Introduction to Heml

Like any other form of human communication, the web comprises a good deal of information about the past. Web sites of historical interest range in size and ambition from vast scholarly endeavours, such Athenians, a record of all known people who lived in ancient Athens, to small-scale personal efforts, like Genjirou Inui's My Guadalcanal. Indeed, web materials in an historical mode are produced in and for nearly every discipline of the humanities and social sciences. As diverse as they are, these sites share one important similarity: as works of history, nearly all make assertions about events that took place in the past or argue about those events. The peculiarities of layout and presentation in each historical site obscures this similarity. For example, Le Musée virtuel de la Nouvelle-France provides a time-line with links, entitled Explorations européennes en Amérique ; in contrast, the Prosopography of the Byzantine Empire collects articles on individuals, with significant dates included in parentheses.

As a result, though there exists a wealth of electronic resources which represent historical events, including course materials generated with Critical Tools or genealogical information collected with GeneWeb, these are each sui generis, and their data cannot be compared or combined. Indeed, typically their data are accessible only in the form of text, either as html data, or in bound volumes typeset from the database. This is unfortunate since data recording historical events are rich with information, such as dates and locations, which can be manipulated by computational means.

If we consider the web to be no more than a huge keyword-indexed repository, then this difference in presentation matters little. Searches on an historical person or event are, after all, sometimes remarkably effective. Yet this approach to historical documents omits much. Ideally an electronic historical document includes interfaces for posing other sorts of historical questions, most obviously, searches for evidence pertaining to a given date or range of dates and to a given spatial range. Today's web cannot effectively respond to the question, "what resources are available which describe historical events in Europe during the first century B.C."

However, if the underlying similarity of historical web materials was exposed and interchanged in the web of the future using a shared semantic markup scheme, such queries on the web itself would be possible. Text markup schemes for the humanities, chiefly the Text Encoding Initiative, have been in use for over a decade. Ideally, such schemes also provide a standard and well thought-out basis for encoding such material, thereby aiding the scholar who marks up a text with them, as well as assuring her that the results will interact well with other material so encoded and be presented in a clear and consistent manner.

The potential for doing the same with historical events has been shown by Hockey et al. (1998). Yet the scholar working with this material currently has no such standard data formats established for her, and so each struggles alone against the pitfalls of designing schemata suited to this purpose. (Townsend provides guidelines for the larger question of marking-up of historical texts as documents. The European Community's ACO*HUM Working Group on History and Historical Informatics suggests that such standards are required in their Principia, but, as far as I know, there is no effort under way to provide them.)

In the past, humanities markup was been built around the Standard Generalized Markup Language, or SGML. However, SGML has proven too complex for direct presentation and manipulation on the web. Recognizing the power of generalized markup on the web, a group of experts have developed a new standard in markup called the eXtensible Markup Language, or XML. XML is a means of defining one's own mark-up and including, for instance, tags like <era> or <revision_number>. The excitement has built around XML's usefulness in business and literary studies, where a very rich mark-up language has already been composed by the TEI. The purpose of the current project is to define XML tags that describe historical events and to show how such markup can be transformed into useful and interesting presentations on the web.

An important extension of XML, known as XML Namespaces, allows parts of one markup scheme to be employed within another. With this advance, an entire document need not conform to one single schema; instead, relatively small XML elements pertinent to one domain can be embedded into a larger markup scheme, such as XHTML, the XML-compliant successor to HTML. A well-established example of this is MathML, the XML-based mathematics markup language. Elements prefixed with the MathML namespace can appear within XHTML documents, allowing the parser and display engine to treat these elements as mathematical notation.

It is the goal of this project to define XML elements that expose and outline historical events asserted in documents across the web and to parse and display these elements in interesting and useful ways.

How could these historical XML elements be used?

Consider, for instance, twelve diaries or memoirs, each a different view of the Siege of Sarajevo in 1994 and published on the web. They would include entries about events such as the stationing of troops or the evacuation of a certain region on such-and-such a day. If a standard form of mark-up were used to tag the text that recorded these events and their date and location, a computer could collect this information and associate it with the document. It would then be possible to search for descriptions of events on a certain day, or in a certain region or both, and retrieve references to the proper section of the pertinent memoir or memoirs.

Other alternative views of the events could be produced from these marked-up documents. It is common, for instance, for historical texts to include a chronological chart of events to aid the reader. From our twelve memoirs, each marked up in the same manner, we could create an exhaustive chronological chart detailing every event the texts' editors thought worthy of inclusion. Furthermore, the entries in our chronological chart would refer to their sources. Alternatively, if the mark-up scheme included information about the events' location, the documents could be used to produce historical maps or to search for events that took place in only a certain quarter of the city.

Consider now the usefulness of such a mark-up scheme if it were applied to even a fraction of the thousands of historical documents and web sites published on the web. It would be possible for a curious student to ask what happened worldwide between the years 1500 and 1200 BCE and to receive a list of events linked to scholarly sources or historical arguments. Time-lines and maps could be generated from disparate sources worldwide. Such a scheme would afford humanity a new and exciting means of communicating about its past.

Among historians and philosophers of history debate continues over what constitutes historical evidence and how it relates to the work of the practicing historian (Carr 1987:10-15; Elton 1969:112-13; Jenkins 1991:26, 48-50). This project does not seek to solve these problems, much less to impose one approach on the scholar who uses it; it hopes instead to provide a new medium for communicating about the past, one in which this debate can be continued. Though the concentration on `events' in this proposal might suggest that it is best suited for political history, we hope that non-traditional approaches to the study of the past will also be well served by this organizational tool, because it puts all encoded events on equal footing. For instance, so-called microhistory, whether temporal or spatial (Ginzburg 1993:14-15), can more easily be pursued in this organizational scheme, in which the life of a slave, recorded in Athenians, is on the same footing as any other event.

It is also true, as Zuern (1999) points out, that the typographical representing historical events is a rhetorical device. We hope that scholars using our tools will allow their audience to choose the format in which they view their information, opening it up to further interpretation and discovery.

This project provides a basic set of text mark-up and transformations for historical information. These include:

These constitute an integrated approach to the electronic representation of historical events; they make it possible to combine, compare and reuse sources for course materials, on-line resources and research project databases, providing this world with new views on its past.

The Heml Project's products are:

The Heml Project is exploring how best to make use of general metadata frameworks such as RDF and TopicMaps. Our first efforts in this will probably involve our jackdaw metadata files.

We have begun making use of Dublin Core metadata in our XHTML-extension documents, and notes relating to the applicability of Dublin Core elements appear throughout this document.

Heml markup has some core terms for, and approaches to, the problems of internationalization and references to unique identities. An introduction to these is useful before examining the basic elements of a Heml event.

As shown in the examples, all elements have the namespace http://www.heml.org/schemas/2002-05-29/heml. The prefix heml: is customarily assigned to this namespace. With the exception of xml:lang and xlink:href, attributes are not namespaced.

Element names always begin with a capital letter, and words within their names are written run-on, each with a beginning capital letter, such as PhysicalSourceSet.

Attributes are written in miniscule; new words within their names are separated with an underscore.

<Location uri="http://testing.com/#New_Carthage"
    xmlns="http://www.heml.org/schemas/2003-09-17/heml" >
  <LocationLabelSet>
    <Label xml:lang="en">New Carthage</Label>
    <Label xml:lang="la">Carthago Nova</Label>
    <Label xml:lang="ja">カルタゴ＝ノヴァ</Label>
    <Label xml:lang="el">Καινη` πόλις</Label>
  </LocationLabelSet>
  <!-- Additional Location information follows -->
</Location>

          
<Participants xmlns="http://www.heml.org/schemas/2003-09-17/heml">
  <PersonWithRole>
    <PersonRef uriRef="http://testing.com/#Hannibal"/>
    <RoleLabelSetRef uriRef="http://testing.com/#general"/>
  </PersonWithRole>
  <!-- Further Persons or PersonWithRoles here .... -->
</Participants>    

The label for each event is expressed in a EventLabelSet element; this groups together one or more heml:Label elements, each of which contains a description of the historical event in a brief headline-style entry suitable for inclusion in a table or printing on a map. heml:Labels have a xml:lang attribute to specifying the language of its label text. It is a schema error to have two Label elements with the same xml:lang value. heml:EventLabelSets use the 'Set' design described in the section called “'Sets'”. Example 3, “Sample EventLabelSet” illustrates.

<EventLabelSet xmlns="http://www.heml.org/schemas/2003-09-17/heml">
  <Label xml:lang="en">Ebro River Treaty</Label>
  <Label xml:lang="ja">エトロ川の協定</Label>
</EventLabelSet>
      

The Chronology element records what can be known about the time within or at which an event took place. It does not aim to be anything like a complete ontology of time; rather it is developed in conjunction with Java xslt extension elements that make sense of its markup, sorting and comparing events. The extension elements in turn are based on IBM's developerWorks' International Calendars in the ICU4J classes and on the Java Date class. At present, the only possible primatives are the XML Schema GYear and Date types, wrapped similarly named tags. However, the previous version of Heml accepted primatives from all calendars supported by ICU4J, and that will be made possible once again quite soon.

These primatives express a point in time, called an AbsoluteDate and illustrated in Example 5, “Sample AbsoluteDate Element”, or a DateRange, which represents a span of time, as shown in Example 6, “Sample DateRange Element”.

<Chronology>
    <Year>-0218</Year>
</Chronology>

   <Chronology>
     <Comment>1 Sierpnia 1944 - 30 Wrzensia 1944</Comment>
       <DateRange>
         <StartingDate>
           <Date>1944-08-01</Date>
         </StartingDate>
         <EndingDate>
           <Date>1944-09-30</Date>
         </EndingDate>
       </DateRange>
     </Chronology>

It therefore includes not only the year, month, day, etc. of the date, but also its calendrical scheme (i.e. Gregorian or Hebrew ), era (i.e. BC or AH ). Dates deriving from diverse calendrical schemes are thus accurately sorted and translated from one scheme, likely the one in which the source material was written, to another, likely the one with which the reader is most familiar. For example:

      <!-- The Battle of al-Yarmuk -->
      <!-- RECORDED IN ISLAMIC CALENDAR -->
      <date calendar="islamicsacred" era="AH" xml:lang="en">
        <year>15</year>
      </date>
      

The University of Virginia's Temporal Modelling Project in the summer of 2001 was most helpful in thinking through this element.

The Location elements appear in the LocationDefinitions element, one of the collections of definitions that appear at the head of the Heml document. It includes information sufficient for a rudamentary mapping front-end. It defines a LocationLabelSet to provide a name for the location and a geographic point expressed in Latitude and Longitude. Note that decimal lat/long values are used in the GeographicalHourLatitude and GeographicalHourLongitude elements.

<Location xmlns:geo="http://www.w3.org/2003/01/geo/wgs84_pos#"
xmlns="http://www.heml.org/schemas/2003-09-17/heml" 
uri="http://heml.org/docs/#Rome">                                 
  <LocationLabelSet>
    <Label xml:lang="en">Rome</Label>
    <Label xml:lang="la">Roma</Label>
    <Label xml:lang="ja">&#x30ED;&#x30FC;&#x30DE;</Label>
    <Label xml:lang="el">&#x1FEC;&#x1F7D;&#x3BC;&#x3B7;</Label>
  </LocationLabelSet>
  <geo:Point>
    <geo:lat>41.8172222222222</geo:lat>
    <geo:long>12.3188888888889</geo:long>
  </geo:Point>
</Location>

      

In accordance with the uri/uriRef pattern, Events would be described as taking place in this location if the Event element contained the following: <LocationRef uriRef="http://heml.org/docs/#Rome"/>.

The heml:Participants element contains information about people or groups that participated in an event and the role or roles each played. Person are represented with the heml:Person, located in the heml:Definitions/heml:PersonDefinitions element and illustrated in Example 8, “Sample Person Element”. A corresponding PersonRef element is used in a Event/Participants/ element to indicate that the specified person participates in the event.

<Person 
xmlns="http://www.heml.org/schemas/2003-09-17/heml" 
uri="http://heml.org/docs/FRAGLongus">
  <NameSet>
    <Name xml:lang="la">
      <NameElement ordinal="1">Sepronius</NameElement>
      <NameElement ordinal="2">Longus</NameElement>
      <NameElement ordinal="3">Tiberius</NameElement>
    </Name>
    <Name xml:lang="ja">
      <NameElement ordinal="1">&#x30BB;&#x30F3;&#x30D7;&#x30ED;&#x30FC;&#x30CB;&#x30A6;&#x30B9;</NameElement>
      <NameElement ordinal="2">&#x30ED;&#x30F3;&#x30B0;&#x30B9;</NameElement>
      <NameElement ordinal="3">&#x30C6;&#x30A3;&#x30D9;&#x30EA;&#x30A6;&#x30B9;</NameElement>
    </Name>
    <Name xml:lang="el">
      <NameElement ordinal="1">&#x3A3;&#x3B5;&#x3BC;&#x3C0;&#x3C1;&#x1F7D;&#x3BD;&#x3B9;&#x3BF;&#x3C2;</NameElement>
      <NameElement ordinal="2">&#x39B;&#x1F79;&#x3B3;&#x3B3;&#x3BF;&#x3C2;</NameElement>
      <NameElement ordinal="3">&#x3A4;&#x3B5;&#x3B2;&#x1F73;&#x3C1;&#x3B9;&#x3BF;&#x3C2;</NameElement>
    </Name>
  </NameSet>
</Person>

<RoleLabelSet uri="http://testing.com/#general"
     xmlns="http://www.heml.org/schemas/2003-09-17/heml">
  <Label xml:lang="en">General</Label>
  <Label xml:lang="la">Dux</Label>
  <Label xml:lang="ja">将軍</Label>
  <Label xml:lang="el">στρατηγός</Label>
</RoleLabelSet>

Persons in disparate documents will be reconciled in the database if they have the same uri.

This element collects references to a resource that the editor considers evidence for the asserted historical event. Since the same piece of evidence might appear in both physical form and in a networked resource, this element can comprise one or both of the PhysicalSourceSet and NetworkedSourceSet elements. These use the approach described in the section called “'Sets'” in order to refer to the document in various languages.

As Example 10, “An Evidence Element” shows, the PhysicalSource element uses namespaced bibliographic elements from DocBook. In truth, I have not found an XML Schema for DocBook, so I defined a small schema for the necessary db:bibliomixed elements. When the DocBook authority publishes an XML Schema, I will import it. It is likely that the namespace will change then, too.

<Evidence 
    xmlns="http://www.heml.org/schemas/2003-09-17/heml"
    xmlns:db="http://www.oasis-open.org/docbook/xmlschema/4.1.2">
  <NetworkedSourceSet>
    <!-- the href attribute value below has been abridged 
         to keep this example legible          
      -->
    <SimpleLink xml:lang="en" 
        xlink:href="http://etext.lib.virginia.edu/etcbin/toccer-new2...">
        Livy 25.8-10
    </SimpleLink>
    <SimpleLink xml:lang="la" 
        xlink:href="http://www.gmu.edu/departments/fld/CLASSICS/liv.25.html#8">
        T. Livius 25.8-10
    </SimpleLink>
  </NetworkedSourceSet>
  <PhysicalSourceSet>
    <PhysicalSource xml:lang="en">
      <db:bibliomixed>
        <db:author>Livy</db:author>. <db:pubdate>1982</db:pubdate>. 
        <db:title>From the Founding of the City</db:title>. Translated by B.O Foster. 
        <db:publishername>The Loeb Classical Library</db:publishername>. 
        <db:pagenums>vol. VI 367-377 (25.8-10)</db:pagenums>.
      </db:bibliomixed>
    </PhysicalSource>
    <PhysicalSource xml:lang="la">
      <db:bibliomixed>
        <db:author>Livy</db:author>. <db:pubdate>1982</db:pubdate>. 
        <db:title>Ab Urbe Condita</db:title>. Translated by B.O Foster. 
        <db:publishername>The Loeb Classical Library</db:publishername>. 
        <db:pagenums>vol. VI 366-376 (25.8-10)</db:pagenums>.
      </db:bibliomixed>
    </PhysicalSource>
  </PhysicalSourceSet>
</Evidence>

Version 0.5 of Heml includes an XSLT that filters Sets (see the section called “'Sets'”) according to their xml:lang attributes. It follows the following rules:

Events may be displayed in several different list views, including the event list in its entirety as well as lists defining events by participant, location and keyword. Single events may also be rendered in popup windows.

A table of events, chronologically sorted and (optionally) translated into a common calendrical scheme, with links to source materials.

A scalable vector graphics map labeled with the events in a given chronological range.

Shows points on a map representing locations for events spanning a given period of time. Mousing over these events will show event locations and a table of events that occured in these locations. Each event in the table contains a hypertext link to further event information.

Figure 1. Sample Heml Dynamic Map

	Moving your mouse over a location point will reveal the name of that location. Events that occured at that location will be listed in a floating table. Each event in the table contains a hypertext link to further information.
	This floating table contains events that occured at a selected location. The table may be dragged, minimized or closed. To open a closed table click on any location point on the map.

Animation runs a slider along a timeline, plotting events on a map as they occur. Each event listed on the map contains a hypertext link to more information on that event. The animation may be manipulated using a control panel which offers the ability to play, pause, stop, fast forward and rewind. The duration of the event may also be selected. The slider on the timeline is draggable by the mouse to the desired time.

Figure 2. Sample Heml Animated Map

	Panel allows you to control the animation via play, stop, pause, fast forward and rewind.
	Moving the slider with the mouse allows you to control the duration of the animation.
	You may also slide this pointer with your mouse to choose a desired place on the timeline.
	Events present as red during the time that they occured. Each event contains a hyperlink to further information.

Timelines are generated by a more complex process. XSLT has no access to information about font size and layout and so cannot effectively produce timelines of arbitrary size and scale. We therefore handle this transformation by a Java2D application, the org.heml.image.timeline package. Using the Batik library, any such application can output its image to an SVG document. The timeline package is handed the heml events though a DOM adaptor package.

Figure 3. Sample Heml Timeline

Heml 0.5 timeline representing events during Stalin's rule

This view provides an experimental raw RDF representation of the selected Heml data in xml form.

A future version of Heml will include the concepts of chronological or chronographic uncertainty, relative offsets and recurring classes of events, such as the coronation of a line of kings.

        <offset id="LifeExpect" type="+">
          <certainty value="95%"/>
          <year>80</year>
        </offset>
      

Though Heml today concentrates its efforts on text, it is just as possible to refer to sources or to annotate documents which are in visual or audio media. The appreciation of a song like Stan Roger's Northwest Passage could be improved with the addition of Heml tags, for instance. Presumably this would be an application of SMIL or similar technology.