@techreport{TR-IC-PFG-18-25,
   number = {IC-PFG-18-25},
   author = {Guilherme {Pereira Gribeler}},
   title  =  {{Semantic  Metadata  Extraction  from Subtitles of Video
                   Lectures}},
   month = {December},
   year = {2018},
   institution = {Institute of Computing, University of Campinas},
   note = {In English, 25 pages.
    \par\selectlanguage{english}\textbf{Abstract}
       Video  lectures  can stimulate learning experiences considering
       individual  needs  and  learning  styles.  Extracting  relevant 
       information from video lectures can be useful to recommendation
       purposes  and  to  interpret  a  concept in a exact moment of a
       lecture  that  a  student  can  be  interested in watching. The
       extraction  of  semantic metadata from a video natural language
       subtitle  involves  challenges in dealing with informal aspects
       of language and the detection of semantic classes from the free
       text.  In  this  work,  we  propose  a technique for extracting
       semantic  metadata,  which  consists  in  developing  a tool to
       extract the subtitle of a YouTube video in a text file and then
       use semantic annotation tools to identify semantic classes from
       the subtitle text file. We conduct an evaluation to compare the
       effectiveness  of  distinct  semantic  annotation tools on this
       task. Obtained results indicate that both AutoMêta tool and the
       our proposed SubAnnotator tool can perform the task of semantic
       annotating  relevant  terms well, but Ontotext and NCBO are not
       very  effective  for  accomplishing  this  task. The difference
       between  SubAnnotator and AutoMêta is the ability of annotating
       multiple occurrence of the terms throughout the input text. The
       SubAnnotator   was   able   to  annotate  a  higher  number  of 
       occurrences  than  AutoMêta. The results also indicate that the
       biggest   challenge  on  the  video  lectures  semantic  metada 
       extraction  process  is  the definition of the ontology used by
       the tools.
  }
}