improved text mining approach for plagiarism detection
Loading...
Date
item.page.authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
newline Plagiarism causes significant damages on innovations in scholarly works such as student
newlineassignments and research. It appears in various forms like Accidental, Unintentional,
newlineIntentional and Self plagiarism and spans across varied data formats: text (articles, scientific
newlinepapers , essays. or just plot ideas in fiction books,Video (scenes, dialogues), music foms
newline(melodies,tunes,compositions ), images (statistics, charts graphics and drawings),internet
newlinesource streams. Text plagiarism comes in multifaceted forms: low-obfuscation (verbatim,
newlineparaphrasing and its variants) as well as high-obfuscation (semantic and conceptual or idea
newlineplagiarism) and more.
newlineThe state-of-the-art and commercially available plagiarism detection merely detect pairs of
newlinesections bearing similar text chunks, both from suspicious and source documents, and hence,
newlinehandling low-obfuscation plagiarism tasks. This leaves us at exploring high obfuscation
newlinelevels of plagiarism as one of the research challenges. The current research objective aims at
newlinedeveloping a novel method of plagiarism detection tool based on complete and precise
newlineguidelines heuristics for anti plagiarism standards both mapped from national and
newlineinternational level. The tool was designed in form of a text semantic frame work to reveal the
newlinetype of plagiarism in the form of word-to-word, paraphrasing, syntactic, semantic or idea
newlineplagiarism. It incorporated logistics for improving the accuracy of existing plagiarism
newlinedetection approaches using semantic based Natural Language Processing (NLP) techniques
newlineupon free text available at different levels of granularity viz. documents, paragraphs and
newlinesentences.
newlineThe research began with drafting the guiding heuristics on all plagiarism detection categories
newlineand mapping them to a precise set of IF-THEN Rules in order to type-cast to an exact
newlineplagiarism category. The second part of the research focused on detection of idea plagiarism,
newlinestill a challenging task for plagiarism detection research community. The outcome of suc