Comparison between fingerprint and winnowing algorithm to detect plagiarism fraud on Bahasa Indonesia documents

Abstract

Plagiarism detection has been widely discussed in recent years. Various approaches have been proposed such as the text-similarity calculation, structural-approaches, and the fingerprint. In fingerprint-approaches, small parts of document are taken to be matched with other documents. In this paper, fingerprint and Winnowing algorithm is proposed. Those algorithms are used for detecting plagiarism of scientific articles in Bahasa Indonesia. Plagiarism classification is determined from those two documents by a Dice Coefficient at a certain threshold value. The results showed that the best performance of fingerprint algorithm was 92.8% while Winnowing algorithm's best performance was 91.8%. Level-of-relevance to the topic analysis result showed that Winnowing algorithm has got stronger term-correlation of 37.1% compared to the 33.6% fingerprint algorithm.

Plagiarism Detection for Indonesian Texts

“…All of these systems do comparison on the scope [94], [89], [45] of document level and none hinted on segmentation or chunking techniques. In measuring the number of shared fingerprints, Dice coefficient is more favoured than Jaccard as it is applied in [100,143,186]. Mardiana et al compared the performance of the system by using two similarity measures: Jaccard and Dice [100].…”

“…The winnowing algorithm needs at least a parameter that is the length of hash window. Window length applied in these 5 systems are 4 hashes [133,186], 30 hashes per window [135], and unspecified window length [100,175]. The illustration on winnowing algorithm is presented here in figure 3.3 with a reason that it is used also as fingerprint selection strategy in Plagiarism Detection (see section 3.4.2.1).…”

“…30 characters [135]. Unlike 5 researches mentioned before, Wibowo et al used word unigram as feature unit and MD5 function for generating fingerprints of a document [186].…”

“…Certainly, there are many strategies on how to select these features to be fingerprints which could be classified into four, namely full fingerprinting, positional strategies, frequency-based strategies, and structure-based strategies [74]. Whether it is coincidence or not, winnowing algorithm becomes the favorite strategy in fingerprinting selection as it is applied in 5 out of 6 papers reviewed in this group [100,133,135,175,186]. Winnowing algorithm is a fingerprint selection strategy which combines the positional strategy with the minimum hash value in a window.…”

“…Among those which build their own evaluation corpus, the number of documents tested varies from 2- [135], and 70 documents [5]. The test documents are mostly a literal copy from one or more than one source documents with no obfuscation at all or with an obfuscation which is done by shufling the order of paragraphs or sentences [2,4,73,100,143,186]. The obfuscation types done on a literally copied d plg from one d src are synonym replacement on the 50% of document length [100], paraphrasing some sequences with paraphrase percentage of 20% [186] or 50% [100] of document length, partial paraphrase on the sentence level [4], summary obfuscation in a small portion [157], sentence structure alterations such as changing the voices from active to passive [5,186].…”