| General Info | |||
| Participant Id | 299 | ||
| Run Id | 1 | ||
| Task | book-toc | ||
| Toc Creation | automatic | ||
| Toc Source | book-toc | ||
| Source files | XML | No | |
| JPG | No | ||
| Yes | |||
| Description | This implementation relies on the data provided in PDFs and it supports books with a ToC available. First, it finds the ToC location. Then, lines are reconstructed grouped with respect to text boxes features: alignments, lines ending with numbers or not, text containing specific words, etc. | ||
| Results | |||
| Precision | Recall | F-Measure | |
| Titles | 42,85% | 40,15% | 40,66% |
| Levels | 30,94% | 29,98% | 29,90% |
| Links | 22,51% | 21,10% | 21,46% |
| Complete entries | 15,38% | 14,99% | 14,96% |
| Entries disregarding depth | 22,51% | 21,10% | 21,46% |