Cultural semantic modeling
Cross-language aesthetic concepts — wabi-sabi, quiet luxury, gorpcore, Y2K — become structured, computable semantic units.
TASTE KNOWLEDGE GRAPH
Taste Knowledge Graph — answering why people like it
Pinweima builds a knowledge graph of brands, culture, products and users, translating tacit aesthetics and lifestyles into a relational network machines can reason with and businesses can use.
Four entity typesRelational reasoningMultilingual culture
What is a Taste Knowledge Graph
A conventional product graph answers “what is this?” A taste knowledge graph answers “why would someone love it?”
The graph takes brands, culture, products and users as its nodes, and taste relations — style, lineage, scene, emotion and identity — as its edges, continuously mapping the deep structure of Asian consumer culture.
- Brands: positioning, style lineages and the taste distance between brands
- Culture: aesthetic motifs, lifestyles, regional and subcultural lineages
- Products: design language, materials and craft, price bands and use scenes
- Users: taste profiles, preference migration and decision context
Graph structure
Four entity types, one relational network
Every edge between nodes is an explainable taste connection.
BrandsCultureProductsUsers
Core capabilities
Making taste computable, reasonable and usable
Multi-relational reasoning
Reason along edges of style, scene, lineage and identity — not just behavioral correlation.
Native Asian multilingual
One ontology across eight languages: the same aesthetic concept is aligned across markets, not fractured by translation.
Dynamic evolution
Continuously absorbs content, reviews, new products and community signals to map taste migration and lifecycles.
Layered ontology
A shared top-level taste ontology plus local cultural layers — global consistency with local depth.
Open APIs
Graph queries, relational reasoning and taste tags available as standard APIs to agents and brand systems.
Construction
How the graph is built
Multi-source understanding
Recognize taste expression across products, content, communities and expert cultural corpora — reading context, not extracting buzzwords.
Relation extraction
Extract relations of style, lineage, scene and emotion between entities, each edge backed by evidence.
Cross-lingual alignment
Align aesthetic concepts across eight languages and markets, removing the semantic loss of translation.
Continuous evolution
Incremental signals update graph weights and structure, so taste evolution is captured continuously.
FAQ
About the Taste Knowledge Graph
How is this different from an ordinary product knowledge graph?
A product graph organizes goods by category, attributes and specs and answers “what is it”. A taste graph organizes brands, culture, products and users by aesthetic and cultural relations, answering “why do people like it — and what might they like next”.
How does the graph handle cultural differences across Asian markets?
A shared top-level ontology keeps markets comparable, while local cultural layers carry each market's own motifs and vocabulary. Cross-lingual alignment maps “same taste, different expression”.
Where does the data come from, and how is quality ensured?
It combines public content, product information, expert cultural corpora and licensed partner data. Every relation keeps its evidence source and passes model extraction, rule-based checks and manual sampling.
How can brands use the graph?
Through APIs, datasets and insight products: taste profiles, trend signals, brand positioning maps, product-innovation inputs and cross-region strategy — see Brand Intelligence Infrastructure.
Build your intelligence on the Taste Knowledge Graph
Graph query and reasoning capabilities, open to decision agents, consumer apps and brand systems.