Information Extraction from Research Papers by Data Integration and Data Validation from Multiple Header Extraction Sources

Massive amount of information is available on the web in form of Research paper publications. Extracting Header information like Conference Name, Title, Authors, Affiliation, Email, Keywords and Abstract can be very useful in performing data mining tasks like finding research trends in particular research area or finding collaboration done among different research groups or universities. Existing Header Parser tools identified by Yao et,al.[1] includes GROBID, ParsCit, Mendeley, HeaderParserService, PDFSSA4MET, PDFMEAT, Zotero and PaperPile. Tools using Machine Learning Algorithms include GROBID, ParsCit, Header Parser Service and Mendeley. Now the problem faced here is that no single tool gives 100% results against all sample research papers. One tool outperforms other in identifying individual elements. For this reason one cannot rely on single tool for all elements extraction. In this paper, we are proposing a hybrid method for the extraction of header information from the papers using GROBID, ParsCit and Mendeley. Results of these tools are merged to achieve accurate header extraction. This proposed method has been applied on 75 sample research papers and the overall accuracy of 95.97% is achieved.