-side on a schedule, recording per-feed status so a silently dead feed is visible.
Enrichment → Local — Materialises enrichment into the local store so dashboards render from your own database instead of a live third-party call.
Summarise (Copilot) — Produces a narrative summary beside the underlying records. It explains; it never creates indicators or assigns attribution.
Generate Report — Assembles a sourced product from the current case or query, with provenance attached to each element.

A note on the boundary: the only skill that involves a language model is Summarise (Copilot), and it writes prose about records that already exist. Nothing else on this list involves generation of any kind. No indicator, relationship or attribution in the platform originates from a model. See the full skill list.

Tradecraft notes

The distinctions that separate a competent analyst from a fast one:

Disambiguation is the analysis. Most bibliometric errors are entity errors made before any counting starts, and they systematically penalise common surnames and transliterated names. Resolve to identifiers first and report your resolution error rate rather than implying it is zero.
Volume is not capability. A group can publish prolifically on theory while lacking the instruments to demonstrate anything. Read methods sections for equipment, materials and containment, because the physical inventory is the real capability signal.
Bridging authors matter more than central ones. The person connecting two otherwise separate clusters is where knowledge actually moves between communities and countries. Betweenness beats degree for any question about transfer rather than prestige.
Silence can be the strongest signal. When a productive group stops publishing while its funding and staffing grow, the plausible explanations include classification, commercialisation or a shift to applied work, and each is worth more than another year of papers.
Normalise before you compare. Publication counts have grown across nearly all fields, so an apparent surge often reflects field growth, index expansion or new journal coverage. Always present a field baseline alongside the subject trend.
Acknowledgements outperform author lists. Facilities, instruments, sample providers, computing allocations and unnamed collaborators appear there and nowhere else, and they frequently identify the real dependency in a research programme.
Guard against nationality as a proxy. Research security work that drifts into flagging people by origin is both ethically wrong and analytically poor, because it produces enormous false positive volume. Target funders, institutions, undisclosed affiliations and controlled technology instead.

Measuring whether it is working

Capability claims should be falsifiable. These are the measures that show whether work on Academic Intelligence is producing anything, and they are worth baselining before you change process or tooling.

Lead time between the first publication signalling a capability and its confirmation through procurement, patents or product release, tracked per case to show whether warning is genuinely early.
Entity resolution error rate measured on a manually verified sample, reported separately for author and institution resolution.
Proportion of capability claims in delivered products carrying independent off-corpus corroboration such as procurement, patent or imagery evidence.
Coverage overlap between the primary and secondary index for each corpus, used as a running check on collection bias.
Share of products where individual-level attribution was reviewed and either justified or converted to institutional attribution.
Reproducibility rate measured by whether a second analyst rerunning the documented query obtains counts within an agreed tolerance.
Number of research security decisions, such as a collaboration review or an export control referral, informed by an academic intelligence product.

Beware of measuring volume alone. Indicator counts and report counts rise easily and say little; time-to-attribution, proportion of findings that survive review, and how often a product changed a decision say a great deal.

Common pitfalls

Counting publications as capability. Output volume can reflect publication incentives or paper mills rather than genuine technical ability
Author name collisions in transliterated names, which merge or split researchers and destroy network analysis at scale
Index coverage bias: no single database covers all languages, regional journals or conference proceedings evenly
Assuming affiliation equals location or loyalty; visiting positions, honorary titles and joint labs all break that assumption
Reading collaboration as intent when many co-authorships are consortium artefacts with no substantive working relationship
Attributing an individual's research to a nationality or ethnicity, which is both analytically wrong and ethically indefensible

Legal and ethical considerations

Bibliographic metadata is generally free to use, but full text is usually copyrighted and licensing terms restrict bulk redistribution; work from open access corpora and cite rather than republish. Profiling named academics engages data protection law, and analysis must target institutions and capabilities rather than individuals' nationality or ethnicity. Research security work should follow published export control and dual-use guidance rather than informal suspicion, and findings about individuals should be handled as sensitive personal data with a documented lawful basis.

Data integrity: no fabrication, no drift, no hallucination

Intelligence that cannot be traced back to a source is not intelligence, it is assertion. Everything in this entry — and everything in the platform behind it — is built on a small number of non-negotiable rules.

Provenance on every record

Every indicator carries the source that supplied it, a first-seen and last-seen timestamp, and a sighting count. Where several feeds report the same artifact, each contribution is recorded separately rather than collapsed, so you can see whether a finding rests on one source or twelve. Source attribution travels with the data into every export, so a recipient can audit a claim without asking you for the working.

Nothing is invented to fill a gap

If the platform has no data for Academic Intelligence, it says so. Empty is displayed as empty — never padded with plausible-looking placeholder values, sample records or illustrative examples that a reader might mistake for observations. A dashboard with no rows is a true statement about collection coverage, and it is treated as a gap to close, not a blemish to hide.

Scoring is deterministic and reproducible

Threat scores, reputation grades and risk tiers are computed from stated inputs with fixed weights, not estimated. The same inputs always produce the same output, and the formula is visible rather than a black box. Aggregates are cached with an explicit time-to-live so a figure on screen is never silently stale — and when a heavy query exceeds its time budget the platform serves the last known-good value and labels it, rather than inventing a fresh number or hanging.

Where AI is used, and where it is not

Language models summarise and explain. They do not create indicators, assign attribution or manufacture relationships. No IP address, wallet, hash or identity in the platform originates from a model — every one is ingested from a named feed, resolved from a reference dataset, or entered by an analyst with a source recorded. Copilot output is presented as narrative alongside the underlying records, never in place of them, so a reader can always check the summary against the evidence.

Guarding against drift

Enrichment is additive and timestamped rather than overwriting. Reference data — sanctions lists, allocations, taxonomies — is re-synchronised from the authority on a schedule instead of being edited in place, so local copies cannot quietly diverge from the source of truth. Attribution is recorded with a confidence level and the reporting it rests on, and inferred relationships are labelled as inferred. When a source retracts or corrects, the correction propagates rather than leaving a stale assertion behind.

What this means for you

You can put a finding from this platform in front of a regulator, a court, a board or a partner agency and show where each element came from. That is the standard the tooling is built to — because in this work, being confidently wrong is more damaging than being usefully uncertain.

By the numbers

The taxonomy this entry belongs to is not a marketing list — it is the actual structure of the platform: 52 mission domains, 52 intelligence disciplines and 65 data points, each with a live dashboard behind it. Supporting that: 18 indicator types, 14 playbooks, 16 AI skills, 18 export formats and a 30-step automated pipeline.

This particular entry connects directly to 6 data points, 4 mission domains, 2 closely related entries — every one of them a tag you can follow, and a dashboard you can open.

Questions analysts actually ask

Which index should I use as my primary source?

Use OpenAlex as the backbone because it is open, has an API, carries ROR and ORCID identifiers, and can be snapshotted for reproducibility. Add Semantic Scholar or a licensed index as a second view to measure coverage bias, and add a domain index where the field demands it: PubMed for life sciences, arXiv for physics and computing preprints, DBLP for computer science conferences. Never rely on a single index for a count you intend to publish. Report which index produced which number, with the snapshot date, since coverage changes materially between releases.

How do I tell whether a laboratory actually has a capability?

Read the methods and supplementary material, not the abstract. Look for named instruments with model numbers, materials with suppliers, computational resources, sample provenance and biosafety containment level. Then corroborate off-corpus: equipment procurement notices, facility images, dissertation acknowledgements, conference demonstrations and staff pages archived over time. Publication of a result using a shared national facility tells you the group had access, not that it holds the instrument. The distinction between access and possession is often the whole question for a capability assessment.

Is preprint activity reliable intelligence?

It is the earliest signal and genuinely useful, but it is unreviewed. Treat a preprint as a claim by the authors, weight it by the group's track record, and check whether it later appeared in a reviewed venue and whether the results changed. In fast-moving computational fields the preprint is effectively the publication of record and the reviewed version adds little. In biomedicine the gap matters much more. Always mark preprint status explicitly in the product, because customers routinely read a citation as peer reviewed.

How should research security work avoid discriminating against individuals?

Anchor the analysis on conduct and objects rather than identity: undisclosed affiliations, undisclosed funding, transfer of controlled technology, contractual breaches, and institutions or funders subject to restrictions. Do not use nationality, ethnicity or place of education as a risk indicator, which is unlawful in many jurisdictions and analytically worthless given the false positive volume. Apply proportionality: name individuals only where necessary for the decision, escalate through the university research office or a designated authority rather than acting informally, and document the basis for every individual-level finding.

What do citation counts actually tell me?

They measure attention within a citing community over time, filtered by index coverage and field norms. They do not measure correctness, importance or capability. Citation rates vary by an order of magnitude across fields, take years to accumulate, and are inflated by review articles, methods papers and self-citation. For intelligence purposes, citation structure is far more useful than citation count: who cites whom reveals which groups are following whose work, and a burst of citations from an applied community signals that a result has moved towards implementation.

How do I handle non-English and grey literature?

Accept that your corpus is biased and say so. Major indexes under-cover Chinese, Russian, Japanese, Korean and Persian language venues, national journals, institutional technical reports and conference proceedings. Use national indexes where licensing permits, work with a language-capable analyst rather than relying on machine translation for technical content, and mine reference lists of English-language papers by the same groups, which frequently cite their own domestic publications. State the language coverage limit prominently, because it usually bounds the confidence of any capability conclusion.

Do I need ethics approval for bibliometric research?

If the work is aggregate and uses published metadata, most institutional review boards treat it as exempt or not human subjects research. If you profile identifiable individuals, link across datasets to build personal histories, or draw conclusions about specific named researchers, that is human subjects research in most frameworks and needs review. Data protection law applies independently of ethics review in the EU and UK, since researcher names and affiliations are personal data. When in doubt, ask the review board and document the answer, particularly for work commissioned by government or industry.

Standards, frameworks and further reading

Work that references a recognised framework is easier to defend, easier to hand over, and easier for a partner to consume:

Leiden Manifesto for research metrics, setting principles for responsible use of bibliometrics including field normalisation and protection against gaming.
San Francisco Declaration on Research Assessment, which rejects journal impact factor as a proxy for the quality of individual research or researchers.
FAIR data principles, requiring findable, accessible, interoperable and reusable research data and applicable to derived bibliometric datasets.
CRediT contributor roles taxonomy, which structures authorship contribution claims and helps distinguish who actually performed the work.
COPE guidelines on publication ethics, governing retraction, correction and expression of concern and therefore the integrity flags in a corpus.
Export control regimes including the EU dual-use regulation, the US fundamental research exclusion and Wassenaar Arrangement lists, which define when published research becomes controlled.
National research security policy such as US NSPM-33 implementation requirements and UK Trusted Research guidance, governing disclosure of foreign affiliations and funding.
Regulation (EU) 2016/679 and equivalent data protection law, which applies to profiling of named researchers regardless of the public availability of the underlying records.

References

Primary sources and authoritative references for this entry. Publishers revise and retire material, so treat the retrieval date as part of the citation and re-check before relying on any of it in a formal product.

OpenAlex open catalogue of scholarly works — OurResearch. Fully open bibliographic dataset with author, institution and funder identifiers and an API
Semantic Scholar Academic Graph — Allen Institute for AI. Open research corpus with citation contexts and author disambiguation used for network analysis
Crossref metadata and funder registry — Crossref. DOI registration metadata including funding acknowledgement and retraction linkage
Research Organization Registry — ROR community initiative. Open identifiers and hierarchy for research institutions, the basis of affiliation disambiguation
ORCID researcher identifier registry — ORCID. Persistent researcher identifiers with affiliation and works records used for author resolution
San Francisco Declaration on Research Assessment — DORA. International statement on appropriate and inappropriate uses of research metrics
Trusted Research guidance — UK National Protective Security Authority. National guidance for universities on protecting research from hostile interference while sustaining collaboration
CORDIS EU research projects database — European Commission. Record of EU-funded projects, consortium members and reported results
NIH RePORTER grant reporting system — US National Institutes of Health. Searchable record of federally funded biomedical projects and linked publications
Retraction Watch database — Center for Scientific Integrity. Structured record of retractions and expressions of concern used for corpus integrity screening

Link integrity: every reference above was verified with a live request when this page was generated. Where a publisher had moved or withdrawn a document, the link was repointed at a preserved copy in the Internet Archive and marked as archived. Anything with no reachable copy anywhere had its link removed rather than left to rot — the source is still credited, it simply cannot be linked.

Put it into practice

The Quantus Intel threat intelligence platform operationalises this entry: maps expertise, funding and collaboration networks and links researchers to entities already in your graph. Explore the platform, or browse the rest of the library by following any tag above.