We are happy to announce the winners of the first part of 15th ECML PKDD Discovery Challenge. In total 17 teams participated in the competition, setting new standards for the new recommendation task of recommending given names. A broad variety of recommendation techniques has been applied and adapted to this new kind of data, ranging from classical collaborative filtering approaches and association rule mining to advanced factorization machines.
| Pos | Diff | Team Name | Score |
|---|---|---|---|
| 1 | uefs.br | 0,0491 | |
| 2 | ibayer | 0,0472 | |
| 3 | all your base | 0,0423 | |
| 4 | Labic | 0,0379 | |
| 5 | cadejo | 0,0367 | |
| 6 | disc | 0,0340 | |
| 7 | Context | 0,0321 | |
| 8 | TomFu | 0,0309 | |
| 9 | Cibal | 0,0262 | |
| 10 | thalesfc | 0,0253 | |
| 11 | Prefix | 0,0203 | |
| 12 | Gut_und_Guenstig | 0,0169 | |
| 13 | TeamUFCG | 0,0156 | |
| 14 | PwrInfZC | 0,0130 | |
| 15 | persona-non-data | 0,0043 | |
| 16 | erick_oliv | 0,0021 | |
| 17 | Chanjo | 0,0016 |
Although the challenge’s evaluation scores indicate a leadership, the results are very close. At the workshop we will present and discuss a thorough evaluation and comparison of the different recommendations.
We are looking forward to see different recommender systems in the online challenge!