When the PRASA conference moved from Stellenbosch to Johannesburg in 2006, the HTML markup for its venue page changed from a simple <p> tag containing a single line of text to a table-based layout that spanned three columns. That seemingly minor edit — preserved in the Internet Archive’s snapshots — marks a turning point in how academic conferences represented physical locations on the early web, and why those representations matter for historians of pattern recognition research.
Why Venue HTML Matters for Conference History
The venue page of a conference website is often dismissed as a logistical afterthought: a block of text with an address, a map link, and perhaps a hotel recommendation. But for researchers tracing the evolution of a scientific community, venue HTML holds useful metadata. It reveals the conference’s institutional affiliations (university vs. convention center), the geographic spread of the organizing committee, and even the economic constraints that shaped the event. In the case of PRASA — the Pattern Recognition Association of South Africa — the venue markup also parallels the shift from static, hand-coded pages to more structured, database-driven sites.
The Mid-2000s Web: Tables, Fonts, and Inline Styles
Between 2003 and 2008, the typical academic conference website in South Africa was built with HTML 4.01 Transitional. Layout relied on nested tables, fonts were set with <font> tags, and venue information lived inside a <td> cell with a width="100%" attribute. For PRASA’s 2004 symposium — hosted at the University of Cape Town — the venue section looked like this:
<table width="600" border="0" cellspacing="5">
<tr>
<td width="150"><b>Venue:</b></td>
<td>Leslie Social Sciences Building, Upper Campus, UCT</td>
</tr>
</table>This pattern dominated until roughly 2007, when CSS-based positioning started to appear — but only after the PRASA 2006 proceedings showed the organizing committee had switched to a <div>-based layout for the first time.
Common Venue HTML Patterns in PRASA Sites (2001–2009)
By examining archived copies of the PRASA conference websites, we can identify four distinct markup strategies used over the decade:
- Plain text paragraph (2001–2003): Venue was a single
<p>element with no semantic structure. Example: “The symposium will be held at the University of the Witwatersrand, Johannesburg.” - Definition list (2004–2005): Some sites used
<dl>/<dt>/<dd>to label “Venue:” and “Date:” separately — an early attempt at machine-readable metadata. - Table cell (2004–2007): The dominant pattern, as shown above, with venue in a table row alongside a label.
- CSS-styled
<div>(2008 onward): A<div class="venue">with a background image of the conference center, often using absolute positioning for the text.
None of these patterns used any formal microformat or schema.org vocabulary — those standards were still embryonic. The closest thing to structured data was the occasional <meta name="geo.position"> tag, which appears in the PRASA 2009 site for the first time.
The 2006 Shift: Why Tables Persisted
Why did table-based layout persist for venue pages long after CSS became available? Two reasons: first, the conference websites were maintained by graduate students who reused templates from previous years; second, the venue information was often the last section updated, so it inherited older markup. The PRASA 2006 venue page, for instance, still used a <table> even though the rest of the site had moved to <div>s. That inconsistency is a gift to historians — it pinpoints exactly when the webmaster changed their workflow.
What Venue HTML Tells Us About Conference Logistics
The level of detail in venue HTML also reflects the conference’s budget and audience. Early PRASA venues (2001–2003) were described with just a building name; by 2005, pages included parking instructions, wheelchair accessibility notes, and even a small photograph of the entrance. The 2007 PRASA venue page, hosted at the CSIR International Convention Centre in Pretoria, included a hyperlink to a PDF floor plan — a rare inclusion that suggests the organisers expected a larger international delegation.
From HTML to PDF: The Venue in Proceedings
Venue information also leaked into the PDF proceedings themselves. The first page of many PRASA papers includes a footnote with the conference location — e.g., “Presented at PRASA 2008, Cape Town, South Africa.” In some cases, the PDF metadata (the /Subject field) contains a truncated venue string. These artifacts are useful for cross-referencing the HTML version of the venue page, especially when the website has been lost. A close reading of the prasa2012_13.pdf header, for instance, reveals that the venue listed in the PDF matches the HTML page only if you account for a typo in the building name — a detail that would be invisible without both sources.
The Legacy of Pre-Semantic Venue Markup
Today, conference websites routinely embed structured data using JSON-LD with Event schema, specifying location as a Place object with latitude, longitude, and address. But the mid-2000s era of loose, inconsistent HTML markup left a fragmented record. For researchers studying the geography of pattern recognition research in South Africa, those old <td> cells are primary sources — they document not only where the community met, but how the community represented itself on a web that was still learning to describe the physical world.
When the PRASA 2011 website finally adopted a <div id="venue"> with a class-based CSS layout, it marked the end of a nine-year tradition of ad-hoc venue markup. That <div> still exists, cached in the Wayback Machine, waiting for a future historian to parse its contents.