A significant amount of the textual content available on the Web is stored in PDF files. These files are typically converted into plain text before they can be processed by information retrieval or text mining systems. Automatic conversion typically introduces various errors, especially if OCR is needed. In this empirical study, we simulate OCR errors and investigate the impact that misspelled words have on retrieval accuracy. In order to quantify such impact, errors were systematically inserted at varying rates in an initially clean IR collection. Our results showed that significant impacts are noticed starting at a 5% error rate. Furthermore, stemming has proven to make systems more robust to errors.

A significant portion of the textual information of interest to an organization is stored in PDF files that should be converted into plain text before their contents can be processed by an information retrieval or text mining system. When the PDF documents consist of scanned documents, optical character recognition (OCR) is typically used to extract the textual contents. OCR errors can have a negative impact on the quality of information retrieval systems since the terms in the query will not match incorrectly extracted terms in the documents. This work introduces sOCRates, a post-OCR text correction method that relies on contextual word embeddings and on a classifier that uses format, semantic, and syntactic features. Our experimental evaluation on a test collection in Portuguese showed that sOCRates can accurately correct errors and improve retrieval results.