Significant Word-based Text Alignment for Text Reuse Detection
2017
One of challenges in text reuse detection is how to detect the source-reused passage pairs which have a wide range of similarity degree by one method or a single alignment algorithm only.However, the academic texts are rich with terminologies which are hardly altered when one reuses the existing texts in his/her writings.In this paper, we introduce and analyze the use of significant words filtered through word local weighting from the field of text summarization to align source-reused passages.We base our alignment process on the paragraph segmentation which is filtered by the use of their weighted and binary vectors.We demonstrate that the proposed text alignment method is capable of detecting the sourcereused passage pairs which are obfuscated by means of literal copy, copy and shake, and paraphrases from light, medium to heavy levels.