Every year, the demand of information extraction from documents is rapidly increasing due to the speedy annual rise of biological documents. We construct a system which can easily search biological documents that users are interested in by using a sea...
Every year, the demand of information extraction from documents is rapidly increasing due to the speedy annual rise of biological documents. We construct a system which can easily search biological documents that users are interested in by using a search tool of the system itself, and extract useful biological information in the documents. In order to perceive Bio-named Entities (DISEASE, DNA, RNA, CELL-TYPE, CELL-LINE, PROTEIN), we use a conditional random fields method which is a machine learning algorithm. Specially, Bio-named entities used NLPBA corpus and the corpus added disease names which are extracted by using MeSH and BioText. Also, we perform the Bio-Named Entities recognition algorithm using MALLET and ABNER API. For searching abundant data, XML format containing data up to the year of 2005, which is included in PUBMED, was parsed by using SAX-API, and database suitable for the system was implemented. Furthermore, users can parse documents by inputting XML file which fits PUBMED XML document type, and extract and save desired information. System was offered in a form of GUI and was implemented as JAVA.