On Embedding Machine-Processable Semantics into Documents

1
On Embedding Machine-Processable Semantics into
Documents

• Krishnaprasad Thirunarayan
• Department of Computer Science Engineering
• Wright State University
• Dayton, OH-45435, USA

2
Talk Outline

• Background and Motivation (Why?)
• Goals (What?)
• Details (How?)
• Conclusions

3
Background and Motivation
4

Content Extraction Formalize doc, using
controlled vocabulary
Heterogeneous Doc.
Spec. Defn. Rep.
5
Problems with this approach to content extraction

• Archiving spec (for human comprehension)
  separately from its formalization is not
  conducive traceability.
• Manual extraction from spec (from scratch) for
  each use is labor intensive, time consuming, and
  prone to typographical errors.

6
Observation

• Conceptually, every piece of information in an
  extraction owes its existence to a phrase in
  spec, and possibly, controlled vocabulary.
• So, explore techniques to maintain correspondence
  between a spec fragment and its formalization.

7
Goal
8
General Problem

• Embed domain-specific mark-up (annotations) into
  human sensible document
• to make explicit semantics of content text and
  complex data, and
• to augment an interpretation in a modular
  fashion.
• Document text Human comprehensible
• Semantic Mark-up Machine processable

9
Details (How?)
10
Nature of Specs

• Semi-structured
• Heterogeneous
• Text
• Tables
• Images
• Constrained technical vocabulary
• Available as MS Word document

11
Pre-processing Spec

• Abstract content from spec document by removing
  display oriented information
• Save text
• Save tabular data, preserving grid layout
• Retain links to images
• Note Save As text option in MS Word inadequate

12
Heterogeneous Document
13
XML generated by Majix
14
ASCII Output
15
Annotating Pre-processed Spec

• Embedding Machine Processable Semantics
• Recognizing and tagging text using controlled
  vocabulary
• By product of Document Indexing and Semantic
  Search
• Tagging tabular data to make explicit its
  semantics Same grid layout, but different
  interpretation and dependencies based on
  headings
• Explore XML-based programming language Water for
  defining data and its behavior (semantics)

16
Locating Controlled Vocabulary Terms
17
Example Table
Thickness (mm) Tensile Strength (ksi) Yield Strength (ksi)
0.50 and under 165 155
0.05 1.00 160 150
1.00 1.50 155 145
18
Example of Tagged Table

• Thickness (mm) Tensile Strength (ksi) Yield
  Strength (ksi)
• table.ltsetHeading thickness strength.tensile
  strength.yield/gt
• 0.50 and under 165
  155
• table.ltaddRow 0 0.50 165
  155 /gt
• 0.50 - 1.00 160
  150
• table.ltaddRow 0.50 1.00 160
  150 /gt
• 1.00 - 1.50 155
  145
• table.ltaddRow 1.00 1.50 155
  145 /gt ...

19
Example of Processing Code

• ltdefclass table rowsrequiredvector
  headingoptionalvectorgt
• ltdefmethod setHeading trequired tsrequired
  ysrequiredgt
• ltset headingltvector t ts ys/gt/gt
• lt/gt
• ltdefmethod addRow smin smax ts ysgt
• ltset rows
• table.rows.ltinsert ltvector smin
  smax ts ys/gt/gt/gt
• lt/gt
• ltdefmethod computeYieldStrengthgt
  lt/gt
• ltdefmethod computeTensileStrengthgt lt/gt
• lt/gt

20
(contd)

• ltdefclass table rowsrequiredvector
  headingoptionalvectorgt
• ltdefmethod computeTensileStrengthgt
• ltset tempfluid.Thickness/gt
• ltset i0/gt
• ltdogt
• ltuntil ltand temp.ltless table.rows.ltget
  i/gt.1/gt
• temp.ltmore_or_equal
  table.rows.ltget i/gt.0/gt /gt gt
• table.rows.ltget i/gt.2
• lt/untilgt
• ltset ii.ltplus 1/gt/gt
• lt/dogt
• lt/gt
• lt/gt

21
(contd)

• ltdefclass table rowsrequiredvector
  headingoptionalvectorgt
• lt/gt
• fluid.ltset Thickness0.60gt
• lttry
• ltset TensileStrengthtable.ltcomputeTensileStre
  ngth/gt/gt
• TensileStrength
• gt
• "TABLE out of range error occurred"
• lt/trygt

22
Water

• XML-based OO Scripting Language
• Facilitates creating Web Services
• Run methods remotely via web-browser
• Generalizes dynamic typing to constraint checking
• Conformance of actuals to formals

23
Pros and cons

• Encoding Improvement
• Amount of tagging can be controlled by suitably
  delimiting table data and annotating it with
  corresponding string-processing method
• Master Copy Update
• Changes to spec requires manual modification to
  archived annotated version.
• Irregular Tables in Specs
• Different units, etc

24
Some Related Work

• Microsoft Smart Tags
• Recognize controlled words in Office 2003
  documents and associate predefined list of
  actions with each occurrence
• SHOE
• Table data in a declarative (logic) language

25
Prolog rendition

• strengthTableRow( 0, 0.50, 165, 155).
• strengthTableRow(0.50, 1.00, 160, 150).
• strengthTableRow(1.00, 1.50, 155, 145).
• ...
• strengthTable(Thickness, TensileStrength,
  YieldStrength) -
• strengthTableRow(L, U,
  TensileStrength, YieldStrength),
• L lt Thickness, U gt Thickness.
• thicknessToTensileStrength(Thickness,
  TensileStrength) -
• strengthTable(Thickness,
  TensileStrength, _).
• thicknessToYieldStrength(Thickness,
  YieldStrength) -
• strengthTable(Thickness, _,
  YieldStrength).
• ?- thicknessToYieldStrength(0.6,YS).

26
Conclusions
27
A Step towards Holy Grail

• Ultimately enable authoring and/or extracting,
  human-comprehensible and machine-processable
  parts of a document hand in hand, and keep them
  side by side.