HyperPlex High-Precision Query-Response Knowledge Repository © Hyper Realm Consulting 2000-2010 © kenablersys 2010, to 2014 All rights reserved.
Sep 11, 2014
HyperPlexHigh-Precision Query-Response
Knowledge Repository
© Hyper Realm Consulting 2000-2010 © kenablersys 2010, to 2014 All rights reserved.
WE Information / Knowledge Seekers
Want answers to Our questions or queries
For that we need a good query facility
For PDF version click the link below
04 JAN 14HyperPlex (c) 2000-2014 All rights reserved
2
Knowledge Seekers
http://www.slideshare.net/putchavn/hyper-plex-high-precision-knowledge-authoring-queryresponse-system-06mar13
Kinds of Queries
Are many and
The expected responses are also many
Here we are considering queries that are:
Serious,
Precise,
Complex,
deep!
Like..
04 JAN 14HyperPlex (c) 2000-2014 All rights reserved
3
Typical Questions
1. Is Washington a city,
state, university,
person or an object?
2. How is electrical
current different
from voltage?
3. Since convex lenses are used
for sight correction in
presbyopia and hyperopia
are they the same? If NOT,
what is the difference and
why does it arise?
04 JAN 14HyperPlex (c) 2000-2014 All rights reserved
4
More Questions & Expectations
How is A different form B &
Why?
What is the cause of X and
what are the conditions for X
to occur or NOT to occur?
What is NOT P but may be
mistaken as P and Why?
Attributes of Answers
Precise
Relevant,
Latest and
Correct / Authentic
04 JAN 14HyperPlex (c) 2000-2014 All rights reserved
5
So what we need is:
A dynamic knowledge builder cum server or repository with
A High-Precision Query Facility
Giving Answers with specific attributes
HyperPlex (c) 2000-2014 All rights reserved
6
Let’s see what we have
04 JAN 14
Q & A
What or Who has the capability we need?
1. Books &
Libraries?Books just have text and graphics
2. Human
Experts?
Humans, YES they have knowledge and
interactive capability to deliver responses
3. Search
Engines?
So far, these are like 1 but are improving
with Semantic Web Technologies
04 JAN 14HyperPlex (c) 2000-2014 All rights reserved
7
Books & Libraries?
The seeker
has to be a
skilled
scholar
To find right
information
04 JAN 14HyperPlex (c) 2000-2014 All rights reserved
8
Yes, they served us well, But we had to work hard
Human Experts?
Professors,
Consultants,
Doctors,
Lawyers
Experts
Yes, they have knowledge
So far the best to deliver answers to queries
04 JAN 14HyperPlex (c) 2000-2014 All rights reserved
9
Query Elicitation by Human Experts
Good at finding out
What the information seeker is looking for
With minimum probing
In natural languages / body language
Then they present
Most relevant units of information
Closely matching the needs of information seeker
04 JAN 14HyperPlex (c) 2000-2014 All rights reserved
10
But this may NOT be true
with every human expert &
it may NOT uniform
Unfortunately…
Human Experts are
Not many
Even those few are not accessible
When one needs
Particularly to the millions of
information seekers
They are growing, as are topics
Alternatives?
04 JAN 14HyperPlex (c) 2000-2014 All rights reserved
11
Computers, Portals, & Search Engines?
Google, Bing,
Wolfram Alpha,
Information
Retrieval
Systems
Recognize only the letters of a word and process them (coarsely)
Their meaning is NOT addressed
Subjectively useful to SKILLED users
Statistics determine Relevance, Precision & Reliability of hits---NOT meaning of queries
04 JAN 14HyperPlex (c) 2000-2014 All rights reserved
12
The problem is with MEANING
Meaning is too complex
Five factors are involved
Machines, including software,
cannot address
Many aspects of meaning
Even approximately
04 JAN 14HyperPlex (c) 2000-2014 All rights reserved
13
See the PPT Semantic Web: The Need, Basics and Benefits
Certain aspects of MEANING can be
Something more
meaningful can still
be done
HyperPlex can
precisely answer the
typical queries
considered
Dealt with mechanically
That is, without human or artificial intelligence or understanding
That’s the scope of Semantic Web, RDF, OWL or UNL
But they have NOT done much
04 JAN 14HyperPlex (c) 2000-2014 All rights reserved
14
Words seen as just strings of letters
Most search engines have
Single field for query &
No way to qualify query terms
Terms like “Of, From, By, when”
are ignored
Only occurrences of words are
considered
Synonyms, or
equivalents are
NOT recognized
No aspects of
meaning of words
are addressed
HyperPlex (c) 2000-2014 All rights reserved
15
04 JAN 14
Too many hits of uncertain relevance
Search Engines give far
too many responses
Relevance ranking
criteria are
Not open & cannot be
defined by the user
Search results could be
irrelevant & misleading
Information seeker needs
To personally search again
To find relevant response
04 JAN 14HyperPlex (c) 2000-2014 All rights reserved
16
Single Field for query
1. Most search engines JUST provide a single field for query
2. Most Information Seekers CAN GIVE supplementary information to make their queries precise
3. But most search engines don’t accept 2 &
4. Can’t use that additional information
04 JAN 14HyperPlex (c) 2000-2014 All rights reserved
17
Query: ________________________________
Advanced Search
exists but is not
good enough
Semantic Web, KIF, RDF, OWLUniversal Network Language UNL
Have recognized the need
for dealing with meaning
Though they cannot capture
full significance of meaning
They promise meaning
centric applications
World Wide Web standards exist since 2000, 2004, 2011
Let’s see recent
Top Semantic Web applications
04 JAN 14HyperPlex (c) 2000-2014 All rights reserved
18
Top Semantic Web Applications
The latest available for 2010 at http://readwrite.com/2010/12/29/top_10_se
mantic_web_products_of_2010#awesm=~oqQ9
9xQoEriPqg
No products with
High-precision knowledge
encoding & query response
Meaning is central to
Queries & Relevant
Responses
Now machines can
mechanically make
sense of
Suitably encoded
knowledge / content
04 JAN 14HyperPlex (c) 2000-2014 All rights reserved
19
Winners of 2013 Sem Web Challenge
http://www.elsevier.com/about/press-releases/science-and-technology/winners-of-the-2013-semantic-web-challenge-announced-at-the-international-semantic-web-conference-held-in-sydney
Indicate that
Proposed HyperPlex has identified
A critical gap in Semantic Web & is
Filling it
04 JAN 14HyperPlex (c) 2000-2014 All rights reserved
20
A Superior Possibility: HyperPlex
HyperPlex is a
High-Precision
Query-Response
Knowledge Repository
Feasible & viable
We know & understand meaning
HyperPlex DOES NOT claim to Know or understand meaning
But it structures & parameterizes knowledge sufficiently
To make sense of queries & giveprecise meaningful answers
04 JAN 14HyperPlex (c) 2000-2014 All rights reserved
21
Precise Answers can only come from,
Precise Queries
So, HyperPlex has a
facility for
High-Precision Query
formulation
But that’s NOT all
The encoded knowledge
Must be fine-grained &
Must have the same
Knowledge structures &
Parameters of precision
Used in query
04 JAN 14HyperPlex (c) 2000-2014 All rights reserved
22
HyperPlex builds on RDF
HyperPlex uses Subject-Predicate-Object Structure of
knowledge representation of
RDF--Resource Description Framework
It is suitable but NOT sufficient
So, HyperPlex has added
Knowledge types & parameters
HyperPlex Goes beyond RDF to add
Expressive power &
Precision to
Encoded Knowledge
04 JAN 14HyperPlex (c) 2000-2014 All rights reserved
23
Knowledge Structures & Parameters of Precision
Each of Subject, Predicate & Object of RDF
Needs machine readable Typing (category) &
Special microstructures (meta tags) for each type
HyperPlex has added them
To encode & process meaning with precision
Still HyperPlex does not claim to
know or understand
meaning
04 JAN 14HyperPlex (c) 2000-2014 All rights reserved
24
HyperPlex
A Software Suite for K Building &
Delivery
HyperPlex is about High-Precision Q&A
04 JAN 14HyperPlex (c) 2000-2014 All rights reserved
25
Data & Information
Query
Knowledge
Seekers
Query Dialog Box
Auto-
Complete
One + Three Field Query
Struggled with typical search engines and
added what they lack
04 JAN 14HyperPlex (c) 2000-2014 All rights reserved
26
1+3 Field Query Box
Gen
Sub Pred Obj
High-Precisioncomes from
One General + 3 Special Fields
High-Precision 1+3 Field Query
Gen: A common field
Subject, Predicate & Object
3 special fields for -- RDF Triple
As you fill-in Gen, HyperPlex Auto-completes the other 3
Typing + microstructures of S, P & O help matching & drilling down to more details
04 JAN 14HyperPlex (c) 2000-2014 All rights reserved
27
Gen
Sub Pred Obj
1+3 Field Query BoxGen
Sub Pred Obj
Rapid Shortlisting of Subject Predicate & Object Options
Query & Response are interleaved
Response is refined by choosing auto-completed options of different fields
The parameters are many & fine-grained
04 JAN 14HyperPlex (c) 2000-2014 All rights reserved
28
HyperPlex
Shortlist of more relevant & precise options
Field
parameters
High-Precision Knowledge Authoring
Text, graphics, video & speech in
computers IS NOT Knowledge
Not machine
processable
That content has to be rewritten more
finely for machine processing
Knowledge
Authoring
Then high-precision Query & Response
are possibleThe need
04 JAN 14HyperPlex (c) 2000-2014 All rights reserved
29
RDF & UNL Also Represent Knowledge
1.But NOT with sufficient precision in our humble opinion
2.
HyperPlex
has
Special typing & micro-
structures for That is our IP.
We can license it
High-Precision encoding of
knowledgeSemi automated
04 JAN 14HyperPlex (c) 2000-2014 All rights reserved
30
Meaning & Understanding of text
There are many meanings of these words but no agreement on them
Understanding is a human ability to act on the meaning of text
It is out side the scope of machines and software
We consider a restricted
meaning of meaning for
humans & machines
Well-formulated text
prompts certain actions
Such action is the
meaning of that text
04 JAN 14HyperPlex (c) 2000-2014 All rights reserved
31
See the PPT Semantic Web: The Need, Basics and Benefits
Restricted meaning of meaning is used
1. In Semantic Web, RDF,
UNL etc.,
2. And also in HyperPlex
3. In both cases the S-P-O
structure & meta tags
4. Represent knowledge
5. There are defined ways
of acting on 3
HyperPlex has many types of
Subject, Predicate & Object
And each has many parameters
So, it is possible to encode
many details of knowledge in
machine readable form
And get precise response
04 JAN 14HyperPlex (c) 2000-2014 All rights reserved
32
HyperPlex Summary
1. High Precision Q-R
Query-Response is the
key
2. 1+3 FQ is something
special and unique
3. High-precision encoding
or Authoring of
knowledge is the
foundation for the Q-R
4. Special Structures and
Micro-structures make
this possible
04 JAN 14HyperPlex (c) 2000-2014 All rights reserved
33
Status and Plans
Claim: An improvement over Protégé
and Google, Yahoo in precision &
relevance --- ckeck it out
HyperPlex is the intellectual property
of joint project partners
Non-exclusively licensed
It is under implementation
04 JAN 14HyperPlex (c) 2000-2014 All rights reserved
34
Think
Confidential Information
will be disclosed under
non-disclosure agreement
Let’s Talk