Research map / Probabilistic and scientific modeling / Molecules and proteins
College Towns, Vacation Spots, and Tech Hubs: Using Geo-Social Media to Model and Compare Locations
Hancheng Ge, James Caverlee
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Cluster cities by tweet interest models instead of demographics | spectral clustering · 112 cities vs expert lists | avg within-cluster similarity: 0.72 -> 0.81 | Improved |
| Cluster cities by tweet interest models instead of LDA topics | spectral clustering · 112 cities vs expert lists | avg within-cluster similarity: 0.56 -> 0.81 | Improved |
| Cluster by interest models instead of demographics | spectral clustering · 112 cities vs expert lists | avg similarity with all others: 0.32 -> 0.31 | Improved |
| Cluster by interest models instead of LDA | spectral clustering · 112 cities vs expert lists | avg similarity with all others: 0.35 -> 0.31 | Improved |
| Apply SMOTE oversampling to topical/non-topical training data | classifiers · Topical vs non-topical tweets | precision: maximum improvement of 4.3% | Improved |
| Use binary plus token features with Random Forest | Random Forest · Topical vs non-topical tweets | precision: 89.1% | Improved |
| Token weighting scheme (4, 2, 1) instead of (3,2,1) or (5,3,1) (ablation) | token features · Topical tweet classification | performance | Improved |
| Use MAX token-feature rule alone instead of majority vote ensemble (ablation) | MMF classifier · Tweet categorization | F1: 92.83% vs 85.65% | Mixed |
| Choose number of clusters via CCC | spectral clustering · 112 cities | CCC: k = 10 | Improved |
| Assign cities to clusters with smaller Euclidean distance (new cities) | MMF city interest model · 56 new cities | avg F1: 100% (cluster 6) vs 34% (cluster 9) | Mixed |
| Use MMLR instead of neural network | demographic-to-interest predictor · 168 cities | RMSE: 0.0186 -> 0.0122 | Improved |
Closest papers by meaning
- Predicting Demographics of High-Resolution Geographies with Geotagged Tweets (AAAI 2017 · similarity 0.86)
- POI2Vec: Geographical Latent Representation for Predicting Future Visitors (AAAI 2017 · similarity 0.86)
- GeoMamba: Towards Multi-granular POI Recommendation with Geographical State Space Model (AAAI 2025 · similarity 0.83)
- Integrating Personalized Spatio-Temporal Clustering for Next POI Recommendation (AAAI 2025 · similarity 0.82)
- GeoLLM: Extracting Geospatial Knowledge from Large Language Models (ICLR 2024 · similarity 0.81)
- SPOT-Trip: Dual-Preference Driven Out-of-Town Trip Recommendation (NeurIPS 2025 · similarity 0.81)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
