| Summary: | 碩士 === 樹德科技大學 === 資訊管理研究所 === 91 === Plagiarism is an important issue in copyright and authorization protection.With the growth of the Internet, it is much more easier for plagiarists to copy from the Internet various materials and put them in their own documents without
the permission of the original authors. Manual detection may be the most accurate detecting approach to this issue. Intuitively, to detect plagiarism of documents is to detect how similar the documents are. There are only several automated systems for plagiarism detection. The complexity and
performance of these methods depend on how documents are copied. Most of the methods employ pairwise comparison of the input strings. For the documents which are copied and past without any modications, detecting the existence of a copied area is simple. Nevertheless, plagiarists usually change
the order of sentences, add or delete a couple of words, modify the writing style, etc., to prevent from being detected. These approaches perform well on the documents which are completely plagiarized from the sources. It is ob-
served that there are a lot of materials available on the Internet and a variety of plagiarizing behaviors. Simple comparison is not sucient for detecting documents that are tailored by sophisticated plagiarists. In this paper, we
present how the knowledge-based approach can serve as a
exible solution to the detection of plagiarized text. We analyze the types and behaviors of plagiarism and describe documents to be test in form of graph structures.
The problem of detecting plagiarism then is rephrased as the comparison of such graphic structures.We consult several senior faculties with our depart-
ment for collecting the knowledge of selecting, identifying, and comparing the structures and contents of documents. Based on this idea, a knowledge-based system is implemented in CLIPS. Experimental results show that our approach is workable and eective.
|
|---|
