Precision–Recall Trade-off Model Indexing and Searching

Searching approaches specifically designed for untranscribed (handwritten)
text images are demonstrated here following the so-called “Precision–Recall
Trade-off Model”:  Rather than exact searching, a confidence threshold
is explicitly or implicitly specified by the user as part of the query in order
to meet the precision-recall trade-off which is considered most adequate
in each query.  Are you interested to see how it works?
 
  Small sample for experimental tests (42 Vol-I pages with ground truth info):
  → Choose an interface:  confidence-oriented  or  rank-oriented
 
  Full Vol-I of PLANTAS, excluding training images (656 page images):
  → Choose an interface:  confidence-oriented or  rank-oriented
 
  Vol-VII of PLANTAS  (739 untranscribed page images):
  → New interface, integrating confidence & rank settings
 
          Indexed is based on a new hierarchical indexing model and the
          demonstration uses a new user interface which supports this model.   
          Morphological (or Optical) and Linguistic Models trained on the
          training set of the ICFHR-2014 HTRtS competition (400 pages).
 
          Small sample for experimental tests (33 pages with ground truth info):
         → Choose an interface:  new hierarchical searching interface
 
          All the indexed BENTHAM pages, excluding training images:
         → Choose an interface:  new hierarchical searching interface