The Joint European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases (ECML/PKDD) 2007 was held in Warsaw, Poland, from September 17–21, 2007. It included two co-located conferences: ECML 2007 (18th European Conference on Machine Learning) and PKDD 2007 (11th European Conference on Knowledge Discovery in Databases). The proceedings were published by Springer in the Lecture Notes in Computer Science (LNCS) / Lecture Notes in Artificial Intelligence (LNAI) series.
The ECML 2007 proceedings are in LNCS volume 4701 (ISBN: 978-3-540-74958-5). The PKDD 2007 proceedings were also published in the Springer LNAI series. Together, they form the ECML PKDD 2007 accepted papers archive, an important snapshot of research advances in machine learning and data mining.
ECML 2007 Regular & Short Papers
- Statistical Debugging Using Latent Topic Models – David Andrzejewski, Anne Mulhern, Ben Liblit, Xiaojin Zhu. Explores latent topic models to detect anomalies in execution traces.
- Learning Balls of Strings with Correction Queries – Leonor Becerra Bonache, Colin de la Higuera, Jean-Christophe Janodet, Frédéric Tantini. Proposes algorithms for learning string languages under correction queries.
- Neighborhood-Based Local Sensitivity – Paul N. Bennett. Introduces local sensitivity analysis for classifiers.
- Approximating Gaussian Processes with 𝓗²-Matrices – Steffen Börm, Jochen Garcke. Scales Gaussian processes with hierarchical matrix approximations.
- Learning Metrics Between Tree Structured Data – Laurent Boyer, Amaury Habrard, Marc Sebban. Metric learning for tree data with applications in image recognition.
- Shrinkage Estimator for Bayesian Network Parameters – John Burge, Terran Lane. Improves Bayesian network parameter estimation with shrinkage techniques.
PKDD 2007 Regular & Short Papers
- User’s Behaviour Prediction Challenge – Description & Solutions – Joanna Jaworska, Hung Son Nguyen et al. Outlines the Discovery Challenge task on predicting web user behavior and presents team solutions.
- Effective Prediction of Web User Behaviour with User-Level Models – Krzysztof Dembczyński, Wojciech Kotłowski, Marcin Sydow. Demonstrates that user-specific models improve predictive accuracy.
- Bayesian Inference for Web Surfer Behavior Prediction – Malik Tahir Hassan, Khurum Nazir Junejo, Asim Karim. Applies Bayesian inference for multi-task prediction on web session data.
- Predicting User’s Behavior by the Frequent Items – Tung-Ying Lee. Uses frequent item methods for predicting web categories and page visits.
- Stacking Heterogeneous Data Resources – Dimitrios Mavroeidis, Charis Brisagotis, Dimitris Drosos, Michalis Vazirgiannis. Combines diverse datasets for stronger performance in the Discovery Challenge.
Paper Categories at a Glance
| Category | Examples of Papers / Topics | Main Focus |
|---|---|---|
| Metric Learning & Kernel Methods | Learning metrics between trees; Gaussian process approximation | Scalability, structured data |
| Web Behavior & User Modeling | User behavior prediction; discovery challenge | Session modeling, personalization |
| Probabilistic & Bayesian Methods | Shrinkage estimators; Bayesian inference | Uncertainty handling, regularization |
| Learning with Structured Data | Trees, strings, correction queries | Beyond vector inputs |
| Scalability & Approximation | 𝓗²-matrices; frequent item methods | Efficiency on large datasets |
External Links & References
Conclusion
The ECML PKDD 2007 accepted papers archive highlights how researchers addressed fundamental problems such as scalable Gaussian processes, metric learning for structured data, and user behavior prediction. The LNAI proceedings 2007 volumes remain a valuable reference for understanding the development of modern machine learning and knowledge discovery. Many of the methods and ideas presented in Warsaw in 2007 paved the way for techniques that are now standard in today’s machine learning research.