/daniel_fintoc2019

Our participation to the "Financial Document Structure Extraction" task 2019

Primary LanguageTeX

=== FINSBD 2020 ===
To come

=== FINTOC 2019 ===

Usage : 
  use_model.py
    the predict_title function takes as input a python dictionary with the following informations:
      "text_line" : text line to be classified (String, mandatory)
      "begins_with_numbering" : 0 or 1 (not mandatory)
      "is_italic": 0 or 1 (not mandatory)
      "is_all_caps" : 0 or 1 (not mandatory)
      "begins_with_cap" : 0 or 1 (not mandatory)
      "page_nb" : integer (not mandatory)
   returns a triplet :
     Model name
     Class predicted
     Title probability  

TIT :
run1 1,3 simple DT10 (run1)
run2 baseline 5 (run2)
run3 1,4 simple DT10


July 20, 2019: system results due to participants
July 27, 2019: shared task system papers due
Aug 10, 2019: reviews due
Aug 17, 2019: notification of acceptance
Aug 24, 2019: camera ready version of shared task system papers due
Sep 30, 2019: Workshop day

Classement

Subtask1: Title detection
---------------------------------------------------
Aiai_2            0.9818997315023511
Aiai_1            0.9766402240293054
UWB_2          0.9723895009266195
FinDSE_1      0.9700572855958501
FinDSE_2      0.9684006306179805
UWB_1          0.9653446892789734
Daniel_1         0.9488117489093626
Daniel_2         0.9417339436713312
YseopLab_1    0.9124937810249167
YseopLab_2    0.9113421072180891


Subtask2: TOC generation
-----------------------------------------------------
Daniel_1          0.4272235845712835
IHSMarkit_1     0.39418792730119045