Abstract

While CMS publishing is bound to the features and limitations of a specific CMS, Web publishing often involves the integration of multiple information sources. Using Web technologies, Web publishing tasks can be solved efficiently and in a way which naturally leads to a publishing pipeline producing good Web content. Based on the context theme of the course, this lecture looks at various ways how context can be leveraged for Web-based publishing.

Linkbase Processing and Publishing

Publishing often assembles content from various sources
Context (as represented by links) can also have various sources
1. clients can use different linkbases (linkbase subscriptions)
2. linkbase servers may be implemented by aggregation
3. linkbase can use any data source which is then exposed as XLink
This publishing scenario uses a link processing pipeline
- having models and schemas for pipeline stages always is a good idea
- setting up example pipelines helps to explore the possibilities
- well-defined formats allow the modular composition of pipelines

Outline (Linkbase Processing)

Linkbase Processing [10]
1. ProtoLSL [9]
Linkbase Merging [5]
1. ProtoLML [4]
Link Creation [1]
Conclusions [1]

Classified Context

Linkbases are sets of links
- very often, all links in a linkbase use the same pattern
- based on a linkbase schema, link processing can be schema-based
Links are sets of classified resources
- there is no order of links in an XLink (which sometimes is a problem)
- the only semantics associated with a resource are its type
Arcs can be considered to be views of linkbases
- for one linkbase, there can be different views (different sets of arcs)
- a second component of a view is the current context (the arc starting point)

Outline (ProtoLSL)

Linkbase Processing [10]
1. ProtoLSL [9]
Linkbase Merging [5]
1. ProtoLML [4]
Link Creation [1]
Conclusions [1]

ProtoLSL Blink Example

<?xml version="1.0" encoding="UTF-8"?>
<lsl xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:noNamespaceSchemaLocation="protolsl.xsd">
 <resource type="post" title="Post" minOccurs="1" maxOccurs="1"/>
 <resource type="timestamp" title="Post Time" minOccurs="1" maxOccurs="1"/>
 <resource type="permalink" title="Post Permalink" minOccurs="1" maxOccurs="1"/>
 <resource type="relpost" title="Related Post"/>
 <resource type="site" title="Referenced Site"/>
 <resource type="blog" title="Blog Home" minOccurs="1" maxOccurs="1"/>
 <resource type="author" title="Blog Author" maxOccurs="1"/>
 <arc from="post" to="timestamp" title="Post Time"/>
 <arc from="permalink" to="timestamp" title="Post Time"/>
 <arc from="post" to="relpost" title="Related Post"/>
 <arc from="relpost" to="permalink" title="Related Post"/>
 <arc from="site" to="post" title="Blog Post"/>
 <arc from="post" to="permalink" title="Post Permalink"/>
 <arc from="post" to="author" title="Post Author"/>
 <arc from="permalink" to="author" title="Post Author"/>
 <arc from="post" to="blog" title="Blog Home"/>
 <arc from="permalink" to="blog" title="Blog Home"/>
 <arc from="blog" to="author" title="Blog Author"/>
 <arc from="author" to="blog" title="Authored Blogs"/>
</lsl>

ProtoLSL Address Example

<?xml version="1.0" encoding="UTF-8"?>
<lsl xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:noNamespaceSchemaLocation="protolsl.xsd">
 <resource type="name" title="Name" minOccurs="1" maxOccurs="1"/>
 <resource type="homepage" title="Home Page"/>
 <resource type="email" title="E-Mail" minOccurs="1"/>
 <resource type="phone" title="Phone"/>
 <resource type="fax" title="Fax"/>
 <resource type="affiliation" title="Affiliation" maxOccurs="1"/>
 <arc from="homepage" to="email" title="E-Mail"/>
 <arc from="homepage" to="phone" title="Phone"/>
 <arc from="homepage" to="fax" title="Fax"/>
 <arc from="hompage" to="name" title="Name"/>
 <arc from="affiliation" to="homepage" title="Staff"/>
</lsl>

ProtoLSL Validator

<?xml version="1.0" encoding="UTF-8"?>
<xsl:stylesheet version="2.0" xmlns:xsl="http://www.w3.org/1999/XSL/Transform" xmlns:xlink="http://www.w3.org/1999/xlink" xmlns:protolsl="http://dret.net/xmlns/protolsl/">
 <xsl:output method="text"/>
 <xsl:param name="lsl" select="'blinks+addresses.lsl'"/>
 <xsl:template match="/">
  <xsl:for-each select="descendant::*[@xlink:type eq 'extended']">
   <xsl:variable name="resources" select="./*[@xlink:type = ('resource', 'locator')]"/>
   <xsl:variable name="link-count" select="position()"/>
   <xsl:for-each select="document($lsl)//resource">
    <xsl:if test="count($resources[current()/@type = tokenize(@protolsl:type, '\s+')]) &lt; ( if ( exists(@minOccurs) ) then @minOccurs else 0 ) or count($resources[current()/@type = tokenize(@protolsl:type, '\s+')]) > ( if ( exists(@maxOccurs) ) then @maxOccurs else 999 )">
     <xsl:message>
      <xsl:value-of select="concat('Wrong number of &quot;', @type, '&quot; resources in link #', $link-count)"/>
     </xsl:message>
    </xsl:if>
   </xsl:for-each>
  </xsl:for-each>
 </xsl:template>
</xsl:stylesheet>

Linkbase Views

The pure linkbase needs only resources and no arcs
- LSL does not need to contain arcs
- arcs are only important for using links, not for creating them
There can be different sets of arcs (different views) for the same linkbase
- the navigable connections depend on the application
- for better separation of content and presentations, separate resources and arcs
- arcs should only be folded into the links once they become necessary

ProtoLAL

<?xml version="1.0" encoding="UTF-8"?>
<xsl:stylesheet version="2.0" xmlns:xsl="http://www.w3.org/1999/XSL/Transform" xmlns:xlink="http://www.w3.org/1999/xlink" xmlns:protolsl="http://dret.net/xmlns/protolsl/">
 <xsl:param name="lsl" select="'blinks+addresses.lsl'"/>
 <xsl:template match="/">
  <view>
   <xsl:for-each select="descendant::*[@xlink:type eq 'extended']">
    <xsl:copy>
    <xsl:copy-of select="@*"/>
    <xsl:for-each select="*[@xlink:type = ('resource', 'locator')]">
     <xsl:copy>
      <xsl:attribute name="label" namespace="http://www.w3.org/1999/xlink" select="@protolsl:type"/>
      <xsl:copy-of select="@*[name() ne 'protolsl:type'] | node()"/>
     </xsl:copy>
    </xsl:for-each>
    <xsl:variable name="current-link" select="."/>
    <xsl:for-each select="document($lsl)//arc">
     <xsl:if test="exists($current-link/*[@xlink:type = ('resource', 'locator')][current()/@from eq @protolsl:type])">
      <arc xlink:type="arc" xlink:from="{@from}" xlink:to="{@to}" xlink:title="{@title}"/>
     </xsl:if>
    </xsl:for-each>
    </xsl:copy>
   </xsl:for-each>
  </view>
 </xsl:template>
</xsl:stylesheet>

Outline (Linkbase Merging)

Linkbase Processing [10]
1. ProtoLSL [9]
Linkbase Merging [5]
1. ProtoLML [4]
Link Creation [1]
Conclusions [1]

Combining Schemas

Combining linkbases is combining contexts
- without some mapping information there is no matching
- mappings can provide information where the schemas overlap
Linkbase merging needs a well-defined way of defining mappings
1. two links are referencing the same resource (use the same URI)
2. the roles of the resource in the links have well-defined values

Outline (ProtoLML)

Linkbase Processing [10]
1. ProtoLSL [9]
Linkbase Merging [5]
1. ProtoLML [4]
Link Creation [1]
Conclusions [1]

ProtoLML Example

<?xml version="1.0" encoding="UTF-8"?>
<lml xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:noNamespaceSchemaLocation="protolml.xsd">
 <merge>
  <resource type="author"/>
  <resource type="homepage"/>
 </merge>
</lml>

The Full Pipeline

Individual linkbases (blinks/addresses) and schemas (blinks/addresses)
protolsl.xsl can be used to validate the linkbases
protolml.xsl merges them according to the merging rules
The merged linkbase can be checked with a merged schema
protolal.xsl the view based on the merged schema
xlink-analyzer.xsl visualizes all navigable structures
xlink2lil.xsl creates the interaction description based on the context
lil2html.xsl creates the YUI HTML code

Link Processing and Publishing

Outline (Link Creation)

Linkbase Processing [10]
1. ProtoLSL [9]
Linkbase Merging [5]
1. ProtoLML [4]
Link Creation [1]
Conclusions [1]

Interfaces from Schemas

LSL is a good foundation for generating interfaces
1. the extension grabs a LSL from a linkbase server
2. for creating links, the extension provides forms and drag&drop
3. if there are arcs in the schema, it can even visualize connections
Creating HTML forms from LSL is trivial (with some scripting help)
- most XLink resources are URI-identified resources
- a drag&drop interface (capturing the title as well) would be preferable

Outline (Conclusions)

Linkbase Processing [10]
1. ProtoLSL [9]
Linkbase Merging [5]
1. ProtoLML [4]
Link Creation [1]
Conclusions [1]

Publishing in Pipelines

Pipelines as a design principle are very flexible
XML pipelines can be distributed in many different ways
Publishing tasks usually need flexibility
More channels can be easily added in pipeline-based design

Publishing Pipelines

Web-Based Publishing (INFO 290-19)

Erik Wilde, UC Berkeley School of Information
2007-04-26

Abstract

Linkbase Processing and Publishing

Outline (Linkbase Processing)

Classified Context

Outline (ProtoLSL)

ProtoLSL Schema

ProtoLSL Blink Example

Blink Schema

Blink Instance

ProtoLSL Address Example

ProtoLSL Validator

Linkbase Views

Blink Schema & Arcs

ProtoLAL

Outline (Linkbase Merging)

Combining Schemas

Outline (ProtoLML)

ProtoLML Example

Merged Link Example

The Full Pipeline

Link Processing and Publishing

Outline (Link Creation)

Interfaces from Schemas

Outline (Conclusions)

Publishing in Pipelines

Publishing Pipelines

Web-Based Publishing (INFO 290-19)

Erik Wilde, UC Berkeley School of Information2007-04-26

Abstract

Linkbase Processing and Publishing

Outline (Linkbase Processing)

Classified Context

Outline (ProtoLSL)

ProtoLSL Schema

ProtoLSL Blink Example

Blink Schema

Blink Instance

ProtoLSL Address Example

ProtoLSL Validator

Linkbase Views

Blink Schema & Arcs

ProtoLAL

Outline (Linkbase Merging)

Combining Schemas

Outline (ProtoLML)

ProtoLML Example

Merged Link Example

The Full Pipeline

Link Processing and Publishing

Outline (Link Creation)

Interfaces from Schemas

Outline (Conclusions)

Publishing in Pipelines

Erik Wilde, UC Berkeley School of Information
2007-04-26