The Open World: The Significance of Constructivism to the Web

Harry Halpin, <H.Halpin@ed.ac.uk>

Textorized Parthenon

Ph.D. Presentation, Oxford Internet Institute, July 2008

The Semantic Web

Is a open world and universal space for machine-readable data.

To a computer, then, the web is a flat, boring world devoid of meaning...This is a pity, as in fact documents on the web describe real objects and imaginary concepts, and give particular relationships between them...Adding semantics to the web involves two things: allowing documents which have information in machine-readable forms, and allowing links to be created with relationship values.TimBL, WWW1994

RDF data...

a set of circles and arrows

...merges just like that.

more circles and arrows superim

Subject and object node using same URIs

Links Join Across Different Sources of Information

 Links between column headings

Verb/predicate/Property using same URIs

A Day in the Life of the Semantic Web

  1. I give a presentation at a conference at you get a new URI for myself
  2. Your Facebook profile produces FOAF with a URI for you.
  3. I have my own FOAF file with my own URI for myself.
  4. The Semantic Web Community Wiki has a URI for me.

None of these URIs match.

Not so fast...

httpRange-14: Everything Must Have a URI?

The Eiffel Tower versus a web-page about the Eiffel Tower: Isn't there something fundamentally different? And what URI do you use for the Eiffel Tower qua Eiffel Tower?

The Symbol Grounding Problem?

Eiffel Tower Homepage

Eiffel Tower Picture

The URI of the web-page is clearly http://www.tour-eiffel. fr/.

Introduction

Background

The Semantic Web uses URIs not just to refer to web-pages, but to things.

For the Semantic Web to work, there must be agreement from decentralized agents on what a URI.

Thesis

URI agreement must take place in a decentralized manner in an analogous manner to how humans come to agreement to terms in natural language.

Approach

We're going to have use an interdisciplinary approach from two rarely connected fields: philosophy and the World Wide Web.

Research Question

  1. How do humans come to agreement on new terms in a decentralized manner?
  2. Can this be formalized?
  3. Can we create a usable program to encourage URI re-use?

Thesis Structure

  1. Introduction
  2. Significance of the Web
  3. Ontology and Principles of Web Architecture
  4. Theories of Naming and Reference for the Web
  5. Generation and Disambiguation of Names on the Web
  6. Experiment
  7. Results
  8. Discussion
  9. Conclusions

Information, Content, and Encoding

Information is portable, modular, public, and about something (i.e. has a meaning or interpretation).

Key Observation: Telling someone how many bits they have does not tell them what information was transmitted.

Dretske's Definition of Information Content:

"A signal r carries the information that s is F iff the conditional probability of s's being F, given r and k is 1 (where k is the knowledge of the receiver)" (Dretske, 1981).

info theory resource

What are Representations?

Haugeland states that the mark of a representation is the standing-in rel ationship (Haugeland, 1991).

  1. Presentation: Process S is in effective local contact with pro cess 0.
  2. Input: An input procedure of S puts "memory" R in corre spondence with (at least partially) process 0. This is entirely non-spook y since A and B are in effective local contact.
  3. Separation: Processes S and 0 change in such a way that the processes are non-local.
  4. Output: Due to some local effect in process S, S uses i ts local effective contact with R to initiate the local dynamic behavior that depends on R for success.
Representation is information with distal (non-local) content.

Content Negotiation

A resource is that content which has invariant content over time. URI Resource Rep diagram

Referent

referent

The Extended Mind Thesis

Extended Mind Thesis: Otto suffers from Alzheimer's disease, and like many Alzheimer's patients, he relies on information in the environment to help s tructure his life. Otto carries a notebook around with him everywhere he goes. When he learns new information, he writes it down. When he needs some old information, he looks it up.

(Clark and Chalmers, 1998)

Andy Clark and Dave Chalmers

What is the difference between using the notebook and an act of memory?

Traditional mental characteristics of concepts, belief, knowledge, attitu des, propositions is therefore extended into the environment.

Otto Redux

Imagine Otto trying to find his way to the Museum of Modern Art, and instead of a notebook having a personal digital assistant with access to a map on the Web.

Palmtop

Inga can have access to the exact same map via her personal digital assistant.

Since both Otto and Inga are sharing the same representation and using it in the same manner, Inga and Otto can be said to share at least some of the same cognitive state, due to the fact that their individual cognitive states are causally dependent on accessing the same representation.

Language, and external media (including the Semantic Web) is for co-ordinating action

Kripke's Causal Theory of Names

A name refers to same referent in all possible worlds and is so independent of any logical description.

A name can be transmitted through time via historical and causal chains, with a name being given its original referent through a process Kripke calls baptism. Saul Kripke

A URI is given its meaning by a baptism by its owner, such as registering a domain name.

OKKAM

OKKAM is a huge EU-funded project that has a Kripkean theory of names.

It will mint new URIs for all named entities on the Web, with links to those names from web-pages it has mined, but no logical statements and (as of yet) no disambiguation.

Saul Kripke

Russell's Descriptivist Theory of Names

Names are shorthand for logical expressions - so the name "Eiffel Tower" includes a host of facts about it such as "in Paris" and "completed in 1889". Bertrand Russell

Patches of sense-data known through direct acquaintance allow one to ground the atoms of logical statements or create descriptions that can form the basis for names.

The priority of reference over sense

We are also assuming the real world (as opposed to any logical idealization of it) is divided neatly into individuals (Strawson).

But how can we deal with referents in the past, or imaginary referents?

URI Declarations

When an agent uses a URI, it is forced to "follow its nose" to the URI, download all the logical statements at that URI (or some sort of ontological closure), and accept them.

Small variation by Peter Patel-Schneider: An agent should not believe any statement that is not explicitly owl:import.

David Booth

Frege and Meaning

Frege, the father of logic, posits that the actual thing in the world is the referent, and a name is a symbol that identifies a referent(s). The sense is the mode of presentation, a type of public, objective(?), knowledge about that private concept among a shared community. The third party of sense (meaning) mediates the reference relationship.Gottlieb Frege

The priority of meaning over reference

Example:Hesperus has a sense ("the morning star") different from that of Phosphorus ("the evening star"), yet both have the same referent, the planet Venus.

As formalized in logic, meaning is taken to be given by interpretations that define the truth conditions of the statement.

Ambiguity is built in: The web of logical statements is the bearer of meaning, and whatever satisfies the sense of the statement could be a referent.

Interpretation

interpretation example

Syntax and Semantics Example: The syntax of the numeral 2 is about the number two. The interpretation of numerals to numbers is arbitrary. 11 is three in binary and eleven in decimal notation.

A Wittegensteinian Theory of Names

Also could be considered Neo-Fregean, as the Fregean "sense" is the "objective" sense shared in a language. The Semantic Web is just a language that allows co-ordinated action. Ludwig Wittgenstein

The amount of co-ordination needed does not need full agreement, but only the minimal agreement needed to get the task at hand completed.

How can we boot-straps URIs for things and re-use these URIs in a way that people will actually use?

Answer: Use a paradigm average Web users are familiar with, such as a search paradigm.

Referent via Serach

referent

An Algorithm

  1. Given a term, retrieve a set of web-pages (Using Live.com).
  2. Given a term, retrieve a set of Semantic Web URIs and all triples (facts) associated with them (Using Falcon-S).
  3. Human searches through web-sites.
  4. For each web-site the human clicks on
    1. Strip out HTML
    2. Extract "shallow semantics" via NLP Pipeline: verb(subject, object) via dependency parser with morphological analysis.
    3. match "shallow semantics" output to each of the facts using string-matching and WordNet for synonyms.
  5. pick the URI with the most "votes" from the shallow semantics.

The Wisdom of Crowds

Each "fact" then has a vote.

So we can use these "votes" to determine what the core set of socially-agreed "facts" are part of our understand.

penguisn

We can then build a statistical model of the most-commonly agreed upon terms.

Not sure how to code this in RDF.

Yahoo! Search Monkey

Companies are trying to deploy this kind of technology...

Yahoo! SemSearch

Term Stabilization

Also view the distribution of what "facts."

Further Thesis: People will disagree of the "long tail" but agree on "top of distribution".

power-law tags

From preliminary results, it appears feedback does not have a significant effect contra work by Catutto et. al. in Semiotic Dynamics.

Data Set: Microsoft Live Search Query Logs

Search queries: 16 million queries

Unique Queries: 7 million

Unique People: 50,000,000

Unique Places: at least 5,000

Unique Concepts: at least 10,000

User Study

Very difficult to evaluate and haven't run yet.

Just running users on queries retrieved from corpus and having them

Asking users if the facts retrieved from the Semantic Web are "about the same thing as the search" on a 1-4 ordinal scale (Poor, Fair, Good, Excellent) or binary?

Some of the facts retrieved by the Semantic Web are confusing - like The Eiffel Tower is a noun.

Any advise?

Null Hypothesis?

What should I compare my system against?

Do I create a really stupid (i.e. randomized) baseline algorithm?

Just use the results of Pagerank over SemWeb Linked Data?

Maybe testDifferent Machine-Learning Algorithms:

Conclusions

Is this a unified picture?

A philosophical thesis: that naming on the Semantic Web relies on minimum co-ordination required for disambiguation.

Wrapping Up

Any questions?