THE WELCH COMPANY
440 Davis Court #1602
San Francisco, CA 94111-2496
415 781 5700
rodwelch@pacbell.net


S U M M A R Y


DIARY: March 21, 1996 02:02 PM Thursday; Rod Welch

Began review Tom Launder's paper on "Plato's Problem".

1...Summary/Objective
2...Define Communication Metrics as New Management Science
............"Management Science" in this context is a set of related
.....The purpose of Communication Metrics is to capture and check the
3...Connectionist Theory May Support SDS Design for Leveraging Thinking
4...Learning Theory Based on Experience
...................Latent Semantic Analysis (LSA)
5...LSA applies a model of a homologue for associative learning.
6...Induction Fills In Missing Information
7...Not Enough Time to Be Confused
8...Latent Semantic Analysis (LSA)
9...Structure, Time, Experience
10...Investing Intellectual Capital Takes Time, But Yields Rewards
......Information, Knowledge, Wisdom, Vision,
......Unconscious Inducation is Subconscious Processing of Information
......The condition that induction is unconscious, occurring on
11...LSA Uses Matrix and Vector Analysis of Context to Learn Meaning
12...Reading and Learning
13...SVD - Singular Value Decomposition
......Encyclopedia Source for Text - a Form of Knowledge Space
......Meaning of Meaning Word Associations Context Everything Connected
......Everything Connected Everything New Information Changes Meaning
......Context Experience of One Word Improves Knowledge of All Words
......Meaning of Everything is Impacted by Anything That is New
......Communication Metrics creates "meaning" from causual association,
......Disambiguation of Context
......How Does LSA Know "Meaning" - Context and Patterns
......Statistical Mirage?
......Shared Meaning: Build and Maintain Common Usage Patterns
......Guessing and Intuition are General Measurements or Metrics
14...Theory Follows Use
15...Writing Text Using the Alphabet is Powerful Knowledge Tool

ACTION ITEMS.................. Click here to comment!

1...What is a "word type" -- need an example, does it just mean

CONTACTS 

SUBJECTS
Knowledge, Communication Metrics, Purpose
Plato's Problem, SDS Explanation
Management Science, Definition
Communication Metrics, Definition
Intelligence Incite Correlations

1307 -
1307 -    ..
1308 - Summary/Objective
1309 -
130901 - Follow up ref SDS 24 0980.
130902 -
130903 - Since Tom did not send the computer file for his printed paper,
130904 - requested by ref DIP 6, I scanned the document and entered it as a
130905 - publication....
130906 -
130907 -           A Solution to Plato's Problem:
130908 -           The Latent Semantic Analysis Theory of Acquisition,
130909 -           Induction, and Representation of Knowledge
130910 -           ref OF 6 0001
130911 -
130913 -  ..
130914 - Review of this paper shows research that supports SDS theory and
130915 - practice presented in the NWO paper on Communication Metrics.
130916 -
130918 -  ..
130919 - Define Communication Metrics as New Management Science
130920 -
130921 - I began a new Subject Index under Research for Communication Metrics
130922 - to set out the various aspects that need to be considered in defining
130923 - a "management science" per the analysis at ref OF 3 1110.
130924 -
130925 -            "Management Science" in this context is a set of related
130926 -            practices and responsibilities to accomplish a specific
130927 -            business purpose, ref OF 4 4224.
130928 -
130929 -        [On 960322 did more on this. ref SDS 27 8888.]
130931 -         ..
130932 -        [On 990215 received definition of "management science" which
130933 -        may be applicable. ref SDS 42 3510]
130934 -
130936 -      ..
130937 -     The purpose of Communication Metrics is to capture and check the
130938 -     connections of cause and effect that convert information on daily
130939 -     details of management into useful knowledge that enables timely
130940 -     follow up.  The short explanation is that Communication Metrics
130941 -     supports the understanding and follow up parts of communication
130942 -     that make leadership effective, see on 950327. ref SDS 5 8492
130944 -      ..
130945 -     This entails professional collaboration with decision makers and
130946 -     staff, and using technology to perform a "process" called the
130947 -     "Management Cycle" (similar to "continual learning") that improves
130948 -     understanding of correlations and implications by showing patterns
130949 -     of conduct and analysis over time from writing out chronologically
130950 -     details of real world, daily experience (a "diary") within a set
130951 -     structure relating external sources (called "References") and
130952 -     assigning classifications of "meaning" relative to organizational
130953 -     objectives called "subjects," which comprise a "knowledge space"
130954 -     to discover and correct ("metrics") errors before they impact
130955 -     performance and earnings, and discover opportunities that are
130956 -     otherwise overlooked.
130957 -
130958 -         [On 001219 steps using SDS new way of working. ref SDS 48 QT6F
130960 -      ..
130961 -     The process helps maintain "shared meaning," or alignment of
130962 -     understandings among team members overtime and so helps avoid
130963 -     knowledge creep, while improving the process of growing new
130964 -     knowledge by providing fast, accurate access to past experience
130965 -     linked to current work and plans for future action.
130966 -
130967 -        [On 960409 "Time" magazine article identifies the mental
130968 -        process of consciousness, described as "soul" in religion, and
130969 -        characterized as "turning water into wine." ref SDS 30 7211]
130970 -
130971 -
1310 -

SUBJECTS
Cognitive Science
Plato's Problem, Tom Landauer
Connectionist Theory

1605 -
160601 -  ..
160602 - Connectionist Theory May Support SDS Design for Leveraging Thinking
160603 -
160604 - Follow up ref SDS 24 0980.
160605 -
160606 - I am interested in support for ideas of "connectionist theory"
160607 - that appear to support a benefit SDS provides, originally considered
160608 - in reviewing Jeromy Campbell's book, "The Improbable Machine" on
160609 - 900303 at ref SDS 1 0000, on the brain using experience as "logic",
160610 - ref SDS 1 3002, converts information into knowledge, ref SDS 1 0052,
160611 - using a method described by "connectionist" theory. ref SDS 1 6006.
160613 -  ..
160614 - The power of writing is reported on 930112. ref SDS 2 3443, and on
160615 - 940609 Kissinger explains need for analysis. ref SDS 3 4238
160617 -  ..
160618 - The SDS design to integrate time, information and meaning, to chain
160619 - records automatically that emulate linked human experience, and to
160620 - permit directed links, and support access to links instantly so they
160621 - can be examined for verification and to stir the mind to investigate
160622 - further, all seem related to the ideas set out in Plato's Problem, ref
160623 - OF 6 line 28.
160625 -      ..
160626 -     See "Time" magazine article on the process of consciousness
160627 -     described as "soul" in religion and characterized as "turning
160628 -     water into wine." ref SDS 30 7211.
160629 -
160630 -
160631 -
1607 -

SUBJECTS
Induction, Continual Learning
Confusion, False Knowledge, Mistakes, Time
Induction is emulated by LSA
Time Impact on Learning/Knowledge
Continual Learning
Homologue of Associative Learning
Experience SDS/Communication Metrics Only Way to Know It
Delimma Knowledge Management

3410 -
341101 -  ..
341102 - Learning Theory Based on Experience
341103 -
341104 - The abstract begins with the question of learning theory that
341105 - attributes to an inherent "induction" mechanism in the human mind the
341106 - ability to enrich, ref OF 6 3019 and ref OF 6 6027, otherwise
341107 - improvished information cited by Jeromy Campbell in his book
341108 - "Improbable Machine," ref SDS 1 6831 and also at ref SDS 1 6689,
341109 - ref SDS 1 8526, by extrapolating and comparing new information or
341110 - stimulai with existing perceptions or "knowledge" that comes from
341111 - "experience." ref OF 6 4838
341112 -
341113 -     [On 960324 role of experience and how induction is a form of
341114 -     "guesing." ref SDS 28 0083.]
341116 -      ..
341117 -     [On 991124 Einstein noticed experience is foundation of knowledge.
341118 -     ref SDS 45 0001]
341120 -      ..
341121 -     [000331 Jack Park's The Scholar's Companion (TSC) seems like a
341122 -     program that may "grow" knowledge. ref SDS 46 0784
341124 -  ..
341125 - The paper describes a mathematical model that emulates reading, see
341126 - ref OF 6 8275, ref OF 6 5005, to explain how induction could occur in
341127 - the human mind, called:
341129 -                    ..
341130 -                   Latent Semantic Analysis (LSA)
341131 -
341132 -     ...ref OF 6 0002, ref OF 6 3071, ref OF 6 2002, see ref SDS 30
341133 -     2627.  Whether it merely emulates or is a direct expression of
341134 -     what the mind does is discussed at ref OF 6 3071.
341136 -  ..
341137 - A source for this paper on the web is....
341138 -
341139 -
341140 -      http://www2.rz.hu-berlin.de/inside/linguistik/institut/syntax/mind/landauer.htm
341141 -
341143 -  ..
341144 - This could support the idea of "continual learning" set out in the NWO
341145 - paper, ref OF 4 7502, as an innate process the mind performs
341146 - automatically, similar to the "management cycle" ...
341147 -
341148 -
341149 -                        plan, perform, report
341150 -
341151 -
341152 - ...described in the POIMS paper at ref OF 3 1104, which operates on
341153 - "automatic pilot" cited in POIMS. ref OF 3 6993
341154 -
341155 -      [On 960409 article in Time on "Consciousness." ref SDS 30 5001]
341156 -
341158 -  ..
341159 - LSA applies a model of a homologue for associative learning.
341160 - ref OF 6 7301
341161 -
341162 -     "homologous" -- corresponding or similar in position, value,
341163 -     structure or function.
341164 -
341165 -     "Associative learning" seems therefore to mean using dimension
341166 -     matching of words to induce meaning, so it is learning using
341167 -     induction from dimension matching, per example explaining the
341168 -     meaning of "meaning." ref OF 6 6351
341169 -
341171 -  ..
341172 - Induction Fills In Missing Information
341173 -
341174 - The process of inducation is summarized at ref OF 6 0005, as:
341175 -
341176 -      ...people have much more knowledge than appears to be present in
341177 -      the information to which they have been exposed.
341179 -      ..
341180 -     See also explanation that new knowledge units are built from old
341181 -     ones. ref OF 6 8500
341183 -  ..
341184 - Inductive processes of LSA depend on and accrue only to large bodies
341185 - of naturally interrelated data. ref OF 6 7301
341187 -  ..
341188 - The paper develops this point by observing that children seem to know
341189 - the meaning of a lot more words in the 7th grade than they have been
341190 - taught or otherwise encountered in speech, including television.
341191 - ref OF 6 0004, ref OF 6 7222  This proposition is derived from
341192 - vocabulary tests showing "knowledge" of far more words than expected
341193 - from forms of directed acquisition.  An assumption is made that the
341194 - only other source for learning the vocabulary is induction (also
341195 - "inference"), ref OF 6 3019, from reading, ref OF 6 5005, which is
341196 - accomplished by "dimension matching" or "dimension reduction" from
341197 - reading. ref OF 6 8444, ref OF 6 0819 also explained in the paper at
341198 - ref OF 6 7302
341199 -
341200 -      [On 960324 continue review induction process. ref SDS 28 8599
341202 -       ..
341203 -      [On 960518 induction explains why "truth" is a moving target in
341204 -      Communication Metrics theory. ref SDS 33 4488]
341206 -       ..
341207 -      [On 960324 role of experience and how induction is a form of
341208 -      "guesing." ref SDS 28 0083.]
341209 -
341211 -  ..
341212 - Not Enough Time to Be Confused
341213 -
341214 - The time and rate of flow of information is not treated with respect
341215 - to the exercise of this faculty, nor of the the ability to recognize
341216 - confusion, error, or to assume accuracy and move on, as discussed in
341217 - the NWO paper. ref OF 4 5203
341219 -  ..
341220 - Actually, "time" is cited as a factor in mental processing at ref OF 6
341221 - 7388 and ref OF 6 1626, see below, ref SDS 0 1234, and on 960325....
341222 - ref SDS 29 6666
341224 -  ..
341225 - Authors indicate LSA assumes a uniqueness which is essential, but not
341226 - apparent. ref OF 6 0370 and the next para, ref OF 6 0375
341228 -  ..
341229 - Possibly it is "time" as used in SDS.
341230 -
341231 -     [On 960325 time/information seem to factors. ref SDS 29 6666]
341232 -
341233 -     [See also that LSA findings support core idea of SDS to integrate
341234 -     time and information ref SDS 33 0550.]
341235 -
341236 -
341237 -
341238 -
3413 -

SUBJECTS
Latent Semantic Analysis (LSA) - Learning Theory
Time Impact on Learning/Knowledge
Induction is emulated by LSA

3705 -
370601 -  ..
370602 - Latent Semantic Analysis (LSA)
370603 -
370604 - This seems to be a mathematical model that applies matrix techniques
370605 - to evaluate frequency of text forms, called "similarity of words,"
370606 - relative to context patterns, or more simply, "text passages," to
370607 - infer meaning of words not previously encountered, ref OF 6 4961,
370608 - ref OF 6 7222,
370610 -  ..
370611 - Dynamic contextual disambiguation can be mimicked using LSA, but the
370612 - acquisition and representation of multiple meanings of single words
370613 - cannot. ref OF 6 6952
370615 -  ..
370616 - Context avoids ambiguity explained below. ref SDS 0 5588
370617 -
370618 -      [On 960324 LSA review continued. ref SDS 28 4158]
370619 -
370620 -      [On 960325 induction reviewed further. ref SDS 29 1100]
370621 -      ..
370622 -      So far I have not figured out how the model knows
370623 -      "meaning," as opposed to calculating frequency and similarty of
370624 -      text forms.
370626 -       ..
370627 -      The paper seems to indicate the LSA model computes contextual
370628 -      statistics, ref OF 6 line 109 and ref OF 6 line 683, ref OF 6
370629 -      line 961, ref OF 6 line 1876, ref OF 6 line 2111.  I still don't
370630 -      know how that relates to "meaning," which seems to me implies a
370631 -      correlation with an objective.  The model is evidently simply
370632 -      finding words have similar "meaning." ref OF 6 line 890. I guess
370633 -      another way to think of "meaning" is as a mind pointer, or guide
370634 -      that moves us toward wider understanding.
370635 -
370636 -         ("Meaning" is discussed in the next record segment, but see
370637 -         ref OF 6 line 888, also Landauer notes this is a natural
370638 -         question at ref OF 6 line 1592.)
370640 -  ..
370641 - The author's also call for "human vocabulary comprehension," ref OF 6
370642 - line 1044.
370643 -
370645 -  ..
370646 - The author's report the LSA model:
370647 -
370648 -     ... learns a great deal about word meaning similarities from text,
370649 -     an amount that equals what is measured by multiple-choice tests
370650 -     taken by moderately competent English readers, ref OF 6 line 1508.
370651 -
370652 -     LSA is a homologue of associative learning, ref OF 6 line 1744.
370654 -      ..
370655 -     Inductive processes of LSA depend on and accrue only to large
370656 -     bodies of naturally interrelated data, ref OF 6 line 1747.
370657 -
370658 -     ...shows that a substantial portion of the information needed to
370659 -     answer common vocabulary test questions can be inferred from the
370660 -     contextual statistics of usage alone, ref OF 6 line 683.
370662 -           ..
370663 -          This appears to question work that has supposed that language
370664 -          acquisition depends on specialized primitive structures and
370665 -          processes, ones that presume the prior existence of special
370666 -          foundational knowledge rather than just a general purpose
370667 -          analytic device, such as LSA, ref OF 6 line 112.
370668 -
370669 -     ...offers a general theory about all human knowledge acquisition,
370670 -     as a homologue of an important underlying mechanism of human
370671 -     cognition in general, ref OF 6 line 623.  In particular, the model
370672 -     employs a means of induction-dimension matching-that greatly
370673 -     amplifies its learning ability, allowing the model to correctly
370674 -     infer indirect similarity relations only implicit in the temporal
370675 -     correlations of experience, ref OF 6 line 121.
370677 -           ..
370678 -          This seems to be saying that what works for text may also
370679 -          work for other patterns the mind builds from its experience.
370680 -          see ref OF 6 line 599.
370682 -           ..
370683 -          I need to flesh the idea of "temporal correlations," see,
370684 -          however, "Structure and Time" below.
370685 -
370686 -
370688 -  ..
370689 - Structure, Time, Experience
370690 -
370691 - The model seems to require that new information be structured, so that
370692 - it can compare (i.e., dimension match), and that it have a large
370693 - reservior of experience from which to make comparisons. ref OF 6 7388
370694 -
370695 - Experience is discussed at ref OF 6 7331, ref OF 6 4838
370697 -  ..
370698 - The importance of "experience" reflects Campbell's book reviewed on
370699 - 900303. ref SDS 1 8526.  In SDS, "time" is a primary structure, i.e.,
370700 - chronology which reflects causal relationships.
370701 -
370702 -     [On 960518 LSA supports core concept of SDS to integrate time and
370703 -     information ref SDS 33 8811.]
370705 -  ..
370706 - This suggests the mind has an indexing method to quickly step through
370707 - an inventory to find patterns that hold the greatest promise of a
370708 - match.  Maybe this relates to the "temporal correlations." This could
370709 - be the "time" factor that seems otherwise to be missing from the
370710 - paper.
370712 -  ..
370713 - The mind searches first the stuff that is closest in time, i.e., fresh
370714 - in the memory, suggesting there is a stack of some kind.  My
370715 - experience using the Subject Index seems to reflect this property,
370716 - which is why today, on 960321, I have added the step of placing cross
370717 - pointers to other indexes. ref SDS 27 7749
370718 -
370719 -
370720 -
370721 -
370722 -
3708 -

SUBJECTS
Experience Improves Induction
Data and Information Converted into
Wisdom v. Common Sense
Investing Intellectual Capital
Learning, Continual
Explaining SDS, POIMS
Lift Capacity to Think, Remember and
Thinking Through Writing

5710 -
571101 -  ..
571102 - Investing Intellectual Capital Takes Time, But Yields Rewards
571103 -
571104 - LSA use of structure and large amounts of "experience," ref OF 6 3015,
571105 - also called "knowledge" ref OF 6 7331, ref OF 6 0660, explains in part
571106 - the phenomena of SDS where the connections and patterns simply grow
571107 - with time, creating a richer environment.  The author's offer this
571108 - rationale for why the LSA method has not received greater attention,
571109 - i.e., considerable effort and time is needed to get an LSA induction
571110 - model to the point of demonstrating significant returns. ref OF 6 3015
571111 - This supports Campbell's point in his book on artificial intelligence
571112 - reviewed on 900303, that humans are "experience" machines. ref SDS 1
571113 - 3002 see also ref OF 6 5021
571114 -
571115 -      [On 960624 Landauer discussed alphabet as a counterintuitive
571116 -      technology. ref SDS 34 1037]
571117 -
571118 - "Expertise" is, also, called "knowledge growth," and is shown to be a
571119 - function of having a lot of existing knowledge, called "experience"
571120 - ref OF 6 3881, derived from "unconscious induction." ref OF 6 7402
571121 -
571123 -       ..
571124 -      Information, Knowledge, Wisdom, Vision,
571125 -      Unconscious Inducation is Subconscious Processing of Information
571126 -
571127 -      The concept that a lot of experience aids the induction process
571128 -      of "thinking" (cited again at ref OF 6 0482) reflects the common
571129 -      occurrance that people with more experience are wiser and have
571130 -      better judgement.  It seems to support as well the more subtle
571131 -      point that the farther back in time we can look for patterns and
571132 -      correlations, the farther ahead we can see, so that wisdom and
571133 -      vision require experience, per analysis of continual learning on
571134 -      950830. ref SDS 8 0096
571136 -       ..
571137 -      Landauer's "unconscious induction" from experience, ref OF 6 7722
571138 -      supports the idea that writing and linking in SDS records that
571139 -      provides traceability to original sources, also, called
571140 -      "alignment," "context," and "understanding," lifts capacity to
571141 -      think, remember and communicate, per POIMS paper, ref OF 3 6528,
571143 -       ..
571144 -      This is supported by the J.C. Penny story on 950419 of a very old
571145 -      person having failing sight but improved vision, ref SDS 6 9933.
571146 -
571147 -         [On 960510 analysis of organic structure human mental patterns
571148 -         from "data" to knowledge, to vision, ref SDS 32 2222.]
571150 -       ..
571151 -      The condition that induction is unconscious, occurring on
571152 -      "automatic pilot," ref OF 6 7402, shows the value of SDS to lift
571153 -      capacity to think, remember and communicate by making the process
571154 -      more conscious to test and thereby measure the correctness of
571155 -      unconscious connections.  This makes writing and linking a
571156 -      "metric" of understanding that grows more accurate knowledge, per
571157 -      reasoning on 950223. ref SDS 4 8477 and also at ref SDS 4 3164
571158 -
571159 -         [On 960731 writing in SDS becomes faster and easier because
571160 -         the stuff that is already there is used to organize and
571161 -         structure the new stuff. ref SDS 37 6499]
571163 -          ..
571164 -         [On 960518 "Meaning Drift" explained. ref SDS 33 3734
571166 -       ..
571167 -      It offers another way to explain "metric" from work on 950223.
571168 -      ref SDS 4 3164.
571169 -
571170 -
571171 -
571172 -
5712 -

SUBJECTS
Marr, mathematical model for LSA
Mathematical Model of LSA's Induction
Grolier's Encyclopedia
Meaning Induced by LSA, Guessing; Reading
Meaning, Maintain Shared
Associations, Guessing and Intuition
Meaning, Maintain Shared/Common
Meaning Everything Changed by New Information
Disambiguity Solved by Context
Meaning, Build and Maintain Shared, Over Time
Landauer, Tom Association of Words in Context Establishes Meaning of

8113 -
811401 -  ..
811402 - LSA Uses Matrix and Vector Analysis of Context to Learn Meaning
811403 -
811404 - The mathematical matrix model called LSA uses vectors to represent
811405 - variable influences from informational input.  The LSA model is
811406 - attributed to work by Marr and Anderson, explained at ref OF 6 3311,
811407 - from a paper:
811408 -
811409 -       Psychological Description of LSA as a Theory of Learning,
811410 -       Memory. and Knowledge
811412 -  ..
811413 - On 950925 a book on technology for using the alphabet to form words
811414 - cited vectors to achieve representational flexibility in connectionist
811415 - models that are highly context dependent. ref SDS 9 4620
811417 -  ..
811418 - Matrix methodology is described at ref OF 6 2002 and at ref OF 6 7302
811419 - also ref OF 6 5083 on disambiguity through frequency of association...
811420 -
811421 -     Dimension matching is further explained on 960324. ref SDS 28 3250
811422 -
811423 -     They use an example of how often the word "model" appears in a
811424 -     paragraph.
811426 -      ..
811427 -     They calculate "event types" as "unitary chunks of perception or
811428 -     memory," ref OF 6 4750, ref OF 6 6839.
811430 -      ..
811431 -     Words are "short events" that occur repeatedly.  Paragraphs are
811432 -     long events that are never repeated. ref OF 6 7302
811434 -      ..
811435 -     Inductive processes of LSA depend on and accrue only to large
811436 -     bodies of naturally interrelated data. ref OF 6 7301
811438 -      ..
811439 -     LSA analysis transforms each cell entry in a matrix from the
811440 -     number of times that a word appeared in a particular context to
811441 -     the log of that frequency. This step approximates the standard
811442 -     empirical growth functions of simple learning.
811443 -
811444 -         The fact that this compressive function begins anew with each
811445 -         context also yields a kind of spacing effect (e.g., the
811446 -         association of A and B will be greater if both appear in two
811447 -         different contexts than if they each appear twice in the same
811448 -         context).
811450 -      ..
811451 -     LSA then divides each cell entry by the entropy for the event
811452 -     type, - E p log p over all its contexts.
811454 -      ..
811455 -     LSA assumes there must exist some way in which highly similar
811456 -     experiences attain unitary status by which a particular
811457 -     representational vector can be part of and be modified by
811458 -     different occasions, ref OF 6 9517
811459 -
811460 -          This may be the integration of time in SDS, where every body
811461 -          of information is uniquely associated with a time.  The mind
811462 -          is aware of time, because it is running out.
811463 -
811465 -  ..
811466 - Reading and Learning
811467 - SVD - Singular Value Decomposition
811468 -
811469 - This is explained as the method by which the matrix can simulate human
811470 - learning at ref OF 6 line 843, by "reading." ref OF 6 line 1029.
811472 -       ..
811473 -      Encyclopedia Source for Text - a Form of Knowledge Space
811474 -
811475 -      They used 4.6 million words from Grolier's encyclopedia arrayed
811476 -      in 30,473 columns by 60,768 rows, ref OF 6 line 860.
811477 -
811478 -      The columns corresponded to each of 30,473 separate articles in
811479 -      the encyclopedia.
811481 -       ..
811482 -      Each row represented a unique word type that appeared in at least
811483 -      two samples.
811484 -
811485 -           What is a "word type" -- need an example, does it just mean
811486 -           a word that appears in at least two of the articles?
811488 -       ..
811489 -      The matrix contained the frequency with which a particular word
811490 -      appeared in a particular text sample, ref OF 6 line 870.
811492 -       ..
811493 -      The cell entries were "transformed" by a calculation.
811495 -       ..
811496 -      The matrix was processed by SVD.
811497 -
811499 -       ..
811500 -      Meaning of Meaning Word Associations Context Everything Connected
811501 -      Everything Connected Everything New Information Changes Meaning
811502 -      Context Experience of One Word Improves Knowledge of All Words
811503 -      Meaning of Everything is Impacted by Anything That is New
811504 -
811505 -      The authors say...
811506 -
811507 -         ... a representation that captures much of how words are used
811508 -         in natural context captures much of what we mean by meaning.
811509 -         ref OF 6 5035
811511 -       ..
811512 -      Seems to reflect review on 950925. ref SDS 9 4620
811513 -
811514 -            [On 960324 "meaning" derived from experiential association
811515 -            differentiating context with classifications, similar to
811516 -            subjects. ref SDS 28 8844
811518 -       ..
811519 -      "Meaning" seems to be derived from language using induction
811520 -      processes explained above, ref SDS 0 7182, as dimension matching
811521 -      patterns of symbols (the alphabet) that can be mapped to patterns
811522 -      of sound from speech or language, shown on 960325, ref SDS 29
811523 -      1100, both associated with patterns of cause and effect from
811524 -      experience that associate results with needs and desires founded
811525 -      in human biology that seeks to live and reproduce, as explained
811526 -      in the paper, ref OF 6 7228.......
811528 -          ..
811529 -         SVD uses everything it can -- all linear relations in its
811530 -         assigned dimensionality -- to induce word vectors that best
811531 -         predict all and only those text samples in which the word
811532 -         occurs. (This expresses a belief that a representation that
811533 -         captures much of how words are used in natural context
811534 -         captures much of what we mean by meaning.)
811536 -       ..
811537 -      Contextural experience with one word improves knowledge of all
811538 -      other words. ref OF 6 4199
811539 -
811540 -            [On 960324 "context" explained, ref SDS 28 3250, and on
811541 -            960518. ref SDS 33 3374
811542 -
811543 -            [On 960324 LSA model applies dimension-matching induction
811544 -            that every node is related to every other through common
811545 -            condensed vector representations... ref SDS 28 8X50
811547 -             ..
811548 -            [On 960325 "meaning" constructed by "induction" using
811549 -            "dimension matching." ref SDS 29 1100
811551 -             ..
811552 -            [On 960518 "knowledge acquisition" by LSA method of
811553 -            constructing meaning. ref SDS 33 R158
811555 -             ..
811556 -            [On 960518 induction causes meaning drift. ref SDS 33 3734
811558 -             ..
811559 -            [On 980405 everything related to everything else, people
811560 -            worry context management and making connections with links
811561 -            to maintain alignment for accuracy is a hopeless task;
811562 -            overlooks long march of civilization using records
811563 -            management and library science to aid human cognition
811564 -            understanding complexity, even though every connection
811565 -            cannot be tracked. ref SDS 41 R63L
811567 -             ..
811568 -            [On 990317 glasses in a cupboard analogy illustrates
811569 -            intelligence process based on Landauer's induction theory
811570 -            that new information adjusts meaning of everything else.
811571 -            ref SDS 43 0594]
811573 -             ..
811574 -            [On 000515 19th century philosopher Charles Peirce
811575 -            indicates meaning derived from experience and
811576 -            communication. ref SDS 47 5026 and ref SDS 47 4078
811578 -             ..
811579 -            [On 000515 Peirce developed semiotic symbol system of
811580 -            existential graphs to improve alphabet technology for using
811581 -            words, ref SDS 47 Y34F and ref SDS 47 0663 that represent
811582 -            continuous thought. ref SDS 47 5G5N
811584 -             ..
811585 -            [On 031215 subject indexing difficult because information
811586 -            has multiple meanings based on context, excellent
811587 -            explanation with examples in Freeman's 1997 PhD thesis
811588 -            paper at Yale presenting Lifestreams to Professor
811589 -            Gelertner. ref SDS 49 K37H
811591 -       ..
811592 -      "Meaning" relates to "subject indexing" associations in the
811593 -      record of daily work, because new information impacts meaning of
811594 -      everything else. ref OF 6 6S5I  The authors explain that all
811595 -      nodes in LSA are related through "...common ...representations,
811596 -      not just through independently acquired pairwise node connections
811597 -      and their composite paths.  This process gives perceptual and
811598 -      observational learning, and the spontaneous generation of
811599 -      abstractions such as chunks, concepts and categories much greater
811600 -      force and flexibility. ref OF 6 6839
811602 -       ..
811603 -      A concept of "meaning" might therefore depend on "pattern" and on
811604 -      "frequency" and "proximity" of association with which information
811605 -      "chunks" occur.
811607 -       ..
811608 -      Recently, on 960304 review of computer history found an article
811609 -      by Vannevar Bush in 1945 who proposed using technology to meet
811610 -      requirements for constructing trails of associations that aid
811611 -      human thinking and memory for making sense of an expanding
811612 -      complex record of daily work. ref SDS 25 L47F
811614 -             ..
811615 -            [On 970324 the paper explains subject indexing and process
811616 -            of "catagorization" at ref SDS 28 8566.]
811618 -             ..
811619 -            [See meaning drift on 960518. ref SDS 33 4488, ref SDS 33
811620 -            3734.]
811622 -             ..
811623 -            [On 960809 "Meaning Space" explained. ref SDS 38 7860]
811625 -             ..
811626 -            [On 980405 Morris recognized the complexity of daily
811627 -            information because it impacts everything that is already
811628 -            known. ref SDS 41 2532]
811630 -             ..
811631 -            [On 990317 glasses in cupboard analogy illustrates process
811632 -            of intelligence based on Landauer's inducation theory that
811633 -            new information adjusts meaning of everything else.
811634 -            ref SDS 43 0594
811635 -
811637 -       ..
811638 -      Communication Metrics creates "meaning" from causual association,
811639 -      derived from sequence inherent in chronologies of events recorded
811640 -      by a diary.  This seems to a big part of "intelligence," which is
811641 -      what the body needs to survive, i.e., converting input into
811642 -      action as output.  SDS records are a "knowledge space" that help
811643 -      a busy mind "understand" cause and effect, and to plan follow up
811644 -      action.  Actions reflect actual understanding relative to
811645 -      percieved needs of the organism, i.e., a person.
811647 -       ..
811648 -      On 940609 explained SDS improves use of alphabet. ref SDS 3 8854
811649 -
811651 -       ..
811652 -      Disambiguation of Context
811653 -      How Does LSA Know "Meaning" - Context and Patterns
811654 -
811655 -      Meaning is discussed again, supplementing above, ref SDS 0 2882,
811656 -      and noting it is not an entirely settled question. ref OF 6 6351,
811657 -      the authors say...
811658 -
811659 -         Two of its most important aspects are usage and reference.
811661 -          ..
811662 -         The similarity relations between words that are extracted by
811663 -         LSA are based solely on usage. The underlying mathematics
811664 -         predicts the use of words in context, and the only reference
811665 -         of a word that LSA can be considered to have learned in our
811666 -         simulations is reference to other words. It might be tempting
811667 -         to dismiss LSA's achievements as a sort of statistical mirage,
811668 -         a reflection of the conditions that generate meaning, but not
811669 -         a representation that actually embodies it.
811670 -
811671 -            ["Context" is explained on 960324, ref SDS 28 3250, and on
811672 -            960518. ref SDS 33 3374]
811674 -             ..
811675 -            [On 000515 Peirce philosopher developed semiotic symbol
811676 -            system of existential graphs to improve alphabet technology
811677 -            for using words, ref SDS 47 Y34F and ref SDS 47 0663 that
811678 -            represent continuous thought. ref SDS 47 5G5N
811680 -          ..
811681 -         Given the strong inductive possibilities inherent in the
811682 -         system of words itself, as the LSA results have shown, the
811683 -         vast majority of referential meaning may well be inferred from
811684 -         experience with words alone. Note that the inductive leaps
811685 -         made by LSA in the simulations were all from purely abstract
811686 -         symbols to other purely abstract symbols. Consider how much
811687 -         more powerful word-based learning would be with the addition
811688 -         of machinery to represent relations other than gross
811689 -         similarity. But for such more elaborate mechanisms to work,
811690 -         language users must agree to use words in the same way, a job
811691 -         much aided by the LSA mechanism.
811693 -       ..
811694 -      Disambiguation of context is explained at ref OF 6 5083 and
811695 -      specifically at ref OF 6 6952.
811696 -
811697 -          Relates to multiple meanings of words, and sentences, which
811698 -          require context to resolve.
811699 -
811700 -          [On 980307 Andy Grove, CEO of Intel, takes copious notes to
811701 -          avoid ambiguity of mental maps.  Seems like an example of
811702 -          developing context. ref SDS 40 3668]
811704 -       ..
811705 -      Text comprehension by LSA is explained at ref OF 6 8452
811706 -
811707 -
811709 -       ..
811710 -      Statistical Mirage?
811711 -
811712 -      This issue is considered at ref OF 6 6351
811714 -       ..
811715 -      LSA does not solve the visual figure-ground, object-parsing,
811716 -      binding (but see conjectures below on unitization), and
811717 -      recognition parts of the problem, but even here LSA may
811718 -      eventually help by providing a powerful way to generate and
811719 -      represent learned and indirect similarity relations among
811720 -      perceptual features. ref OF 6 6201
811722 -       ..
811723 -      In LSA, if two words appear in the same context, and every other
811724 -      word in that context appears in many other contexts without them,
811725 -      the two will acquire similarity to each other but not to the
811726 -      rest, ref OF 6 1774, per Appendix figures 3 & 4.
811727 -
811728 -          [On 960809 "meaning space" reviewed. ref SDS 38 7860]
811729 -
811731 -       ..
811732 -      Shared Meaning: Build and Maintain Common Usage Patterns
811733 -
811734 -      Discussion of building shared meaning is at ref OF 6 8599.
811735 -
811736 -          It seems to support the notion that such meanings must be
811737 -          continually reinforced where there is the potential for other
811738 -          meanings from subsequent events to arise.
811740 -       ..
811741 -      Possibly the authors are saying that they are showing LSA can
811742 -      identify how a word is similar in meaning to another word based
811743 -      on common usage patterns.  Likewise meaning of causation or life
811744 -      objectives can be induced by dimension matching with other
811745 -      patterns stored in the mental experience base.  In effect the
811746 -      mind builds or constructs "meaning" as it goes along dimension
811747 -      matching everything it encounters.  How, though, does it select
811748 -      what to dimension match?  Must relate to what has come up
811749 -      recently and the biological state.  If so, it supports the
811750 -      Communication Metrics notion of building and maintaining shared
811751 -      meaning.
811753 -       ..
811754 -      It also seems to support the notion that writing should first and
811755 -      foremost be a process to discover, shape and preserve what we
811756 -      "mean," "know" and "understand," especially since "truth is a
811757 -      moving target."
811758 -
811759 -           [On 960518 support concept truth is moving target.
811760 -           ref SDS 33 4488]
811762 -       ..
811763 -      Guessing and Intuition are General Measurements or Metrics
811764 -
811765 -      My sense is that the authors theorize the mind recognizes, in
811766 -      some degree of accuracy, the frequency of using words in various
811767 -      contexts, and from that induces "meaning" without being directly
811768 -      "taught;" so, it teaches itself.  "Context" seems to be applied
811769 -      as the relationship between one word and another:  is it
811770 -      adjacent, next to, or several words away in some consistent
811771 -      pattern?   The author's say....
811772 -
811773 -         ...the mind (or brain) stores and reprocesses its input in
811774 -         some manner that has approximately the same effect [as SVD
811775 -         matrix modeling described in the paper], ref OF 6 5021
811777 -       ..
811778 -      This seems like a "guessing" method that relies on experience to
811779 -      establish correctness, per analysis of Jeromy Campbell.
811780 -      ref SDS 1 3002
811781 -
811782 -         [On 960324 more about "guessing." ref SDS 28 0083.
811783 -
811784 -         [On 960626 "intuitive" and the Internet, ref SDS 35 9594
811786 -          ..
811787 -         [On 960628 Internet, intuition and common sense. ref SDS 36
811788 -         0001
811790 -          ..
811791 -         [On 000515 Peirce evidently recognized guessing is a big part
811792 -         of human reasoning. ref SDS 47 5893
811794 -  ..
811795 - How the brain's biology might accomplish the matrix analysis of LSA is
811796 - set out at ref SDS 29 7335, is explained at ref OF 6 5021
811798 -  ..
811799 - Testing the model is explained on 960518. ref SDS 33 0001
811800 -
811801 -
811802 -
811803 -
8119 -

SUBJECTS
Usability Engineering
Use (Technology Application) Drives Theory
Theory, Science, Engineering, Use

8405 -
840601 -  ..
840602 - Theory Follows Use
840603 -
840604 - Like SDS, POIMS and Communication Metrics, the authors report the
840605 - theoretical underpinnings and implications with respect to cognitive
840606 - science of the LSA model arose from using it. ref OF 6 5296
840607 -
840608 - This reflects the point in Tom's book on the "Trouble with Computers."
840609 - ref SDS 7 8303
840610 -
840611 -
840612 -
840613 -
8407 -

SUBJECTS
Writing, Connecting, Alphabet
Pictures Not Text
Thinking Through Writing
Text is Powerful Knowledge Tool
Communication Metrics

9107 -
910801 -  ..
910802 - Writing Text Using the Alphabet is Powerful Knowledge Tool
910803 -
910804 - The "Plato's Problem" paper offers significant treatment of using
910805 - "text" as a primary knowledge acquisition medium.  The LSA method
910806 - seems to be predicated on how the mind can learn new words based on
910807 - reading text, ref OF 6 line 34.  I have not noticed explanations of
910808 - how this might apply to pictures, but presumably some of the same
910809 - theory would apply, since text is an assemblage of small pictures
910810 - called letters in the "alphabet."
910812 -  ..
910813 - Its power is in forcing the mind to build the larger picture and
910814 - thereby dig out of the mind more than can be consciously perceived in
910815 - the abscence of writing, per Kissinger's point at ref SDS 3 4238.
910816 -
910817 -     [On 960518 review Landauer's discussion of connected text as
910818 -     "context" and applyication of the alphabet. ref SDS 33 3374]
910820 -      ..
910821 -     [On 991108 history of alphabet foundation of civilization.
910822 -     ref SDS 44 5628]
910824 -  ..
910825 - This has importance to SDS because it is a knowledge tool predicated
910826 - on creating and arranging text, integrating it with time and organic
910827 - subject structure.  This is not generally a welcome prospect to busy
910828 - executives seeking an easy solution to the complex question of growing
910829 - useful knowledge.
910831 -  ..
910832 - Marketing trends to produce graphical solutions for computers, and
910833 - dialog skills from executive communication training, require credible
910834 - evidence for the value of using text, per explanation in NWO paper,
910835 - ref OF 4 1675, and analysis on 940609. ref SDS 3 4933, and on 930112
910836 - explaining the hisotry of writing. ref SDS 2 3443
910837 -
910838 -
910839 -
9109 -