General Info
Participant Id 125
Run Id MDCS
Task book-toc
Toc Creation automatic
Toc Source full
Source files XML Yes
JPG No
PDF No
Description In the first step each page in a book is assigned with a logical page number. Then, all TOC pages are detected and TOC structure extraction is performed on each. After all words in TOC pages are labeled the words are grouped into entries. Each entry is given a depth level acording to the entry clustering. Also, each entry is assigned with a link by searhing for the entry title on the target page (even if the entry does not have page number; if it does a logical page numbers are used).
Results
  Precision Recall F-Measure
Titles 52,44% 56,96% 53,11%
Levels 43,82% 47,00% 44,17%
Links 48,09% 52,07% 48,72%
Complete entries 40,40% 43,17% 40,75%
Entries disregarding depth 48,09% 52,07% 48,72%