Our great sponsors
-
-
charles-dickens_a-christmas-carol
Epub source for the Standard Ebooks edition of A Christmas Carol, by Charles Dickens
-
WorkOS
The modern identity platform for B2B SaaS. The APIs are flexible and easy-to-use, supporting authentication, user identity, and complex enterprise features like SSO and SCIM provisioning.
-
-
llama2_aided_tesseract
Enhance Tesseract OCR output for scanned PDFs by applying Large Language Model (LLM) corrections, complete with options for text validation and hallucination filtering.
-
InfluxDB
Power Real-Time Data Analytics at Scale. Get real-time insights from all types of time series data with InfluxDB. Ingest, query, and analyze billions of data points in real-time with unbounded cardinality.
Editorial commits are all marked as such and contain no non-editorial changes. The tools for compiling ebook files are available at https://github.com/standardebooks/tools, so creating your own versions with only the work you're interested in is straightforward (and can be at least partially automated).
Yeah, I was just looking through A Christmas Carol and observed a handful of editorial changes in the commits <https://github.com/standardebooks/charles-dickens_a-christma...> (bran-new → brand-new, frouzy → frowzy, and “Lowercase some gratuitously uppercased words”). Frouzy → frowzy I’m ambivalent about. Ditching bran-new definitely loses character. One or two of the lowercased words were mildly strange capitalised (e.g. Idol was inconsistent with the previous paragraph); but the lowercasing of many introduces stylistic inconsistency in many places, including direct local inconsistency sometimes; and most of the capitalisations were not gratuitous. In fact, more than a few were clearly to be pronounced, as a form of emphasis (e.g. Poor, One, Us); and some were distinctly proper nouns in the context the removal of which increases the parse difficulty (e.g. One, /(Cold )?(Roast|Boiled)/); and some reflect customs still common or even preferred in their domains (e.g. Act, Angelic, Apostles, Star). I just reckon that commit should be reverted, because from my perspective it’s mostly actively bad, and the rest subjective.
But yes, I see that you’re practising some editorial oversight and not aiming to faithfully represent the original in all regards, which I gather is more generally Project Gutenberg’s goal; and this would obviously contraindicate upstreaming.
On the other hand, when it comes to more stylistic matters, I tend to wish Project Gutenberg had more consistency. There’s too much gratuitous variation in presentation and ridiculous 256-colour backgrounds. It’s often too obvious much of it is the work of a group of individuals rather than a coherent effort.
I’m curious about the footnote-to-endnote thing, because I’m not sure how the various formats in question handle them all, but in print endnotes are almost always just awful. If anything, I’d be expecting to replace endnotes with footnotes.
Yeah, I was just looking through A Christmas Carol and observed a handful of editorial changes in the commits <https://github.com/standardebooks/charles-dickens_a-christma...> (bran-new → brand-new, frouzy → frowzy, and “Lowercase some gratuitously uppercased words”). Frouzy → frowzy I’m ambivalent about. Ditching bran-new definitely loses character. One or two of the lowercased words were mildly strange capitalised (e.g. Idol was inconsistent with the previous paragraph); but the lowercasing of many introduces stylistic inconsistency in many places, including direct local inconsistency sometimes; and most of the capitalisations were not gratuitous. In fact, more than a few were clearly to be pronounced, as a form of emphasis (e.g. Poor, One, Us); and some were distinctly proper nouns in the context the removal of which increases the parse difficulty (e.g. One, /(Cold )?(Roast|Boiled)/); and some reflect customs still common or even preferred in their domains (e.g. Act, Angelic, Apostles, Star). I just reckon that commit should be reverted, because from my perspective it’s mostly actively bad, and the rest subjective.
But yes, I see that you’re practising some editorial oversight and not aiming to faithfully represent the original in all regards, which I gather is more generally Project Gutenberg’s goal; and this would obviously contraindicate upstreaming.
On the other hand, when it comes to more stylistic matters, I tend to wish Project Gutenberg had more consistency. There’s too much gratuitous variation in presentation and ridiculous 256-colour backgrounds. It’s often too obvious much of it is the work of a group of individuals rather than a coherent effort.
I’m curious about the footnote-to-endnote thing, because I’m not sure how the various formats in question handle them all, but in print endnotes are almost always just awful. If anything, I’d be expecting to replace endnotes with footnotes.
I made a tool like that, and I bet with a more powerful LLM like GPT4, and perhaps a better baseline OCR tool (like GPT4 vision), it could work really well for this sort of thing:
https://github.com/Dicklesworthstone/llama2_aided_tesseract