Unverified50% confidenceDecision RuleExact time
The author gave the classifier an approved type list of about 30 values, but the classifier in practice only chose 8 of them, because giving a model an open list leads it to invent plausible new types for notes that don't match.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/19/2026
First Seen
Valid until: 1/17/2027
Sources
Managing 900+ Obsidian Notes with a Knowledge Graph
rss10/19/2026
Related Entities
Related Claims
UnverifiedOnly commands starting with a known test/build prefix are eligible for replay; the test-category whitelist has 20 prefixes and the build-category has 14 prefixes.68% similarUnverifiedBy default, claude-code-setup gives only 1-2 suggestions per category, expanding to 3-5 when the user explicitly requests a certain category.61% similarUnverifiedSequence classification models are typically based on encoder architectures such as BERT or DeBERTa, with inference costs several orders of magnitude lower than GPT-4.1-mini.59% similarUnverifiedTypeScript's type narrowing allows the compiler to obtain more precise type information in specific code branches through conditional checks.58% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1069139API
curl https://kongchang.com/api/v1/knowledge/claims/1069139MCP
get_claim(id=1069139)