• DocumentCode
    627340
  • Title

    Automated Bangla essay scoring system: ABESS

  • Author

    Islam, Md Minarul ; Hoque, A.S.M.L.

  • Author_Institution
    Dept. of CSE, Mymensingh Eng. Coll., Mymensingh, Bangladesh
  • fYear
    2013
  • fDate
    17-18 May 2013
  • Firstpage
    1
  • Lastpage
    5
  • Abstract
    Essays are the most useful tool to assess learning outcomes. However, teachers have not enough time to evaluate a student´s writing properly because of their other assigned responsibilities. Several Automated Essay Grading (AEG) systems have been developed to evaluate the human written (not hand written) essays easily. But, most of the AEG systems are used for grading English language or essays written in pure European languages. We have developed an Automated Bangla Essay Scoring System (ABESS) using Generalized Latent Semantic Analysis (GLSA). GLSA is an improved information retrieval technique. The ABESS has been evaluated using essay sets for two domains: standard Bangla essays titled “Bangladesher Shadhinota Songram ()”, and “Karigori Shiksha ()”. We have gained a higher level of accuracy as compared to human grader.
  • Keywords
    computer aided instruction; information retrieval; natural language processing; teaching; ABESS; AEG; Bangladesher shadhinota songram; English language; European languages; GLSA; automated Bangla essay scoring system; automated essay grading systems; generalized latent semantic analysis; information retrieval technique; karigori shiksha; learning outcomes; teachers; Accuracy; Indexes; Information retrieval; Matrix decomposition; Semantics; Training; Vectors; Automated Bangla Essay Scoring System; Automated Essay Grading; Latent Semantic Analysis; N-grams; Singular Value Decomposition;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Informatics, Electronics & Vision (ICIEV), 2013 International Conference on
  • Conference_Location
    Dhaka
  • Print_ISBN
    978-1-4799-0397-9
  • Type

    conf

  • DOI
    10.1109/ICIEV.2013.6572694
  • Filename
    6572694