Tools for Semantic Annotation

1
Tools for Semantic Annotation

• Atilla ELÇI
• Dept. of Computer Engineering
• Eastern Mediterranean University

2
Semantic Annotation

• TranscodingTransformation of info from one form
  to another
• Web content transcoding if along the Web
  transaction path
• Objective repurposing
• Available externally
• Cant be internally due to HTML being rigid.
• Even if inserted, not recognized (skipped) by
  browsers

3
Annotation Passin Ch. 4.

• Def. The act of adding notes (WordNet)

4
Annotation continued

• Semantic Web can play an important role in
  improving the role of annotations through
  following features
• Annotation discovery by agents
• Machine-understandable annotation
• Intelligent filtering of annotations (need
  annotation metadata)
• Improved searching
• Enhancement by agents
• Collaboration over annotations and supporting
  data
• Information extraction

5
Web Annotation Systems

• Inline editing download edit an HTML page and
  save in your domain
• External annotation browser first downloads a
  page and its annotation (from your domain or
  other server) then merges them on the fly and
  displays
• External server creates annotated view and serves
  it
• Important criteria to consider
• Ownership of annotations
• Tool becoming obsolete / unavailable
• Server scalability

6
Current Web Annotation Systems

• Wiki Collaboratives allows members to edit Web
  content on the fly. Check a dictionary.
  Examples
• The original Wiki by Ward Cunningham
• Wikipedia and others of Free Encyclopedia Project
• Ontolog Wiki
• W3C Annotea Project and Amaya editor-cum-browser.
  Uses RDF to describe annotations.
• Multivalent Browser
• Multivalent Home Page
• Wikipedia entry.

7
External Annotation Framework

• Ref. Hori and in ACM PD Bookshelf.
• Def. A scheme for representing annotation files
  and a way of associating original documents with
  external annotations.

• Contains metadata adressing part of the Web
  document.
• - XPath XPointer are used to link the two.

8
Annotation File

• RDF/XML syntax is used.

                                                
     Figure 14.2 Framework of external
annotation.
9
Annotation RenderingClient Preference
Capability

• Content adaptation requires dealing with Client
  Feature Set
• User preferences
• Device capabilities.
• W3Cs Composite Capability / Preference Profiles
  (CC/CP) is used in describing such information
  profiles.
• CC/PP specifies that client profiles can be
  delivered to a proxy server over HTTP.
• Thus proxy server is able to consider together
  the original document, annotation(s) on it, and
  clients CC/PP specs in transcoding the content.

10
Annotation-Based Transcoding System

• Transcoding Architecture Abstract architecture
  based on intermediary (proxy) between client
  server.

11
Ex Page Splitting

• Annotation vocabulary
• pcd consists of
• Alternatives,
• Splitting hints(a in the ex.),
• Selection criteria (b in the ex.)

12
Ex Page Splitting (continued)

• Adaptation engine

13
Information Extraction (IE) (Bontcheva et al
_at_Davies Ch. 3)

• A technology based on analysing natural
  language in order to extract snippets of
  information and produce fixed format, unambiguous
  data as output.
• Types of info sought (ref. Message Understanding
  Conference, MUC-7, definitions
• Entities (NE) such as people, places,
  organizations, quantities of commodities, dates,
  etc.
• Mentions (CO) places of references to entities
  in the text
• Descriptions (TE) of the entities
• Relations (TR) between entities
• Events (ST) involving entities
• Semantic annotation assigning to entities and
  relations in the text links to their semantic
  descriptions in an ontology.

14
HLT Semantic WebFigure 3.1 in Bontcheva et al
_at_ Davies
15
Applying IE in SemWeb Traditional Cases

• Traditional IE annotating with metadata
  ontology is not incorporated into annotated text
  such as Web pages
• AeroDAML annotation tool (2001) auto generates
  DAML annotations from Web pages
• Amilcare IE system (2003) a machine learning
  system that produces extracted info in triples
  for use by an annon tool
• MnM semantic annon tool (2002 semi-auto
  piggy-backed to Amilcare.
• S-Cream (2002) auto annon using Onto-O-Mat
  manual annon tool implementing CREAM framework
  for creating relational metadata, and Amilcare

16
Applying IE in SemWeb Ontology-Based Cases

• Ontology-based IE annotating using a formal
  ontology as one of the systems resources
• Pankow (Pattern-based Annotation through
  Knowledge on the Web, 2004) gathers surface
  pattern wrt a given ontology.
• ltinstancegt ltconceptgt, ltinstancegt is a ltconceptgt
• Checks validity through google queries
• Auto performance 24.9 against human perf of
  62.09 semi-auto 49.56
• SemTag large-scale semantic annon wrt TAP
  Ontology.
• KIM (Knowledge and Information Management system
  by OntoText Lab, 2005) now taken over by SEKT
  Project.

17
Semantic HTML?

• W3C has now issued a draft HTML 5 aiming to
  formalize semantic annotations as next version of
  HTML tag vocabulary
• Tony Patton's blog
• Differences of HTML 5 from prev version
• HTML5 specs

18
References

• Thomas B. Passin. Explorers Guide to the
  Semantic Web, Manning 2004. Ch4.
• Masahiro Hori Semantic Annotation for Web
  Content Adaptation, Ch. 14 in Spinning the
  Semantic Web, (Ed. Dieter Fensel et al.), MIT
  Press, 2003. (Check ACM Digital Library Books at
  Professional Development Center)
• Adobes annotation reference.
• Yuces thesis project.
• Bontcheva et al.(2006). Semantic Annotation and
  Human Language Technology. Ch. 3 in Semantic Web
  Technologies, Davies et al (eds). Wiley.