XINAPI — China Research & Analysis Platform

Core Capabilities

Continuous Source Monitoring

継続的ソース監視

持续来源监控

Scheduled ingestion of 6 priority source layers — normative documents, party/state signals, legislative process, ministry implementation, military/security, and auxiliary references. Every fetch is versioned and hashed.

Structured Diff & Version Tracking

構造化差分とバージョン管理

结构化差异与版本追踪

Block-level and metadata diffs between source versions. Raw text diff, parsed block diff, and semantic diff note placeholders. Every diff references explicit version pairs.

Multilingual Entity Resolution

多言語エンティティ解決

多语言实体解析

Canonical entity IDs with native_name, japanese_name, english_name, and aliases. Language-independent identity. Designed for ja/en/zh from day one.

Evidence-Bound Claims & Relations

エビデンス紐付き主張と関係

证据绑定声明与关系

Claims and relations are extracted with traceability to source version and parsed block. No detached summaries. Evidence remains inspectable at block or span level.

Human Review Workflow

人間によるレビューワークフロー

人工审核工作流

Claims and relations go through review states (pending, accepted, rejected, revised). Only accepted_only content flows into dossiers. RBAC roles enforce review authority.

Publish-Safe Dossier Assembly

公開安全なドシエ生成

可发布安全卷宗生成

Dossiers are derived outputs referencing entities, claims, relations, and source-linked evidence. Export with classification headers and banners for downstream handling.

Priority Source Layers

L1 Normative Documents State Council, national law/regulation, national standards

L2 Party/State Signals Xinhua, People's Daily, CCP portals, Qiushi

L3 Legislative Process NPC, standing committee, legislative explanations

L4 Ministry / Regulatory MOFCOM, CAC, MIIT, NDRC, SAMR, MFA

L5 Military / Security China Military Online, PLA Daily, MND

L6 Auxiliary Reference JETRO, CISTEC, Huanqiu, universities, institutes

Analysis Pipeline

Step 1

Seed

Step 2

Ingest

Fetch, version, hash

Step 3

Diff

Block-level version comparison

Step 4

Alert

Watchlist-driven notifications

Step 5

Review

Human accept / reject / revise

Step 6

Dossier

Evidence-bound assembly

Step 7

Export

Publish-safe output

Why XINAPI

vs. Alert-Only Workflows

アラートのみのワークフローとの比較

Gap: Alert

Alert services notify but don't structure. XINAPI goes from alert through version diff, claim extraction, review, and dossier — closing the gap between notification and actionable analysis.

vs. Structure-Only Workflows

構造化のみのワークフローとの比較

Gap: Structure

Structured databases capture snapshots but miss change tracking. XINAPI tracks every version, computes diffs, and links claims to specific source changes over time.

vs. Generic Investigation Tools

汎用調査ツールとの比較

Gap: Generic

General-purpose investigation platforms lack China-specific source handling, multilingual entity resolution, and review-aware dossier assembly. XINAPI is purpose-built for China research.

Deployment Architecture

Surface	Subdomain	Purpose	Access
XINAPI Landing (you are here)	xin.tsuchiyalab.com	Product overview, entry points, documentation gateway	Public
XINAPI Platform	platform.xin.tsuchiyalab.com	Operator UI: sources, diffs, entities, dossiers, review	Authenticated
XINAPI API	api.xin.tsuchiyalab.com	Public v1 endpoints + admin behind Zero Trust	Public (v1) / Zero Trust (admin)
XINAPI Infrastructure	infra.xin.tsuchiyalab.com	Health, queue management, audit viewer	Zero Trust

Quick Start

Clone and install

git clone && pnpm install

Start API and platform

pnpm dev:api (port 8787) and pnpm dev:web (port 5173)

Seed demo sources

Navigate to Demo and run the seed action. 23 priority sources across 6 layers.

Ingest, diff, review

Trigger ingestion, view version diffs, review claims, and build your first dossier.

Follow the onboarding checklist

The Onboarding page walks through each step with progress tracking.

Continuous monitoring, structured analysis, and evidence-bound dossiers for China research