Document understanding is an essential task in various applications areas such as data invoice extraction, subject review, medical prescription analysis, etc., and holds significant commercial potential. Several approaches are proposed in the literature, but datasets' availability and data privacy challenge it. Considering the problem of information extraction from documents, different aspects must be taken into account, such as (1) document classification, (2) text localization, (3) OCR (Optical Character Recognition), (4) table extraction, and (5) key information detection. In this context, graph-based approaches are attractive methods for document processing. In fact, graphs are a natural way to represent the connections among objects (text, blocks, images, etc.) and aim to discover novel and hidden knowledge from data. The extracted text from scanned documents can be represented in the shape of a graph to exploit the best features of their characteristics. On the other hand, understanding spatial relationships is critical for text document extraction results for some applications such as invoice analysis. The aim is to capture the structural connections between keywords (invoice number, date, amounts) and the main value (the desired information). An effective approach requires a combination of spatial and textual information.
In this second edition of the GLESDO workshop, we encourage the description of novel problems or applications for document image analysis in the area of information retrieval that has emerged in recent years. We also encourage works that include NLP tools for extracted text, such as language models and Transforms. Finally, we also encourage works that develop new scanned document datasets for novel applications. The GLESDO workshop aims to bring together an area for experts from industry, science, and academia to exchange ideas and discuss ongoing research in graph representation learning for scanned document analysis.
All submissions will be handled electronically via the EasyChair website. All authors must agree to the policies stipulated below. We welcome the following types of contributions:
Full research papers (8-10 pages): Finished or consolidated R&D works, to be included in one of the Workshop topics.
Short papers (4-6 pages): Ongoing works with relevant preliminary results, opened to discussion.