RecomendeMe Research
RecomendeMe Research is an observational research environment built on a platform where people share and discover human-curated recommendations. It gives universities, researchers and academic institutions access to real, human-generated data on cultural preferences, recommendation behavior and social patterns.
Unlike synthetic datasets or lab simulations, it enables empirical observation of how people recommend, discover and engage with cultural content in naturalistic settings. The platform is currently used in undergraduate research, capstone studies (TCCs) and interdisciplinary work spanning cultural studies, communication, sociology and digital humanities.
RecomendeMe Research is infrastructure for scholarly inquiry, not a commercial product. Its purpose is to support rigorous, ethical research into culture, identity and social behavior in digital contexts.
RecomendeMe Intelligence
The investigative intelligence practice of the RecomendeMe ecosystem. It combines knowledge graphs, human curation and AI to convert complex public data into structured maps, timelines and reports — for institutions, journalists, legal offices and governments.
The methodology was presented at the ITU AI for Good program of the United Nations, which recognized its potential for integration into global reports and initiatives used by governments, international institutions and justice bodies.
Pipeline
Structured public data, investigative human curation and AI-assisted analysis surface connections, timelines and patterns not visible to conventional tools. Output is delivered as structured reports, interactive maps and visual timelines for institutional, journalistic and legal use.
Case log
Brazil
Banco Master
Documentation and accessible content prepared for authorities and the public, supporting understanding of a complex financial case.
MPF
Epstein case
Deep-layer analysis that led to investigation by the MPF (Brazilian Federal Public Ministry) — a direct outcome of the methodology.
Open for partnership
RecomendeMe Intelligence is seeking strategic partners across:
RecomendeMe Intelligence operates independently of the RecomendeMe Cultural platform, which remains focused solely on culture, human recommendations and curation.
Inquiries: admin@recomendeme.com.br
Tools & repositories
Code from the RecomendeMe Intelligence practice, released as it stabilizes — investigation tooling, OSINT case management, and the GraphRAG research pipeline behind the papers in the index below.
MasterZap
Interactive archive of publicly leaked conversations and connections in the Banco Master case, built entirely from Polícia Federal disclosures and Brazilian press reporting (G1, Folha, Poder360, Metrópoles). Chat-style interface with a connection graph, inspired by the Epstein e-mail archive project.
RecomendeMe Intelligence — Case OSINT
Local-first investigative case management platform for detectives and intelligence analysts: person profiling, OSINT source tracking, relationship graph, timeline, HEXACO / Dark Triad behavioral scoring, and LGPD-compliant data handling. Runs entirely on the analyst's machine — no cloud, no external server.
GraphRAG Serialization Benchmark
Source for Paper 1: benchmarks three subgraph serialization formats (triples, Markdown, JSON) across three open-source models, run on decentralized GPU infrastructure (Nosana / Solana) so investigation data never touches a third-party cloud API.
GraphRAG — Real-World Investigation Eval
Follow-up to Paper 1: applies the same serialization pipeline and winning model (Qwen2.5-3B) to a real, dense investigation subgraph instead of a synthetic one, adding entity grounding, faithfulness and semantic alignment metrics. Findings and open issues are logged directly in the README.
All repositories are public and open source. Issues and pull requests are welcome — see each repo's README for setup and contribution notes.
Research scope
Cultural and digital identity
Observational data on how people construct and express cultural identity through recommendations — taste formation, identity signaling, and the relationship between personal recommendations and broader cultural narratives.
Human-curated recommendation systems
Insight into recommendation behavior that occurs outside algorithmic mediation: what people choose to recommend, how recommendations function as social acts, and how human curation differs from automated delivery.
Social behavior and collective trends
Aggregated data for studying collective cultural phenomena, emerging trends and shared consumption patterns — including temporal dynamics, geography and demographic variation.
Media, culture and society
Questions of canon formation, cultural gatekeeping, representation and the social functions of recommendation sharing, at the intersection of media studies, cultural analysis and sociology.
RecomendeMe as infrastructure
Not a single research project, but shared infrastructure that multiple researchers and institutions can use for distinct inquiries — a stable, ethically maintained environment for longitudinal and comparative study.
How researchers use it
- Undergraduate research — observational studies, qualitative analysis and empirical projects using platform data.
- Capstone projects (TCCs) — final-year research drawing on RecomendeMe data.
- Interdisciplinary studies — communication, sociology, cultural studies, anthropology, digital humanities.
- Datasets and documentation — anonymized, aggregated data for approved research, with methodological documentation.
- Qualitative and quantitative analysis — supports interpretive and statistical approaches alike.
RecomendeMe Research operates under principles of open science and reproducible research, with clear documentation of collection methods, ethical protocols and data structure.
Publications and studies
Peer-reviewed articles, technical reports, experimental findings and academic studies developed using RecomendeMe Research data, methodology or infrastructure.
Benchmarking Serialization Efficiency and Model Performance in Decentralized GraphRAG Pipelines
From Open Data to Actionable Intelligence: AI & Knowledge Graphs in a Real-World Brazilian Investigation
Análise Descritiva do Perfil do Usuário e Padrões de Consumo Cultural na Plataforma RecomendeMe
Algoritmos em Redes Sociais: Evolução, Mecanismos de Engajamento e Impactos nas Plataformas
RecomendeMe Feed Algorithm: A Hybrid Human-Centric Ranking Model
CULTID: Viabilidade de um Protocolo de Identidade Cultural Universal
Academic Analysis: Multimedia Recommendations and National Emotional Climate — A Hypothetical Study on the Power and Influence of RecomendeMe
RecomendeMe como Ecossistema Digital de Curadoria Cultural: Inovação Comunitária e Projeção
Updated as research projects reach completion and dissemination. Published or presented work using RecomendeMe data can be listed here — contact us.
Ethics and data use
- Privacy
- All research data is anonymized. Personal identifying information is never shared with researchers.
- Public data
- Only publicly shared recommendations — those users have explicitly chosen to make visible — are included in research datasets.
- Informed consent
- Users are informed their public data may be used for research and can opt out.
- Institutional review
- Academic collaborators are expected to obtain ethical approval from their institutions before conducting research.
- Transparency
- Collection methods, aggregation procedures and limitations are documented for all researchers.
- Responsible use
- We do not support research that could harm individuals or communities, and reserve the right to decline access requests that raise ethical concerns.
RecomendeMe Research adheres to principles established by professional associations and research ethics boards in the social sciences and humanities.
Partnerships and collaboration
We welcome collaboration with universities, research groups and individual scholars. RecomendeMe Research can support:
- Institutional partnerships for ongoing research access
- Collaborative projects with multiple research teams
- Student research programs and thesis supervision
- Methodological workshops and training
- Data sharing agreements with ethical oversight
We are especially interested in partnerships advancing understanding of cultural practices, digital behavior and social patterns through empirical observation.
For investigative intelligence partnerships — companies, journalists, legal offices and public institutions — see RecomendeMe Intelligence above.
Academic inquiries
For academic inquiries, research proposals and collaboration:
Email: admin@recomendeme.com.br
Please include:
- Your institutional affiliation
- A brief description of your research project
- Specific data or infrastructure needs
- Relevant ethical approval or review status
We typically respond within one week.