Wrong reading order

[lfoppiano]

s41588-024-01785-9.pdf

This block should come after:

Mutation rates varied across the different cancer types with cuta-<lb/>neous melanoma having the highest single nucleotide variant mutation <lb/>count and meningioma the lowest (Extended Data Fig. 2). A total of 945 <lb/>samples, notably colorectal and uterine cancers, were hypermutated, <lb/>either as result of defective mismatch repair (dMMR) or POLE mutation. <lb/>Invasive ductal carcinoma of the breast had the highest power for driver <lb/>gene detection (&gt;90% power for a mutation rate of at least 2% higher <lb/>than background) and large cell lung cancer had the lowest power (Fig. 2 <lb/>and Supplementary Table 4). Compared with the recent Pan-Cancer <lb/>Analysis of Whole Genomes analysis 12 , the 100kGP cohort was better <lb/>powered to identify a driver mutation for 19 cancers, notably for breast, <lb/>colorectal, esophageal and uterine cancer, lung adenocarcinoma and <lb/>bladder transitional cell carcinoma where the sample sizes were more <lb/>than tenfold higher. <lb/>Spectrum of cancer driver genes <lb/>Across all cancer types we identified 770 unique tumor-driver gene <lb/>pairs corresponding to 330 unique candidate cancer driver genes <lb/>(Fig. 3, Extended Data Fig. 3 and Supplementary Table 5). When <lb/>

This block should come first:

vision of precision oncology through WGS to National Health Service <lb/>(NHS) patients as part of their routine care 11 . <lb/>Here, we report an analysis of WGS data on 10,478 patients span-<lb/>ning 35 cancer types recruited to the 100kGP (Fig. 1a). Across all cancer <lb/>types we identify 330 candidate driver genes, including 74 which are <lb/>new to any cancer. We relate these to their actionability both in terms <lb/>of currently approved therapeutic agents and through computational <lb/>chemogenomic analysis to predict candidacy for future clinical trials. <lb/>Results <lb/>We analysed 10,478 cancer genomes spanning 35 different cancer types <lb/>(Fig. 1b and Supplementary Tables 1 and 2). While broadly reflecting <lb/>the spectrum and frequencies of cancers diagnosed in the UK popula-<lb/>tion, there were differences, with an over-representation of colorectal <lb/>and kidney cancers and a paucity of prostate and pancreatic cancers <lb/>(Extended Data Fig. 1). Additionally, for the main cancer types, the <lb/>patients recruited to 100kGP tended to be younger and had earlier <lb/>stage tumors compared to patients in the general UK population (Sup-<lb/>plementary Table 3). <lb/>

[lfoppiano]

Another example: ijn-183907-corrigendum-antibacterial-properties-and-toxicity-from-metal-101518.pdf

<div
                xmlns="http://www.tei-c.org/ns/1.0">
                <p>© 2018 Vimbela et al. This work is published and licensed by Dove Medical Press Limited. The full terms of this license are available at 
                    <ref type="url" target="https://www.dovepress.com/terms.php">https://www.dovepress.com/terms.php</ref> and incorporate the Creative Commons Attribution -Non Commercial (unported, v3.0) License (
                    <ref type="url" target="http://creativecommons.org/licenses/by-nc/3.0/">http://creativecommons.org/licenses/by-nc/3.0/</ref>). By accessing the work you hereby accept the Terms. Non-commercial uses of the work are permitted without any further permission from Dove Medical Press Limited, provided the work is properly attributed. For permission for commercial use of this work, please see paragraphs 4.2 and 5 of our Terms (
                    <ref type="url" target="https://www.dovepress.com/terms.php">https://www.dovepress.com/terms.php</ref>). International Journal of Nanomedicine 2018:13 6497 International Journal of Nanomedicine Dovepress submit your manuscript | 
                    <ref type="url" target="www.dovepress.com">www.dovepress.com</ref> Dovepress 6497 C o r r I g e N d u m open access to scientific and medical research open Access Full Text Article S183907 Antibacterial properties and toxicity from metallic nanomaterials [Corrigendum] Vimbela GV, Ngo SM, Fraze C, Yang L, Stout DA. Int J Nanomedicine. 2017;12:3941-3965.
                </p>
                <p>Page 3943, Figure 1, the following text should be included in the Notes section: Adapted by permission from Springer Nature. Springer, Journal of Nanoparticle Research. A review of the antibacterial effects of silver nanomaterials and potential implications for human health and the environment. Marambio-Jones C, Hoek EMV, 2010;12:1531-1551. Copyright © 2010. 145 The figure is excluded from the CC-BY-NC license under which the article is published by Dove Medical Press. Page 3965, References, a reference was excluded from the reference list, the missing reference is "145. Marambio-Jones C, Hoek EMV. A review of the antibacterial effects of silver nanomaterials and potential implications for human health and the environment. Journal of Nanoparticle Research. 2010;12:1531-1551". International Journal of Nanomedicine Publish your work in this journal Submit your manuscript here: 
                    <ref type="url" target="http://www.dovepress.com/international-journal-of-nanomedicine-journal">http://www.dovepress.com/international-journal-of-nanomedicine-journal</ref> The International Journal of Nanomedicine is an international, peerreviewed journal focusing on the application of nanotechnology in diagnostics, therapeutics, and drug delivery systems throughout the biomedical field. This journal is indexed on PubMed Central, MedLine, CAS, SciSearch®, Current Contents®/Clinical Medicine, Journal Citation Reports/Science Edition, EMBase, Scopus and the Elsevier Bibliographic databases. The manuscript management system is completely online and includes a very quick and fair peer-review system, which is all easy to use. Visit 
                    <ref type="url" target="http://www.dovepress.com/testimonials.php">http://www.dovepress.com/ testimonials.php</ref> to read real quotes from published authors. Dovepress International Journal of Nanomedicine downloaded from 
                    <ref type="url" target="https://www.dovepress.com/">https://www.dovepress.com/</ref> For personal use only.
                </p>
            </div>

[lfoppiano]

Article: 38_2006_Article_5071.pdf

This paragraph

The partnership framework also privileges consensus with regard to the interpretation, the presentation and the dissemination of research fi ndings. In practice, this commitment means that all the partners are to be given access to the fi ndings before they are published. In cases of disagreement over the conclusions and recommendations, a partner cannot block the diffusion of the results, but the partnership framework affi rms that all parties may express dissension publicly and have input in alternative interpretation as the fi ndings are diffuseThe foregoing description of the development of a research partnership framework to support a university-based Chair refl ects a complex negotiation process. The Chair's mission d. Lastly, the partnership framework stipulates that the partners collaborate in developing measures or action plans stemming from the research. The foregoing description of the development of a research partnership framework to support a university-based Chair refl ects a complex negotiation process. The Chair's mission 

should come before this:

[...] statement defi ned a negotiated zone. Reaching out to potential partners with just a mission statement for the Chair in hand permitted them to participate in structuring a research partnership. At the stage of the initial contacts with the potential partners, "what is important is not so much the clarity of the fi nal objectives or the yet-to-be-structured mode of operating, but the creation of a dynamic through which guidelines and a new operating logic gradually emerge" [...]

[blazej222]

I can confirm the issue is commonly present - I've parsed about 800 pdf files that have exactly the same problem (each single one of them).
Is there any kind of workaround for now or any explanation of why that happens?

[lfoppiano]

HI @blazej222 I don't think there is an easy fix / workaround, it's the way the PDF are created that often does not result in blocks being extracted in the correct order.

We've been working hard to update pdfalto and this is definitely one of the issue we are going to tackle in the following months. Unfortunately, the task is complex and we have limited time, so it might take some time.

[kermitt2]

Hello !

@lfoppiano @blazej222, by default, pdfalto follows the stream order, with some particular ordering mechanisms for columns and at token level. When using the option --readingOrder, the blocks are re-ordered following left-to-right top-down order based on coordinates, managing overlaps, etc. - however this is restricted to the first page if I remember well. Given that, does using this option improve the result in your case (only first page)?

Using the current reading order option implementation appeared doing more harm than good in general as compared to the stream order, which is why it is not used by default, and why it is limited to the first page (where reading order matters more - but even in this case, this was moderately successful). So this is a tradeoff, at this stage toward stream order. We could apply the reading order option to the whole document easily with a few edits. Improving the current reading order mode is challenging but there is room for improvement.

[AaronNGray]

A global document column flag/count could be a good idea but it might not apply to all pages sometimes there are mixed landscape single or multi page tables or images within mainly portrait documents. So it would have to be overridable on individual pages, but somehow remain sticky ? Also citations can sometimes be a different format than the rest of the document.
I would like to see a facility for an OCR interface for older camera ready documents that do not have complementary embedded text in too. And for those that do we do not have the embedded formatting information with this either. A lot of early PDF paper documents like scientific papers and PhD thesis fit into these categories, particularly in Computer Science and Type Theory. I should go through my papers library and compile a sample. Maybe we can get this into the next version after this coming release of Grobid ?

[lfoppiano]

Here another example: grobidOrg/grobid#1352