Optical Character Recognition Engines Performance Comparison in Information Extraction

Named Entity Recognition (NER) is often used to acquire important information from text documents as a part of the Information Extraction (IE) process. However, quality affects accuracy data obtained, especially for acquired involving Optical Character (OCR) process, which never reached 100% accuracy. This research tried examine OCR engine with highest performance IE using NER by comparing three engines (Foxit, PDF2GO, Tesseract) over 8,562 government human resources within six document categories, two structures, and four measurements. Several essential entities such name, employee ID, number, publishing date, rank, family member's name were trying be extracted automatically documents. processes done Python programming language, preprocessing tasks separately Foxit, Tesseract. In summary, each has its drawbacks benefit, Tesseract better extraction conversion time but lack in number acquired.