Skip to content

line of reasoning

Last updated on July 26, 2021

My line of reasoning as to what Visual-Meta can enable in a large research setting such as the NIH:

There is a lot of important research and work needing doing and health is surely core, since we can’t work well if we are not well. If the NIH mandated Visual-Meta (let me fantasise about this from an extreme outcome first, as I sit here with my espresso macchiato this slightly foggy Wimbledon morning), then these are among the things I think would happen. I hope my excitable rambling provides a useful perspective:First of all, researchers, whether student or graduate, can simply copy and paste to make a robust citation. This, in itself, makes the citation chain less likely to introduce errors.

A researcher can expect to investigate a citation in-situ. Rather than just clicking on a number (reference marker) in body text and jumping to the References section and so loosing their context, all useful info can be read via a pop-up over the clicked citation reference marker. This is possible since the Reference information within the document is accessible to the viewer software in a parseable, structured, form.

This researcher can also expect to be able to click on a citation and, if the cited document is already downloaded onto their device, have it open up—and at to the specific page cited. This instead of opening a browser window to a download page to which they may then have to log in, before clicking to download the document and locate the cited passage.

The extraction of names of authors becomes much easier and more robust. This allows the researcher to use any tools they want for citation mapping, to see who has cited who and so on. Also–and this is important–it becomes trivially easy for anyone to build such tools. This shown by Adam Wern’s citation timeline based on Mark Anderson’s HT conference data (latter made pre Visual-Meta, but it illustrates what Visual-Meta makes easier) shows:

Regarding the building of tools, we are giving away Visual-Meta parsing code because having the citation data be clean and reliable will make it easy for any developer to build new interactions into their PDF viewers, and to make visualisations and interactions of large volumes of documents (q.v. demos above).

Because Visual-Meta includes headings information. Thus, ML analysis such as simply only showing real names in a document (which is becoming easier and easier for developers to add) becomes instantly more useful when you can see the analysis in context of the headings in the document. Also, by having the document retain structural information it becomes easy for viewing software to add instantly accessible views—and we know how important Doug Engelbart felt views were. Bertrand Russell, writing more abstractly about views, used the example of how binocular vision provides a richer sense of what is being viewed than is afforded by a single eye, with a single point of view. Now we must imagine what an almost unrestricted amount of views can give us.

The clean VM data surfaces the contents of the document in computationally useful ways, describing the data they contain, for easy search of just the data itself.

An issue which I think is becoming more important the more I work on it is how this environment can treat concepts as fundamental units of workable knowledge, making it very easy for the author to define terms and relationships, and even easier for the reader to access them, thus giving researchers a leg-up when accessing work outside their core discipline, as presented here:

Furthermore, Visual-Meta now contains a vm-id which is a unique ID for the document, helping management of documents. These are currently being tested as being either data based or hash, a discussion you are very welcome indeed to join:

vm-id = {2021-7-17T18:00:00:00Z;vintongcer} < start with the ‘created’ data and time (in UTC) of the document, then the first 10 characters of the author’s name (order: first, middle and last, with no spaces, all lower case), using random characters if name is shorter

vm-id-appended = {98765435678987654;reader(3.0)} < hash of original PDF before Visual-Meta page is added, using a sha256 of the raw string value of the document.

In closing, I’d like to mention how robust Visual-Meta is and that it is cheap to add. Most of what is added is metadata in the manuscript which is translated into Visual-Meta rather than discarded upon export to published PDF document.

And finally, Visual-Meta allows anyone to add any custom tags, they just have to explain what they are.

Beyond closing, there are mechanics we are working on where the publisher can issue errata pages of Visual-Meta (which are brief PDF pages themselves) so that the researcher can add them (simply clicking ‘ok’ if the reading software announces that one is available) after the original Visual-Meta for any post-publish errors which might have crept in or other information the publisher wants the reader to be aware of. This can be as simple as a typo or as extreme as pointing out that an error was found in the data or that a cited document has been found to be suspect. Remember, this is just a virtual sheet of (PDF) paper added to the back of the document, it does not change the document but it provides a clean mechanism for updating it (this is Peter Wasilko’s idea, there are some seriously clever people investing time, thought and labour in Visual-Meta). This also addresses an initial idea from the inception of Visual Meta: that of cleaning the overall citation ecosystem, with good data—over time—either enriching incomplete data or supplanting erroneous data.

When elements become addressable as documents, sections and concepts, powerful interactions become possible.

Published inVisual-Meta

Be First to Comment

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.