PDA

View Full Version : Semantics, Vector Modeling and Fractals


orion
03-31-2005, 04:02 PM
NEURO-FRACTAL COMPOSITION OF MEANING: TOWARD A COLLAGE THEOREM FOR LANGUAGE (http://www.cs.wlu.edu/~levy/pubs/bics2004.pdf)
Simon D. Levy, from Washington & Lee University, Lexington, VA


Following the interest on self-similar patterns (fractals) in language and semantics, Levy's paper discusses the well known collage theorem and Iterated Function Systems (http://www.miislita.com/fractals/motifs-iterated-function-systems.html) applied to language, semantics and the Vector Space Model. Prof. Levy writes,

"Self-similarity in language appears in the guise
of stories within stories, or sentences within sentences (”I
know what I know”), and has been represented in the form
of recursive grammar rules by Chomsky and his followers.
Having observed this common property of language and images,
we present a formal mathematical model for putting
together words and phrases, based on the iterated function
system (IFS) method used in fractal image compression.
Building (literally) on vector-space representations of word
meaning from contemporary cognitive science research, we
show how the meaning of phrases and sentences can likewise
be represented as points in a vector space of arbitrary
dimension. As in fractal image compression, the key is to
find a set of (linear or non-linear) transforms that map the
vector space into itself in a useful way. We conclude by
describing some advantages of such continuous-valued representations
of meaning, and potential implications."


No, this is not another paper on fractal images. It is fractal theory applied to the Term Vector Space Model and IR in general. Another excellent contribution for the understanding of patterns in words, language and semantics.


Prof Levy concludes

CONCLUSIONS, IMPLICATIONS, AND FUTURE WORK

"Describing the linguistic composition of meaning with
fractals instead of grammars allows us to approach a number
of important questions in an entirely new way. For example,
it is generally agreed that the linguistic data to which children
are exposed is of insufficient quality to enable them to induce
general structural patterns without some pre-existing mechanism
for acquiring language (Chomsky 1965). The traditional approach
has been to view this mechanism as a sort of “Universal
Grammar” (more accurately, grammar schema) constraining the
sorts of languages that human beings can acquire."


"Under the approach described in this paper – where the
lexicon consists of co-occurrence vectors and the “grammar” is
encoded as a set of IFS neural network weights – this “poverty
of the stimulus” phenomenon can be viewed as follows: Essentially,
the problem is to find a set of tree vectors and network
weights such that the frontiers of the trees generated by the IFS
match the sentences (strings) to which the learner is exposed.5
We would like to offer, very tentatively, that the universal
mechanism by which such a process might be constrained could
be something like Barnsley’s Collage Theorem. That is, the
“correct” set of tree-vectors and weights could be those that produce
an IFS whose attractor covers the set of lexical vectors, so
that the IFS effectively maps the lexicon onto itself. The notion
that the child explores a set of “candidate grammar” hypotheses
while learning language could then be seen as an exploration
of the (real-valued) space of network weights. Again, this view
is supported by the work of Tabor (2000), who describes how
fractal encoding of grammars allows accepting machines for
those grammars to be located in a spatial relationship to one
another."


Enjoy it.


Orion

orion
04-01-2005, 04:48 PM
I just have the honor of receiving email from Prof. Levy. He is just a great and dedicated scientist. We need more scientists like him. I believe we all can learn a lot from the academic world.

Anyone interested along these lines of research and thoughts, feel free to contact me.

Orion

xan
04-01-2005, 07:40 PM
;) thank you Orion, we do our best!

I think that its quite an interesting take. It's an abstraction of language that this achieves in my opinion.

Thank you for the paper.

orion
04-02-2005, 12:26 AM
Oh, no problem, xan.

A lot of research work have been done on fractal semantics and the composition of meaning.


Orion

xan
04-03-2005, 08:29 AM
...and a lot of other research has also been carried out with other methods much less abstract than this. If you try and use this in machine translation for example, you will encounter problems with that.

Levy et al. refer to Tabor's work. His research showed that:

"Despite interesting results, this work is far from completion and raises challenging problems as research in fractals often does. Among the set of possible questions, one may wonder how the information of boundaries adjustment used during the correlation process (so-called cropping) can be added to the grammar. More generally, is it necessary to keep this information stored or is the structural information the only thing needed to reconstruct the fractal set?"

but he does agree that:

"As we pointed out in [4], a dual approach would be to study the importance of the function which maps the words to the geometrical space."

Tabor moved on from fractals since 2000. His work is now more centered around psycholinguistics and pure maths.
Tabor publications (http://www.sp.uconn.edu/~ps300vc/papers.html)

This method has been applied to atmospheric clouds and gave interesting results (NATO project).

Note: “grammar” - does this indicate that its not the best word for it?

"accpeting machines" refers to computers.

A context-free grammar is a grammar that generates a context-free language. Context-free grammars are also known as Type-2 grammars in the Chomsky hierarchy. The Chomsky hierarchy specifies Type-2 grammars as consisting only of production rules of the form ... "
Planet Math (http://planetmath.org/encyclopedia/ContextFreeLanguage.html)

I sometimes feel unsure whether the dicussion leans more towards computer language grammar.

A richer approach is needed to be satisfactory, as fractals do not take into account the relationships between the coponents. They miss out a lot about the actual nature of language, in the same way as we found that fuzzy logic did.

Lanuage has mathematical features, but it is just part of language. Using fractal methods like that omits to consider all of the other aspects.

orion
04-04-2005, 02:52 PM
I just received email communication from Hening Fernau, whose PhD thesis is on
Valuation of Languages, with Applicatios to Fractal Geometry (http://citeseer.ist.psu.edu/cache/papers/cs/1071/http:zSzzSzwww-fs.informatik.uni-tuebingen.dezSz~fernauzSzpub-fernauzSzpaperszSzpszSzval34.pdf/fernau93valuations.pdf). As Prof Levy, his work is enlightening.

There is another interesting work; Pavements as Embodiments of Meaning for a Fractal Mind (http://sphere.math.utsa.edu/sphere/salingar/Pavements.html) Here is an abstract

'This paper puts forward a fractal theory of the human mind that explains one aspect of how we interact with our environment. Some interesting analogies are developed for storing ideas and information within a fractal scheme. The mind establishes a connection with the environment by processing information, this being an important theme seen during the evolution of the brain. The authors assert that pavements play a role in connecting human beings to surrounding structures by acting as a vehicle for conveying meaning, and argue that the design on pavements transfers meaning from our surroundings to our awareness."

INTRODUCTION
FRACTALS AND HIERARCHICAL LINKING
THE CONCEPT OF MIND
MEMORY AND THE FRACTAL MIND
FRACTAL TUNING AND COMMUNICATION
PROBLEMS OF MISCOMMUNICATION
SHAPING THE BUILT ENVIRONMENT
PAVEMENTS AND HIERARCHY
THE IMPORTANCE OF MEANING STRUCTURES IN THE FLOORING OF URBAN SPACE
CONNECTION ESTABLISHES A PHYSIOLOGICAL STATE
THE NATURE OF MEANING
CONCLUSION: SOME GUIDELINES FOR PAVEMENT DESIGNS


Based on their work and communications with others IR folks. As IR and search engines move beyond on-page and off-page factors, the race is up for an unified fractal model that would integrate natural language, semantics, vector analysis, neuro-grammar, sentence construction (http://www.goertzel.org/books/complex/ch9.html) and affine concepts.



Orion

xan
04-04-2005, 03:49 PM
I'm not so sure....

but I'm at a pretty good conference at the end of the week, so I'll check it out for you.

orion
04-05-2005, 02:51 PM
I'm not so sure....

but I'm at a pretty good conference at the end of the week, so I'll check it out for you.

Enjoy your conference. What's its name?

Here is a delightful book on fractals and semantics as it relates to cultural diversity; the african culture in this case, but the overall findings can be found in other cultures. http://www.nexusjournal.com/reviews_v2n4-Bangura.html

A brief abstract of a review does not show all the merits of this research work, but it points to the obvious.

"The book is divided into three parts with 14 chapters. The first part introduces fractal geometry for people without any mathematics background, fractals in African settlement architecture, fractals in cross-cultural comparison, and intention and invention in design. The second part discusses geometric algorithms, scaling, numeric systems, recursion, infinity, and complexity. The third part focuses on theoretical frameworks in cultural studies of knowledge, the politics of African fractals, fractals in European history of mathematics, and futures for African fractals."

Orion

xan
04-05-2005, 07:26 PM
Not my kinda flight reading I'm afraid Orion, I'll stick to CL and IR for now!

The conferences is hosted by infonortics, its the search engine meeting:

"The Search Engines Meetings bring together commercial search engine developers, academics and corporate professionals to learn from each other."

http://www.infonortics.com/searchengines/

So its researchers, developers, and companies involved in the provision of search. Should be fun and some good people are there too.

orion
04-05-2005, 07:36 PM
Thanks

I know some of those folks. Some are from here in California. Avi also posted here http://forums.searchenginewatch.com/showthread.php?t=832&highlight=searchtools

Orion

orion
04-27-2005, 06:10 PM
Back to the original post, few weeks ago Prof Levy sent me email. His work on fractals and IR is really interesting. Here is another work in which fractal theory is applied:

Dynamical Parsing to Fractal Representations (http://delivery.acm.org/10.1145/550000/544442/p89-levy.pdf?key1=544442&key2=7204264111&coll=GUIDE&dl=ACM&CFID=11111111&CFTOKEN=2222222)

Interesting parser he has developed. Even arithmetic expressions are parsed. While far from perfect, in my view, it is a right step in the right direction. Levy says

"A connectionist parsing model is presented in which traditional
formal computing mechanisms (Finite-State Automaton;
Parse Tree) have direct recurrent neural-network
analogues (Sequential Cascaded Net; Fractal RAAM Decoder).
The model is demonstrated on a paradigmatic formal
context-free language and an arithmetic-expression
parsing task. Advantages and current shortcomings of
the model are described, and its contribution to the ongoing
debate about the role of connectionism in languageprocessing
tasks is discussed."

Orion