A novel extreme learning fault diagnosis based supervision applied to mathematical formula contrastive analysis

Focusing on the issue of few contrastive analysis technologies existing on mathematical formula in literatures, two methods of mathematical formula contrastive analysis subject to PDF and XML document formats are proposed. First, a novel extreme learning machine (ELM) fault diagnosis based supervision is presented and applied to mathematical formula contrastive analysis subject to PDF document. The strategy employs its merits such as fast learning speed, ease of implementation high efficiency and minimal human intervention, and not rely on the hidden neurons tuned. At the same time, the supervision mechanism is combined to realize the novel mathematical formula contrastive analysis based ELM supervision (ELMS). Second, a new mathematical formula contrastive analysis based MathML subject to XML document formats is presented. The idea first recognizes and extracts mathematical formulas in the detected XML document, normalizing the markup code of mathematical formulas. And then creating the tree presentation of mathematical the formula according to its presentation markup code, normalizing the tree structure by rule base, level traversing the tree to normalize the variable names and to get structure code of the tree. At last, searching the formula information table that named by the structure code. If the table exist, then searching the preorder traversal sequence of the tree in the table. If the records exist, then searching the postorder traversal sequence of the tree in the records. The searching result confirm if the mathematical formula belong to plagiarism. The experimental results reveal that the proposed algorithms can complete the contrastive analysis accurately of mathematical formula in XML documents whatever it existings PDF or XML format and has high performance on detection accuracy and speed.