citation-brainstorming
Citation Brainstroming
Contributors
- ...
- ... (a bunch of good folks!)
- Tantek Çelik
- Tim White
- Michael McCracken
See also
Use Cases
To focus the discussion, please add use cases below that will help show what problems the citation microformat will be solving.
I've included two, focusing on consuming information - I've assumed that use cases for generating microformatted content would just involve the desire to enable your content to be consumed better, but I'm interested to see if there's something I'm missing here -Mike
Acquiring reference information from the web
A user either finds an author's papers page, or is viewing the results of a search and would like to import the information about the displayed papers into their local reference database, for the purposes of cataloging things they've read, adding notes, and using the information to generate later citations, potentially in other forms, such as BibTeX or Docbook, for inclusion in a publication of their own.
Notes: In this case, it isn't important to the user what format the citation takes as displayed on the page where they find it. What *is* important is that it contains enough information to allow generation of the format they will ultimately re-publish it in. This implies that it may be worthwhile to err a little on the side of verbosity.
Also, links to downloadable full representations of the cited work are very important - e.g. a link to the PDF of a journal article, or to a music file.
Subscribing to reading lists, periodicals, etc
I would like to be able to leverage my news aggregator with hAtom to subscribe to a remote source for citation information, for example:
- a reading list for a seminar
- The publication list for a conference (e.g., subscribe to SIGGRAPH and see the updated conference proceedings every year)
- the issues of a journal
- a particular research group or researcher's publications
- Not just research: a popular author's publications (e.g., Malcolm Gladwell's Archive)
Aggregating reading lists and reviews
A citation microformat-specific aggregator could provide a decentralized version of CiteULike. Libraries, authors, research groups, and publishers could mark up their collections, while other people on weblogs or review sites could add tags and reviews.
At least, having a well-adopted microformat would make writing tools like CiteULike much better, since it relies in some cases on screen-scraping publisher web-sites.
Cut & Paste from web pages
Capturing/copying HTML from web pages for use in other applications (especially when those apps present HTML as output), such as pasting into Word, or a specialized application like Google Notebook, Onfolio or Kaboodle. When such captures are made, it makes sense to keep track of the full citation data, including the date it was accessed, which may or may not be the date it was published.
Blogs quoting other resources, including blogs
Any blog that cites online content, whether a blog or news article, could use an hCitation to properly link to the cited reference. Such citations could include the access date when the blogger made the citation, because resources on the other side of those links can change without notice.
Instead, today we have simple formating with a link to the permaURL. The citation data is completely lacking. See Doc Searl's blog for a style of referencing that could benefit from proper a citation uF.
Fascinating... after I added the last two use cases, I realized they focus on potentially marginal cases. The first because it is missing the "output" part of the cut & paste, where the uF would actually be used as part of the paste. The latter because bloggers have a working citation mechanism that is just a link to the URL (hopefully a permaURL). One could argue they wouldn't want a full hCitation. And in fact, until a tool exists that makes it easy, they probably won't. However, a tool that cuts & pastes from anywhere on the web into a blog with a full citation seems like a nice tool. But again, I'm not really paving the cowpaths with these ideas. -Joe Andrieu
Original hBib Discussion
During the WWW2005 Developer's Day microformats track, Rohit Khare gave a presentation where he discussed the microformats process, and then did a quick demonstration wherein a bunch of us got on a shared Subethaedit document, and brainstormed some thoughts on what an "hBib" bibliography citation microformat would look like. Rohit placed the document on his Commercenet site.
An attempt to summarize and inline the linked document follows. -Mike
Two major goals were outlined by the group:
- Avoid re-keying references
- Adapt to new journal styles by changing CSS
The fundamental problem was discussed in terms of display - the ability to transform XHTML+hBib into the many journal-specific formats. For example, how to display "et.al" when all authors are present in the source, and how to re-order the elements if a style defines a set order of elements that conflicts with the ordering in the source. Using hCard for authors was agreed on, and the beginnings of an example were shown.
XHTML Structure
With my exprience working X2V and hCa* has taught me what elememts are easy to find and which are not. Since the Citation microformat is very new it is possible to not make a lot of the same errors twice and to make things easier for extracting application to find and imply certain properties.
- There should be some sort of 'root node' that implies all child elements are for the Citation microformat.
- Since most people will have multiple Citation there should be away to represent each Citation object as a unqiue block independant of another. This is to keep the parse from finding 'author' and applying to all citations. Each citation should be in a container (class="???") that scoped from others.
- Perhaps class="hcite" with
<cite>recommended as the root element. E.g.<cite class="hcite">
Citation vs. media-info
What distinguishes a cite from say media-info (e.g. media-info-examples) is that a cite is a reference to something explicitly external to the current piece of content or document, whereas media-info describes information about content embedded or inline in the current document.
Semantic Meaning
One of the guiding priniciple of Microformats is to use the most semantically rich element to describe each node (Point 2 of Semantic XHTML Design Principles: Use the most accurately precise semantic XHTML building block for each object etc). Since we are dealing with HTML and citations, several elements are candidates to be used to enrich the semantic meaning. CITE, BLOCKQUOTE, Q, A, (are there more?)
The Citation Brainstorming Page has a few development and ideas about how to give another person credit for a link. Some of the semantic ideas behind their choices of tags can be applied to a full bibliographic type reference. Does this sentence make sense only historically? -Mike
OCLC's WorldCat for titles
Question: what about using something like OCLC's WorldCat for linking titles? - Tim White
This and That
After reading through alot of different citation encoding formats, i noticed that each format was being used in onw of two ways. It was either to describe the Current page (THIS.PAGE) or being used to encode references that point to external resources (THAT.PAGE)
The informatation being encoded was identical for both resources (author, date, name, etc) they just reference different things. For this microformat, i'm not sure if we want to try to solve both problems, or just one? The meta tags in the head element would be the ideal place for information about the THIS.PAGE, but that is not in following with the ideals of microformats where information is human-readable. The THAT.PAGE idea where a list of references is at the end of a document in the form of a bibliography is more inline with the ideals of a microformat where the data is human-readable. That doesn't mean that data about the current document shouldn't be human-readable, so some of the same properties used to reference extermal resources can be used for the current document (THIS.PAGE). To do this a different root item could be used and transforming applications could either extract the citation data about the current page, or information about this page's references.
This is open for discussion, but either way, i believe that the properties used to describe a page will be the same for both THIS and THAT. brian suda
More on This and That
Citation microformats are being explored as a possibility for citing genealogical information at Dan Lawyer's blog.
This is a case where frequently the citation would refer to (THIS.PAGE), but would have nested within it a reference to (THAT.PAGE), possibly a few levels deep. For instance, a web page might contain data extracted from a microfilm of a census. The citation would need to include information about the web page, information about the microfilm, and information about the census. Genealogical citations are expected to include the repository (where can this book or microfilm be found. Is this the same as venue?). So, at each level the information should contain the repository of the referenced item. A nesting (recursive) mechanism for citation microformats would be useful in this case. Is this the function of the "container" element in the Straw Format?
Date Formatting
Since microformats are all about re-use and the accepted way to encode Date-Time has been pretty much settled, then this is a good place to start when dealing with all the different date citation types.
These are all the different fields from various citation formats that are of temporal nature:
* Date (available | created | dateAccepted | dateCopyrighted | dateSubmitted | issued | modified | valid) * originInfo/dateIssued * originInfo/dateCreated * originInfo/dateCaptured * originInfo/dateOther * month * year * Copyright Year * Date - Generic * Date of Confernce * Date of Publication * Date of update/revisou/issuance of database record * Former Date * Entry Date for Database Record * Database Update * Year of Publication
There are several common properties across several citation domains and will certainly be in the citation microformat, the unique instances will need further consideration, otherwise there could be no end to posiblities.
There are also several properties (year, month, Year of publication) that can be extracted from another source. Therefore, if you only encode a more specific property such as; Date of Publication, you can extract the 'year of publication' from that. Since the date-time format we are modeling after is the ISO date-time format, just the Year portion is an acceptable date. So if you ONLY know the year of publication, the you can form a valid 'Date of Publication' as a microformat (which inturn is a valid 'year of publication') - you milage may vary when it comes to importing into citation applications.
...
It seems to me that these can be collapsed to maybe one or two different date properties. As far as the specific human readable formatting of the date, that can be chosen per whatever the presentation style guide says, and the datetime-design-pattern used to simplify the markup. - Tantek
Important
Sometimes we need a date range and not simply a date (e.g. 4-6 May 2006). See Conference Citation examples later on this page. - Discoleo
Tags
Some of the citation formats has a place for 'keywords' or 'generic tags', etc. This might be a good place to re-use the RelTag microformat. The downside would be that they are then forced to be links, which might be the correct way to mark-up these terms.
MARC / MODS / Dublin Core
The MODS (example) and Dublin Core (example) transformations of MARC21 may contain some useful ideas.
Here's a first attempt at rewriting the linked examples in XHTML (written in response to a mailing list query about encoding book information with microformats):
<div class="book" lang="en">
<h3 class="fn">Arithmetic /</h3>
<p>By <span class="creator"><span class="fn">Sandburg, Carl</span>,
<span class="date">1878-1967</span></span>,
and <span class="illustrator">Rand, Ted</span></p>
<p>Publisher: <span class="publisher"><span class="fn">Harcourt Brace Jovanovich</span>,
<span class="locality">San Diego</span></span></p>
<p>Published: <span class="issued">1993</span></p>
<p class="description">A poem about numbers and their characteristics. Features
anamorphic, or distorted, drawings which can be restored to normal by viewing
from a particular angle or by viewing the image's reflection in the provided
Mylar cone.</p>
<p class="note">One Mylar sheet included in pocket.</p>
<p>Subjects:</p>
<ul>
<li class="subject">Arithmetic</li>
<li class="subject">Children's poetry, American.</li>
<li class="subject">Arithmetic</li>
<li class="subject">American poetry</li>
<li class="subject">Visual perception</li>
</ul>
</div>
Basic Citation Stuctures
There are basic structures to any citation, this is an overview of some of the types http://www.users.muohio.edu/darcusb/misc/citations-spec.html
Concerns not addressed by existing formats
There are some aspects NOT adequately covered by existing formats. I have addressed this issue on the OpenOffice.org wiki page, too. [see http://wiki.services.openoffice.org/wiki/Bibliographic_Database for an extending discussion, the paragraph on Reference Types]
These issues pertain mainly to Errata, Comments and Authors Reply and Article Retractions.
- a bidirectional link could be necessary to implement these features (original article <=> eratum, reply, retraction letter)
- IMPORTANT: Errata
- Erata: one or more Corrections might be posted in various issues of the journal
- this is usually cited as: Orininal Article Citation Data (Correction available in Journal, Issue Nr, Year, Pages) (repeat for more than one correction)
- it is possibly never cited alone
- there should be a link to the original article, while the original article should contain a link to this Errata
- IMPORTANT: Commentary and Author Reply
- similar to Errata, there might be one or more Comments and Author Replys; this should be stored, too
- however, it is usually not included in the original citation
- it might be used however in a citation, but I do not know exaclty how to cite it optimally (original article should be provided as well)
- IMPORTANT: Article Retraction
- an article may be retracted because of plagiarism or some other flaw
- this should not be used any further in the research
- however, it might be used e.g. for an article on plagiarism or flawed research
- there should be therefore one field storing this information, too, and a link to:
- the published withdrawal letter (which explains why the article was retracted)
- this issue may need a time-controlled event
- IMPORTANT: electronic publishing ahead of print (EPUB)
- more and more articles are initially posted online, before the published article gets actually printed
- How should this be used/cited?
- Is this changed, after the print version becomes available?
Outstanding Issues
The 3 main points i (Brian) came across so far are: 1) IDENTIFIERS 2) FORMAT TYPES 3) NESTING
1) In hCard/hCalendar there is a UID field. Added with URL it makes for a great unique identifier. There are loads of other identifers besides URL, ISBN, LOC call number, SKU, ISSN, etc. Many of these are unique in their domain, but not globally unique. So how to they get marked-up? Much like the hCard TEL/ADR properties, we can use something like:
<div class="uid"><span class="type">ISBN</span>: <span class="value">123456</span></div>
This makes the encoding the most extensible... if we start use class="isbn" then it is an enumerated list, with class="type" it is open ended.
2) I keep mis-using "format", format is the medium - hardback, softback. The TYPE (there probably is a better word - container?) is book, article, conference, manifesto, etc. Much like the identifers we can make an enumerated list of values, class="book", class="article", but that boxes us in, whereas something like:
<span class="type">article</span>
leaves things more open.
3) Nesting citation data in a citation. The ability to nest the same microformat inside itself is something that other microformats don't explicitly handle.
The two options are: i) Using class="book"
<div class="hcite">
<div class="book">
<span class="fn">Book Title</span>
<div class="chapter">
<span class="fn">Chapter Title</span>
</div>
</div>
</div>
This makes things easy to nest and to figure out exactly what is associated with what, but the downside is that we have enumerated lists of values for the class properties.
ii) using the TYPE for book
<div class="hcite"> <div class="type">book</div> <span class="fn">Book Title</span> <div class="type">chapter</div> <span class="fn">Chapter Title</span> </div>
now the class="fn" is not nested inside the class="book" or class="chapter" so there would have to be some other mechanism to associate the data with the type.
Brian's Straw format
implied schema (examples)
+ publisher + language + description + title + creator + journal + volume + issue + page + edition + identifier + tags + format + date published + copyright - audience
implied schema (formats)
+ publisher + language + description + title + creator + volume + pages + edition + issue + identifier + tags + format + date published + date copyrighted - subtitle - image - excerpt - index terms - series title - publication - journal - part (1 of X)
UNION of the two schemas
+ (PLUS) means common properties - (MINUS) means unique to the schema
Example
<ul class="bibliography"> <li class="hcite" xml:lang="en-gb"> <!-- publisher data as hCard--> <div class="publisher vcard"> <span class="fn org">ABC Publishing Co.</span> <span class="country-name">United Kingdom</span> ... </div> <!-- author(s) data as hCard --> <div class="creator vcard"> <span class="fn n"><span class="given-name">John <span class="family-name">Doe</span></span> ... </div> <!-- location data --> <span class="fn">Foobar!</span> <span class="description">World Class Book about foobar</span> <span class="volume">1</span> <span class="issue">1</span> <span class="edition">1</span> <span class="pages">1-10</span> <span class="format">article</span> <!-- differed to the UID debate --> <span class="identifier">12345678</span> <!-- keywords --> <a class="keyword" rel="tag" href="/tags/foo">foo</a> <span class="keyword">bar</span> <!-- date properties --> Published <abbr class="dtpublished" title="20060101">January 1st 1006</abbr> Copyright <abbr class="copyright" title="20060101">2006</abbr> </li> ... </ul> <p class="citation">Have you read <span class="title"><abbr title="book" class="format">Foo Bar</abbr></span>? It was written by <span class="author vcard"><span class="fn">John Doe</span></span>. It only came out a <abbr class="dtpublished" title="20060101">few months ago</abbr></p>
Note: the "format" property above is incorrect. Format would refer more the physical characteristics of an item, rather than its type or genre (e.g. "article", "book", etc.). I'd rather have the main class for the li be "article" in this context, than the fairly meaningless "citation." Of course, one could have both, which would be fine too. -- bruce
Note: Could we use ROLE from hCard to identify editors, translators, authors, etc? This was discussed on the mailing list and the idea was dropped [1]
Comments : singpolyma 08:03, 16 Jun 2006 (PDT) : keywords should be rel-tag, and probably also XOXO (the same way the citation list is)
Mike straw format suggestion
In the interests of starting debate and having something concrete to fix, I suggest the following structure for a format. It is probably very incomplete and I claim no microformat expertise. I'm just trying to follow existing patterns. Comments and ridicule are both solicited. -Mike
In General
The citation format is based on a set of fields common to many bibliographic data formats, which are often implied by standard citation display styles but not explicitly marked up in practice on the web.
Schema
The citation schema consists of the following:
- cite
- title: required, text (class = fn)
- subtitle: optional, text
- authors: optional, use hCard
- publication date: optional
- link(s) to instantiations, optional, url or use rel-enclosure? (class=url)
- UID, optional (for ISBN, DOI - use existing uid class) | permalink
- series (aka volume/issuenum) , optional (not as sure how to handle these - suggestions?)
- pages: startpage & endpage, optional, text
- venue, optional (hCard)
- publisher, optional (hCard)
- container: optional (nested hCite)
- abstract, optional (blockquote + class="abstract" ?)
- notes, optional (blockquote + class="notes" ?)
- keywords, optional (rel-tag)
- image, optional (for inclusion inline, unlike the url)
- copyright, optional (rel-license)
- what else am I missing?
- language, optional
Looks good, but I question the use of hCard for names. Due to ambiguity issues, requring hCard would lead to extra markup in order to apply just a name, hence the need for a root element. We should extract the N optimization of hCard like we did with adr, in order to ease this problem. --Ryan Cannon
Perhaps a Retrieved Date or Access Date would be appropriate for citing online resources. For example at http://www.crlt.umich.edu/publinks/facment_biblio.html you see citations like this:
Chief Academic Officers of the Big 12 Universities (2000). Big 12 Faculty Fellowship Program. Retrieved December 20, 2000 from the World Wide Web: http://www.k-state.edu/provost/academic/big12/big12guide.htm.
Examples
The following are translations into the citation format.
Note: some of these are just placeholders right now. Please feel free to fill them in!
Citing Private Communication
- published-date seems a weird fit, but it works...
private communication, Michael Jordan, May 2004
Needs a formatted example.
Citing Legal Cases
Needs an example. Here's some info I found about citing law:
I'm not a lawyer, so I'm relying on the published "blue book" standard, at least the only part of it I can get without paying $25. I'd be happy to hear improvements from experts in the field - how do lawyers mark up references to case law in HTML now?
From groklaw.net and eff.org, I find mostly just links to PDFs with the name of the case as the link text. Or just this, from EFF:
<h1>The Betamax Case</h1> <h2>Sony Corp. of America v. Universal City Studios, 464 U.S. 417 (1984)</h2>
From an example at the sample bluepages: http://www.legalbluebook.com/pdfs/bluepages.pdf 5 basic components:
- 1 name of the case (citation title)
- 2 published source in which case may be found (citation containing publication?)
- 3 a parenthetical indicating the court and year of decision (citation venue?)
- 4 other parenthetical information, if any (citation notes?)
- 5 subsequent history of the case, if any (citation notes?)
Here's two examples from the bluebook. Note that there are very strict rules about abbreviations in that source!
Holland v. Donnelly, 216 F. Supp. 2d 227, 230 (S.D.N.Y. 2002), aff'd, 324 F.3d 99 (2d Cir. 2003).
Green v. Georgia, 442 U.S. 95, 97 (1979) (per curiam) (holding that exclusion of relevant evidence at sentencing hearing constitutes denial of due process).
Citing a Book
needs an example
Citing a journal article
needs an example
Citing a magazine article
needs an example
Citing a Patent
Patents are often just cited by number. Here's a citation that accomplishes the same thing with some extra information:
<cite class="citation"> <a class="fn url" href="http://patft.uspto.gov/netacgi/nph-Parser?Sect1=PTO2&Sect2=HITOFF&p=1&u=/netahtml/search-bool.html&r=3&f=G&l=50&co1=AND&d=ptxt&s1=tevanian&OS=tevanian&RS=tevanian">US Patent #6,704,928</a> <span class="author vcard">Richard Shann</span> <abbr class="dtpublished" title="20000828T0000-0500">August 28, 2000</abbr> <blockquote class="abstract"> An executable program is prepared from a plurality of object code modules, at least one of the object code modules including section data specifying a plurality of code sequences each associated with relocation instructions identifying condition parameters. Only one of the code sequences is selected for inclusion in the executable program, determined by whether the condition for that parameter is satisfied. A linker for preparing the executable program includes a stack, a relocation module for reading the relocations, carrying out the relocation operations and selecting code sequences for inclusion in the executable program in dependence on values taken from the stack, a section data module for holding section data which is subject to the relocation operations, and a program forming module for preparing executable programs. Also disclosed is a method of assembling an object code module such that the assembled object code module includes the conditional code sequences.</blockquote> </cite>
Citing a conference publication
Based on the following reference, plus some more information from the ACM site and a little of my own input (the tags)
L. Hochstein, J. Carver, F. Shull, S. Asgari, V. Basili, J. K. Hollingsworth, and M. Zelkowitz, “Hpc programmer productivity: A case study of novice hpc programmers,” in Proceedings of ACM/IEEE Supercomputing Conference, 2005.
<cite class="citation">
<span class="author vcard">Lorin Hochstein</span>,
<span class="author vcard"> Jeff Carver </span>,
<span class="author vcard"> Forrest Shull </span>,
<span class="author vcard"> Sima Asgari</span>,
<span class="author vcard"> Victor Basili</span>,
<span class="author vcard"> Jeffrey K. Hollingsworth</span>, and
<span class="author vcard"> Marv Zelkowitz</span>,
<a class="fn url" href="http://dx.doi.org/10.1109/SC.2005.53">HPC Programmer Productivity: A Case Study of Novice HPC Programmers</a>.
<cite class="container citation">
<a class="fn url" href="">Proceedings of ACM/IEEE Supercomputing Conference</a>
<abbr class="dtpublished" title="20051126T0000-0800">2005</abbr>
</cite>
page <span class="startpage">35</span>
<div class="publisher vcard">
<span class=" fn">IEEE Computer Society
</span>
<div class="adr">
<span class="locality">Washington</span>,
<span class="region">DC</span>
</div>
</div>
<a class="url instantiation" href="http://portal.acm.org/ft_gateway.cfm?id=1105800&type=pdf&coll=portal&dl=ACM&CFID=68330711&CFTOKEN=39187329">PDF of full text from ACM</a>
DOI: <a class="url uid" href="http://dx.doi.org/10.1109/SC.2005.53">10.1109/SC.2005.53</a>
Tags: <a href="http://citeulike.org/tag/productivity" rel="tag">productivity</a>, <a href="http://citeulike.org/tag/hpc" rel="tag">hpc</a>, <a href="http://citeulike.org/tag/performance" rel="tag">performance</a>
<blockquote class="abstract">In developing High-Performance Computing (HPC) software, time to solution is an important metric. This metric is comprised of two main components: the human effort required developing the software, plus the amount of machine time required to execute it. To date, little empirical work has been done to study the first component: the human effort required and the effects of approaches and practices that may be used to reduce it. In this paper, we describe a series of studies that address this problem. We instrumented the development process used in multiple HPC classroom environments. We analyzed data within and across such studies, varying factors such as the parallel programming model used and the application being developed, to understand their impact on the development process.
</blockquote>
</cite>
IMPORTANT
- sometimes, the citation must include Town/Country and Precise Date/Date Range, e.g.
- Gillespie SH, Dickens A. Variation in mutation rate of quinolone resistance in Streptococcus pneumoniae [abstract P06-17A]. In: Abstracts of the 3rd International Symposium on Pneumococci and Pneumococcal Disease (Anchorage, 5-9 May 2002).Washington, DC: American Society of Microbiology, 2002.
- Bassetti, M.; Righi, E.; Rebesco, B.; Molinari, MP.; Costa, A.; Fasce, R.; Cruciani, M.; Bassetti, D.; Bobbio Pallavicini, F. 44th Annual Interscience Conference on Antimicrobial Agents and Chemotherapy (ICAAC). Washington, DC; 2004. Epidemiological trends in nosocomial candidemia in ICU: A five-year Italian perspective.
- Peacock JE, Wade JC, Lazarus HM, et al. Ciprofloxacin/piperacillin vs. tobramycin/piperacillin as empiric therapy for fever in neutropenic cancer patients, a randomized, double-blind trial [abstract 373]. In: Program and abstracts of the 37th Interscience Conference on Antimicrob Agents and Chemotherapy (Toronto). Washington, DC: American Society for Microbiology, 1997.
Citing an external website
This is based on a formal citation of a website in the references section of a research paper, but could also be used for in-line links that had added information. Here's the original:
[25] David Stern, "eprint Moderator Model", http://www.library.yale.edu/scilib/modmodexplain.html (version dated Jan 25, 1999)
<cite class="citation">
<a class="fn url" href="http://www.library.yale.edu/scilib/modmodexplain.html">eprint Moderator Model</a>
<span class="author vcard">
<a href="http://pantheon.yale.edu/~dstern/dsbio.html" class="url fn">David Stern</a>
</span>
<abbr class="dtpublished" title="19990125T0000-0500">
Jan 25, 1999
</abbr>
</cite>