multi-source-research

Search 13 academic and technical sources in one command.

No API keys needed for most sources. Results ranked by relevance, deduplicated, returned as structured JSON or markdown.

🇮🇹 Versione italiana disponibile in fondo

---

Sources

Source	Type	Access
arXiv	Preprints (CS, Physics, Math, AI)	Free
Semantic Scholar	Academic graph, 200M+ papers	Free
OpenAlex	Open academic metadata	Free
BASE	Bielefeld Academic Search Engine	Free
GitHub	Code repositories	Free (rate-limited)
Unpaywall	Open-access versions of paywalled papers	Free
POLITesi	Politecnico di Milano theses	Free
CrossRef	DOI metadata, citations	Free
CORE	Open research papers	Free
PubMed	Biomedical literature	Free
Internet Archive	Historical and digitized texts	Free
CNKI (partial)	Chinese academic literature	Free preview
Baidu Scholar (partial)	Chinese research	Free preview

---

Quick start

pip install requests beautifulsoup4 aiohttp

# Search across all sources
python research_agent.py "transformer attention mechanism efficiency"

# Limit sources
python research_agent.py "query" --sources arxiv semantic_scholar github

# Save results
python research_agent.py "query" --output results.json

# Top N results per source
python research_agent.py "query" --top 5

---

Output

{
  "query": "transformer attention efficiency",
  "results": [
    {
      "source": "arxiv",
      "title": "FlashAttention: Fast Memory-Efficient Exact Attention...",
      "authors": ["Dao", "Fu", "Ermon", "Rudra", "Re"],
      "year": 2022,
      "url": "https://arxiv.org/abs/2205.14135",
      "abstract": "...",
      "relevance_score": 0.94
    }
  ],
  "total": 47,
  "sources_queried": 13,
  "elapsed_sec": 4.2
}

---

Architecture

Queries run in parallel (asyncio + aiohttp). Each source has an independent adapter. Results are deduplicated by title similarity before ranking.

---

Changelog

Version	Date	Status
v1.1	2026-05-28	ACTIVE — 13 sources, async parallel, dedup
v1.0	2026-04	REPLACED by v1.1 — 8 sources, sequential
v0.1	2026-03	REPLACED by v1.0 — arXiv + Semantic Scholar only

---

Part of

TITANIUM_OS — personal cognitive OS. Full source: NODES/RESEARCH_AGENT/research_agent.py

---

Versione italiana

Cerca su 13 fonti accademiche e tecniche con un solo comando.

Nessuna API key necessaria per la maggior parte delle fonti. Risultati ordinati per rilevanza, deduplicati, restituiti come JSON o markdown.

Fonti principali: arXiv, Semantic Scholar, OpenAlex, BASE, GitHub, Unpaywall, CrossRef, CORE, PubMed, Internet Archive, POLITesi, CNKI (parziale), Baidu Scholar (parziale).

Architettura: le query girano in parallelo (asyncio + aiohttp). Ogni fonte ha un adattatore indipendente. I risultati vengono deduplicati per similarità del titolo prima del ranking.

Codice sorgente: TITANIUM_OS → NODES/RESEARCH_AGENT/research_agent.py