Übungsblatt 5

Vorlesung „Intelligente“ Systeme im WWW

Sommersemester 2002

Zum Online Bearbeiten am 10.7.

Aufgabe 1 (10pt. – also bitte mehr als 3 Sekunden denken)

Finden Sie Beispiele für Strukturelle, Beschreibende und Administrative Metadaten für folgendes Bild:

Carl Spitzweg

Der kurzsichtige Liebhaber

WV 1043

frühe Arbeit

Öl auf Leinwand, 46,0 x 34,0 cm

Museum Georg Schäfer Schweinfurt

bei Roennefahrt wird dieses Bild als "Die Angebetete" aufgeführt

Aufgabe 2 (10 Pt):

Der International Press Telecommunication Council (IPTC). Hat einen Metadaten-Standard mit gleichem Namen geschaffen, der bei der professionellen Informationsverarbeitung von Bildinformationen (bsp.: in Verlagen) genutzt wird. Er wid auch von Applikationen für die professionelle Bildbearbeitung unterstützt (zum Beispiel sind die Textinformationen in Adobe Photoshop eine Teilmenge der IPTC- Informationen).

Ordnen Sie die Metadaten-Attribute den Kategorien Beschreibend, Strukturell und Administrativ zu.

Objektname

Überschrift

Bildunterschrift

Besondere Hinweise

Objektbeschreibung

Bildrechte

Quelle

Copyright-Vermerk

Schlagwörter (Menge)

Titel der Bildunterschrift

Autor

IPTC Kategorie

IPTC Erstellungsdatum

IPTC Erstellungszeitpunkt

Freigabedatum

Freigabezeitpunkt

Versanddatum

Versandzeitpunkt

Ort

Staat/Provinz

Land(Code)

Land (Text)

Aufgeber-Code

Erstellungsprogramm

Programmversion

Job-Kennzeichnung

Service-Id

Bearbeitungsstand

Dringlichkeit

Objektzyklus

Referenzdienst

Referenzdatum

Referenznummer

Aufgabe 3 (30 Pt):

Für die Beschreibung von Nachrichtenartikeln existieren mehrere Standars, die versuchen

Metadaten zusammen mit den eigentlichen Nachrichten zu erfassen. Beispiele dafür sind NITF (http://www.nitf.org/) und NewsML (http://www.newsml.org/).

Erfassen Sie (mit Hilfe des beiliegenden NIFT Tutorial). Metadatenbeschreibungen für den folgenden

Artikel.

Weather and Tide Updates for Norfolk

By Alan Karben
NITF Network News Online

The weather was great today in Norfolk, Virginia. Made me want to take out my boat, manufactured by the Acme Boat Company .

Tides in Norfolk are running normal today. This weeks article highlights many of this week's fishing issues, and also presents a reference table of tide times.

The Tides are High

As can be seen from the table below, the shores of Oceanview again present the brightest spots for fishermen and sandcastle-builders alike.

	today		tide		tomorrow		next day		third day
beach	high	low	in	out	high	low	high	low	high	low
Sunset	30	14	09:23	18:51	28	11	31	12	33	9
Oceanview	31	15	09:25	18:56	26	11	31	11	31	9
Shellfish	29	15	09:25	18:53	26	9	29	11	30	11

Based on these tide tables, I believe you can see that this weekend stands to be an excellent one for small- or large-scale fishing exhibitions.

Photo: Volz

The tides, captured on film late yesterday.

There are many local nooks that fishing fans may want to keep a special eye one.

• Deer Creek: This area has proven to be a bass-lover's haven. Wilt Monthaven reports that examples of bass over 30 inches long have been reeled-in by both the casual and the professional angler.
• Fox Run: A bit more difficult, logistically, to navigate. However, the reports of Sturgeon in this area have put it on the fishing map for the first time since 1996.
• Pheasant Hollow: If you don't mind the crowds, this old favorite has come through this year. More fish than you can shake a stick at. Or a rod at.

Happy fishing everybody!

Stewart Klometers contributed to this article.

NIFT Elemente:

a	Anchor for hypertext links.
abstract	Story abstract
addressee	Person or organization to whom the postal item is being sent.
alt-code	An alternate symbol for the phrase.
bibliography	Free-form bibliographic data.
block	A group of releated containers.
body	The content portion of the NITF document.
body.content	Actual body content.
body.end	Information at the end of an article body
body.head	Metadata intended to be displayed to the reader.
bq	Block quote.
br	Forced line break.
byline	Container for byline information.
byttl	Byline title. Often contains an organization.
caption	Text for the caption of a table.
care.of	Poste restante.
chron	Date and time.
city	City, town, village, etc.
classifier	Generic holder for metadata. Could be used by researchers and archivists to qualify documents.
col	Column.
colgroup	Column group.
copyrite	Container for copyright information.
copyrite.holder	Copyright holder.
copyrite.year	Copyright year
correction	Correction information.
country	Geographic area with a government.
credit	Names the source of the block quote.
custom-table	A holder for a namespaced XML fragment for custom-tagged data, or for an alternative set of non-parser-breaking content.
datasource	Source of the information grouped in a block element.
date.expire	Date/time at which the document has no validity.
date.issue	Date/time document was issued.
date.release	Date/time document is available to be released.
dateline	Container for dateline information.
dd	Definition data.
del-list	Delivery trail of delivery services.
delivery.office	Postal city or town.
delivery.point	Street, PO Box No.
denom	Fraction denominator.
distributor	Information distributor.
dl	Definition list.
doc-id	Registered identification for document.
doc-scope	Indicates an area where the document may be of interest.
doc.copyright	Copyright information for document header.
doc.rights	Rights information for use of the document.
docdata	Document metadata.
ds	IIM Record 2 dataset information.
dt	Definition term.
du-key	Dynamic Use Key, created daily. Has tree structure indicated by defined form.
ed-msg	Non-publishable editorial message from provider or editor of item.
em	Emphasis.
event	An event.
evloc	Event location.
fixture	Specification for named document, such as Heard on the Street or On Language.
fn	Footnote.
frac	Fraction.
frac-sep	Fraction separator.
from-src	Delivery service identifier.
function	Role played by a person.
head	Holds metadata about the document as a whole.
hedline	Container for main headline and subheadlines.
hl1	Headline 1 (main-headline).
hl2	Headline 2 (sub-headline)
hr	Horizontal rule.
identified-content	Holds content identifiers that can apply to document as a whole.
iim	IIM Record 2 Data Container.
key-list	List of keywords.
keyword	Keyword. Can also be a phrase.
lang	Language identifier.
li	List item.
location	Significant place mentioned in an article.
media	Generalized media object.
media-caption	Text describing media.
media-metadata	Generic metadata placeholder.
media-object	Inline media data.
media-producer	Byline of media producer.
media-reference	Reference to an external media object, OR to its following media-object.
meta	A construct for sending generic metadata.
money	Monetary item.
name.family	Family name.
name.given	Given name.
nitf	The root element for NITF.
nitf-col	A holder for a namespaced XML fragment for custom-tagged data.
nitf-colgroup	A collection of nitf-col elements.
nitf-table	A holder for a table, and content-filled metadata.
nitf-table-metadata	A holder for a namespaced XML fragment for custom-tagged data.
nitf-table-summary	Textual description of the table.
note	Document cautionary note.
num	Numeric data.
numer	Fraction numerator.
object.title	Title of inline object such as book, song, artwork, etc.
ol	Ordered list.
org	Organization.
p	Paragraph.
person	Human individual.
postaddr	Mailing address.
postcode	Postal code.
pre	Preformatted information.
pronounce	Pronunciation Information.
pubdata	Information about specific instance of an item's publication.
q	Quotation.
region	Geographic area.
revision-history	Information about the creative history of the document; also used as an audit trail.
rights	Information on rights holder.
rights.agent	Rights agent.
rights.enddate	Rights end date.
rights.geography	Area to which rights apply.
rights.limitations	Limitations (exclusive / nonexclusive) of rights.
rights.owner	Rights owner
rights.startdate	Rights start date.
rights.type	Type of rights claimed.
series	Series information.
state	State or province or region.
story.date	Date of story.
sub	Subscript.
sublocation	Named region within city or state.
sup	Superscript.
table	Table of data.
table-reference	A pointer to a table that is elsewhere in the document.
tagline	A byline at the end of a story.
tbody	Table body.
td	Table data cell.
tfoot	Table footer.
th	Table header cell.
thead	Table heading.
title	Document Title.
tobject	Subject code.
tobject.property	Subject code property.
tobject.subject	Assigns subject information to news material based on a Subject Code system.
tr	Table row.
ul	Unordered list.
urgency	News importance.
virtloc	Virtual Location.

NITF is XML

NITF is an XML-conforming vocabulary. This means that NITF uses the constructs standardized by XML to describe elements of content within a document, and the descriptive attributes of that content.

For example, if a publisher wants to use NITF to distinguish a company name from the surrounding text, the <org> element would be used:

Today, <org>Microsoft</org> announced the release of....

If the publisher wants to embed Microsoft's NASD stock symbol, the value and org-id attributes of the <org> element would be used:

Today, <org value="MSFT" org-id="NASD">Microsoft</org>
announced the release of....

For more information on how XML works, visit our page listing XML resources.

This tutorial covers the most widely used sections of NITF. For details about each element, consult the NITF documentation page. Within this tutorial, elements and attributes are displayed in purple. Comments are included within the XML <!-- comment markers -->.

---

Basic Structure of NITF

NITF is divided into two sections, the <head> and the <body>.

<nitf>
 <head>
 <!--
 Metadata about the document as a whole
 goes here.
 -->
 </head>
 
 <body>
 <!--
 Contents for direct display to the user
 go here.
 -->
 </body>
</nitf>

In this respect, NITF is just like HTML -- But that's where the resemblance ends. Web authors use HTML to describe the display of their pages. NITF, on the other hand, is designed to describe the substance of news articles.

---

NITF <body>

The NITF <body> element is itself divided into three sections:

<body>
 <body.head>
 <!--
 This section holds core news components,
 such as headline and byline, that are commonly
 displayed before the text of an article.
 -->
 </body.head>
 
 <body.content>
 <!--
 This section holds the article,
 generally consisting of paragraphs of text,
 but perhaps with embedded tables, lists,
 photos, and other items. These can also be
 referenced by specifying a location on the
 Internet or another computer.
 -->
 </body.content>
 
 <body.end>
 <!--
 This section holds core news components
 that are commonly displayed at the end of
 an article.
 -->
 </body.end>
</body>

Here is an example of a simple NITF article:

<nitf>
 <head>
 </head>
 <body>
 <body.head>
 <hedline>
 <hl1>This is the main headline</hl1>
 </hedline>
 
 <byline>
 By Joseph Q. Reporter
 </byline>
 </body.head>
 
 <body.content>
 <p>This is the content of the first
 paragraph of the article.</p>
 
 <p>This is the content of the second
 paragraph of the article.</p>
 </body.content>
 </body>
</nitf>

---

Containers within <body.content>

Within body.content, NITF allows for several other types of text container at the same level as the paragraph. These include:

• <table>
• <list>
• <block> (groups together related content, for example: into a sidebar)
• <hl2> (a sub-headline)
• <media> (a photo or video clip, an associated caption, etc.)

---

Enriched Text

NITF contains many elements that can distinguish content appearing within headlines, bylines, tables, lists, and paragraphs. These "enriched-text" elements allow the publisher to index and highlight documents better, and add hyperlinks to richer sources of related and archival information.

Attributes of these enriched-text elements allow publishers to store -- inline with the text -- descriptive codes. These attributes can provide consistency and dependability among authors, writing styles, and languages.

Enriched-text elements provide facilities for marking up:

• people
• places
• companies and organizations
• values (such as numbers, money, and dates)
• titles and functions
• emphasized words
• hyperlinks

Here is a more enriched and expanded example of an NITF article:

<nitf>
 <head>
 </head>
 <body>
 <body.head>
 <hedline>
 <hl1>This is the main headline</hl1>
 <hl2>This is a sub-headline</hl2>
 </hedline>
 <byline>
 By <person value="JQR-412"
 idsrc="newspub-corp">
 Joseph Q. Reporter</person>
 </byline>
 </body.head>
 
 <body.content>
 <p>Today, <org value="MSFT"
 org-id="NASD">Microsoft</org>
 announced the release of....</p>
 <p>That company is <em>so</em> big
 that they....</p>
 <media media-type="image">
 <media-reference
 mime-type="image/jpeg"
 source="gates.jpg"
 alternate-text="Bill
 Gates at the podium."
 >
 </media-reference>
 <media-caption>
  Gates makes speech.
 </media-caption>
 </media>
 </body.content>
 </body>
</nitf>

More details on what is allowed in the NITF <body> element can be learned from our documentation and examples.

---

NITF <head>

The <head> element contains metadata that describes the article as a whole. It is divided into five main sections:

<head>
 <title>
 <!--
 A short, plain-text title of the document.
 Often used for display in a listing of
 search-results.
 -->
 </title>
 
 <tobject>
 <!--
 Codes for the type of article, and the
 subjects covered by the article.
 -->
 </tobject>
 
 <docdata>
 <!--
 Contains metadata about this document in
 particular. Includes publication date, an
 urgency rating, the column name (if it's
 a regular feature), series name
 (if it's part of a series of articles),
 and information on copyright ownership and
 distribution rights.
 -->
 </docdata>
 
 <pubdata/>
 <!--
 Information on where and how this article
 was published.
 -->
 
 <revision-history>
 <!--
 Specifies who revised the document, and why.
 -->
 </revision-history>
</head>

---

NITF <title>

The <title> element holds a plain-text string of characters that sums up, usually in one line, what the article covers. Most often, the <title> element is just the headline of the article, with any line breaks or enriched-text elements stripped.

Many NITF processing systems display a list of <title> elements when users search an NITF archive. Hence, it is important that this element be included.

Here is an example:

<title>
 President's State of the Union Address Makes Big Splash
</title>

---

NITF <tobject>

The <tobject> section (which stands for "topic object") contains elements that distinguish feature news stories from, say, news analyses or obituaries. It also holds elements that describe what subject the article is about, via a three-level Subject Codes taxonomy.

The IPTC has issued controlled Subject Code vocabularies that it strongly recommends the publisher use when including <tobject> elements. These vocabularies are available in several languages, and the IPTC is open to requests for additions to the list.

Here is an example:

<tobject>
 <tobject.property
 tobject.property.type="Wrapup"
 />
 
 <tobject.subject
 tobject.subject.refnum="4008006"
 // required
 
 tobject.subject.code="FIN"
 // optional three-letter code for Level 1
 
 tobject.subject.type="Economy, Business & Finance"
 // optional display-name for Level 1
 
 tobject.subject.matter="Macro Economics"
 // optional display-name for Level 2
 
 tobject.subject.detail="Foreign Exchange Markets"
 // optional display-name for Level 3
 
 />
</tobject>

---

NITF <docdata>

The <docdata> element (which stands for "document data") holds elements that identify, date, and register the article. It breaks out as follows:

<docdata>
 <correction
 info="fixes typo in paragraph 3"
 id-string="4000.21.a"
 />
 <!--
 Indicates this article was issued as a
 correction, and which document it corrected.
 -->
 
 <doc-id
 id-string="4000.21.b"
 />
 <!--
 ID of this document.
 -->
 
 <del-list>
 <!--
 Indicates who has delivered / distributed
 this article.
 -->
 <from-src src-name="ScreamingMedia"/>
 </del-list>
 
 <urgency
 ed-urg="1"
 />
 <!--
 Highest.
 -->
 
 <fixture
 fix-id="Investigative Business Feature"
 />
 <!--
 The name given to a regular feature.
 -->
 
 <date.issue
 norm="2000-05-21T09:05:00-5:00"
 />
 <!--
 Date the item was published.
 -->
 
 <du-key
 generation="3"
 part="2"
 version="12"
 key="microsoft-trial"
 />
 <!--
 Provides a mechanism for grouping related
 stories.
 -->
 
 <doc.copyright
 year="2000"
 holder="The Toronto Globe & Mail"
 />
 <!--
 Who owns the copyright to this document.
 -->
 
 <key-list>
 <!--
 Holds a set of keywords relevant to the article.
 -->
 <keyword key="software"/>
 <keyword key="antitrust"/>
 </key-list>
</docdata>

---

NITF <pubdata>

The <pubdata> element (which stands for "publication data") has attributes that identify and date and register the article. It breaks out as follows:

<pubdata
 type="print"
 name="The Toronto Globe and Mail"
 date.publication="20000521"
 />

---

NITF <revision-history>

The <revision-history> elements provide a creative history of the document. It breaks out as follows:

<revision-history
 name="Pat Q. Editor"
 function="editor"
 norm="20000521T08:43:00Z"
 comment="fixed rampant typos"
 />

Aufgabe 4 (10 Pt)

(a)

Diskutieren Sie die Vor- und Nachteile von XML-basierten Metadatenformaten (wie NIST) gegenüber

RDF-basierten Metadatenvokabularen (wie RSS).

(b)

Diskutieren Sie, inwiefern die Herausforderungen an Metadatenformate (Slide 45 – Metadatenfoliensatz) von den Sprachtypen addressiert werden. Zur Erinnerung:

- Konsistenz

- Proper Reference (schwer zu übersetzen, sorry)

- Vermeidung von Redundanz

- Relationierung / In Bezugsetzung von Metadaten untereinander

- Wartbarkeit

- Einfachheit der Nutzung

- Effizienz