For Human Search - For Agent Calls

Next-Generation Knowledge Infrastructure for Humans & AI

SciBase covers open academic papers, books, patents and other knowledge assets, continuously processing cross-source data into computable, interpretable, traceable, and agent-callable AI-Ready Knowledge Objects.

SciBase knowledge infrastructure overview
01 - Overview

Knowledge Overview Across Papers, Books & Patents

SciBase is not a traditional literature retrieval system - it is an AI-Ready academic knowledge infrastructure for the LLM era, emphasizing comprehensiveness, timeliness, and agent usability.

SciBase knowledge record and discipline coverage overview
455M

Total knowledge records across papers, books, patents and more

374M

Paper records, aggregated across disciplines, languages and sources

81.06M

Book records, including modern books, rare editions and manuscripts

70M

Patent records for prior art search, tech intelligence and industry research

30.53M

AI-Ready OA full texts

2.36M

Journals / conferences / venues covered, with source and discipline filtering

Top 5 Languages

English en296M
Chinese zh48.85M
German de19.57M
French fr11.41M
Spanish es7.73M

Four Discipline Domains

Physical Sciences85.86M
Social Sciences61.61M
Health Sciences42.62M
Life Sciences34.59M
02 - Scenarios

Core Scenarios for Humans & Agents

The core value of SciBase goes beyond retrieval - it serves as the upstream knowledge data infrastructure for agents and enterprise systems.

03 - Data Foundation Layers

SciBase Five-Layer Data Architecture

From multi-source ingestion and raw retention to standard parsing, knowledge objects, evidence layers and index governance, SciBase forms a traceable, computable data foundation callable by agents.

1Multi-Source Ingestion
Ingests papers, patents, books, standards, datasets, code and other source types
2Raw Data Layer
Preserves original responses, full texts, files and collection context for traceability
3Standardization & Parsing
Unified schema, full-text structure parsing, extracting sections, figures, citations and claims
4Knowledge Objects & Evidence
Generates canonical objects (Paper, Patent, Book) and evidence spans
5Index Governance
Builds full-text, vector, graph, quality, copyright and versioning indexes
05 - Entry

Entry Points

Endpoints

API

Intelligent search API. Input natural language questions, get relevant text snippets from academic literature and trusted web pages with relevance scores. Supports full-text retrieval, vector semantic search, and hybrid.

POST /agentic-search
GET /content
GET /resource
POST /meta-search
GET /meta-catalog
View API Docs
Agent Search

Online Demo

Focused on Agent Search: ask research questions in natural language, get papers, patents, books and evidence spans with source provenance, quality signals and traceable links.

Ask research questions
Inspect evidence spans
Trace source provenance
Try Online Demo
Email

Contact Us

For API trials, data collaboration, agent scenario co-development and enterprise knowledge infrastructure integration, reach out via email.

Contact UsSend Email