SymbolFinder: Brainstorming Diverse Symbols Using Local Semantic Networks
summary
SymbolFinder is an interactive tool that generates local semantic networks from a root concept using Small World of Words, clusters associations with Louvain community detection, and ranks terms by concreteness to surface diverse, imageable symbol candidates. A lab study with 10 novices shows SymbolFinder yields 14.8 symbols per concept versus 9.92 with Google Images (statistically significant) and reduces mental effort. A case study with three designers produces 32 low‑fidelity and six high‑fidelity prototypes for news-cover visual metaphors.
AI-written summary · about
chapters (11)
- 0:00Start
- 1:59Formative study
- 3:40Use word associations to generate a local semantic network
- 4:09Cluster the local semantic network into diverse associations
- 4:22Sort words within each cluster by concreteness.
- 4:39Evaluation: SymbolFinder vs. Google Images
- 5:10Users found significantly more symbols with SymbolFinder.
- 5:35Control results
- 5:54Dangerous results
- 6:20Case study: SymbolFinder in practice
- 6:42Case study: Findings
related talks
24mSF Symbols. Icons for everyone?
21mBeyond Just Text: Semantic Emoji Similarity Modeling to Support Expressive Communication 👫📲😃
56mIcons and the Web: Symbols of the Modern Age
33mVisual thinking - simplify complex
46mLets Design an Icon Set
49mExploring the Anatomy of Design Systems
24mUnleash your visual superpower as a communication pro
18mThe space-filling curve of design