The Arc in the Center

Web 3.0 - The Short Version

gilad.goren@gmail.com — Wed, 29 Aug 2007 10:04:00 GMT

Web 1.0: We read
Web 2.0: We read and write
Web 3.0: We read and write and link

Gil'ad

What is Web 3.0?

gilad.goren@gmail.com — Wed, 22 Aug 2007 19:22:00 GMT

As a newcomer to this area I tried to define to myself what stands behind this buzzword. My answer is simple: Web 3.0 is about new techniques to organize, retrieve, share and combine information using Graphbases.

We can define four generations of information-organization methods:

Generation	Method	Data Structure
Web 0.0	Directories	Hierarchical
Web 1.0	Indexing	Flat
Web 2.0	Tagging (folksonomy)	Flat
Web 3.0	Linking	Graph

(I guess that every time the volume of the web, i.e. the amount of documents, grows by four or five orders of magnitude, a new technology emerges.)

Data retrieval in general is about creating metadata 'off-line', and using it 'on-line' to quickly find the information items we need. The essence of the Web 3.0 is to use metadata in the form of typed links, i.e. links that express a specific type relationship between the linked elements, to create annotated graphs, or networks, of linked information items.

We can define three basic classes of items we should link together: Objects, Concepts, and Attributes:

Objects could be real-world objects ( ), desktop objects ( ) or web objects ( ) .
Concepts ( ) are representations of abstract topics or categories (e.g. 'service providers'), instances (e.g. Verizon), or concrete objects (e.g. Verizon' CEO) [don't confuse the concept and the real-world object it represents. See the semiotic triangle below].
Attributes Includes Tags ( ) like "V-e-r-i-z-o-n" or "I-v-a-n-S-e-i-d-e-n-b-e-r-g", as well as Geotags, URLs ( ), etc.

There are various ways to link together objects, concepts and attributes. For example:

When we link attributes to web objects ( ) we call it tagging, or indexing if it is done automatically.
Ontologies, Taxonomies and/or Topics Maps, link together concepts:

Semiotic triangle looks like

Graphbase is an information-base based on graphs in a fashion that can work at Internet scale. When URIs ( ) are used to annotate information items, items can be uniquely identified, so Graphbases allow data to be processed outside the particular environment in which it was created.

Combining together concepts, tags, objects and URIs we can create simple Graphbase like:

In the Web 3.0 environment there will be multiple different applications and ways to use Graphbases for exposing, sharing, and connecting pieces of data, information, and knowledge. For example:

Intra-document information analysis based on domain ontologies to identify the latent knowledge found in text. In simple words, it enables better search engines.
Inter-document data-web: linking together heterogeneous objects enabling better browsing.
Extract structured meaning (i.e. concept-attribute pairs) from documents, and linking them to the document to present knowledge in a standard form, permitting software agents to better find and combine information.

...

So, to summarize this simplistic view, Web 3.0 is about new ways to organize and retrieve information.

The next question one would like to answer is "when does it worth the effort?" In other words, "why and when indexing and tagging are not good enough?" I am not sure I have good answers yet, so I will be happy to have some feedback on this.

Cheers,
Gil'ad

For Further reading

Metadata? Thesauri? Taxonomies? Topic Maps! [2004]

Web Quads [Folksonomy vs. Ontology]

gilad.goren@gmail.com — Fri, 17 Aug 2007 15:13:00 GMT

Two years ago, Folksonomy vs. Ontology became a trendy debate between the social-web buddies who have faith in collective human intelligence, and the semantic-web advocates who look for a strict, machine-friendly, knowledge organization. Typically, tagging is done by the persons who consume information, while ontologies are provided by the information providers.

Tagging is a key phenomenon of the social web since 2002 or so. Folksonomic tagging is intended to make information easy to search and discover, and since folksonomies are user-generated, advocates of folksonomy believe that it provides low-cost alternative to traditional, top down, centrally controlled, taxonomic categorization.

It seems to many, that folksonomies are like taxonomies in that they share the same purpose: classification. So, a lot of discussions went on since early 2005 about "which is better".

In a famous blog post, Clay Shirky (2005) made the argument that "ontology is overrated" and tags are "a radical break with previous categorization strategies...much more organic ways of organizing information than our current categorization schemes allow." Shirky illustrated how engineered taxonomic categorization schemes are limited, and how free-form, massively distributed tagging is resilient against these limitations, making ontologies quite redundant.

Several blogers criticized Shirky's approach. Nova Spivak wrote "Tagging systems are just special, highly simplistic cases of ontologies". Thomas Gruber, in his article "Ontology of Folksonomy: A Mash-up of Apples and Oranges" expressed very different view: "folksonomies and ontologies are like apples and oranges".

Recently, "taggers" became more relaxed with ontologies. Thomas Vander Wal, the "Father of Folksonomy", said last year: "The beauty of tagging is that it taps into an existing cognitive process without adding much cognitive cost" (or did he quote Rashmi Sinha?), but he also "stressed the importance of folksonomies and taxonomies working together, the folksonomy recognizing gaps in the taxonomy".

The Semiotic Triangle [Semiotics: The study of signs]

In semantics and linguistics, the semiotic triangle is a figure that is used to explain the relationship between Concepts, Symbols (or signifier or signifiant) and Objects (or signified or signifié). It is also known as the semantic triangle [Wikipedia].

The Semantic Triangle differentiates between three separate dimensions:

The conceptual domain - thoughts that are in our minds
The symbolic domain - worlds and symbols that we use to communicate with others
The real world - things in the real world that we refer to in our thoughts and with symbols

The semiotic triangle is used to discuss the differences between objects, concepts and symbols. A key point is that a symbolic representation of an object can never refer directly to objects, but only through concepts within the mind. Therefore the link at the bottom of the triangle (which is a dashed line) is only an implied relationship.

We can extend this triangle by adding another domain: percepts which are mental images of real-world objects.

The Web Quad

The equivalent of the Semiotic Triangle in the web-space is the Web Quad which includes four domains:

The conceptual domain (ontologies) - abstract concepts and instances identified by URIs
The symbolic domain (folksonomies) - tags that we use to communicate with others.
The objects world - things in the real world that we refer to in semantic web documents.
The percepts domain – information elements on the web (e.g. multi-media documents) that presents things in the real world.

How the symbolic domain differs from the conceptual domain? The symbolic domain either lacks any internal structure, or it only incorporates associations between a word and its syntactic derivatives. But the major difference between the symbolic and conceptual domains in their relations to real-world objects. Like the human case above, tags can never directly refer to real-world objects, but only through concepts or geo-tagged information elements. Therefore the link between tags and real world objects (which is a dashed line) is an implied relationship.

So now the situation is somewhat clearer. Both folksonomies and ontologies are part of the semantic web. Both can be presented as triplets. But tags are not "highly simplistic cases of ontologies". Tags belong to different domain. Geotags seems to be yet another type of relationship directly connecting information elements with real-world objects.

In other words, tagging is not directly about categorizing. In the social web tagging provides a means to associate information elements with common tags, which in turn are associated with categories in the minds of the users.

The Social Web only involves the symbolic domain (tags) and the Information Elements domain. The conceptual domain is in the minds of the people who interpret the symbols (tags) and the Information Elements. The Semantic Web adds a third dimension, the conceptual domain (ontologies), into the web.

The only way humans may interact is through signs (i.e. words). Human-beings can't directly share concepts (well, most of us can't ). The situation on the web could be different. Computers can directly exchange ontologies which present concepts. But we should leave this discussion to another post.

This is a first version of this document, trying to put some order into semantics and tagging. I guess I will revise it as I learn more. Feedback is very welcome.

Cheers,
Gil'ad

For Further reading

Thomas Vander Wal, bottom-up-tagging [2007]
Csaba's Blog: Tagging: Words versus Magic [2007]
Webology [2007]
ONTOLOGIES & FOLKSONOMIES [2007]
Tom Gruber, Ontology of Folksonomy [2006] <<added>>
Rashmi Sinha, A cognitive analysis of tagging [2005], and A social analysis of tagging [2006]
John F. Sowa, Ontology, Metadata, and Semiotics [2000]

Why blog?

gilad.goren@gmail.com — Tue, 07 Aug 2007 22:26:00 GMT

The Arc in the Center. This blog is created by Gilʻad Goren - an entrepreneur by soul, physicist by education and software engineer by practice.

In the last twelve years or so I was involved in the Telecom industry. Metro Ethernet and T-MPLS were in the center of my world. Now, I am shifting my attention from packet networks to Web X.0 (X>1). I am currently interested in various aspects of the web, from complex networks to the semantic web.

To begin with, I will be happy to use this platform to exchange ideas with those who are interested in RDF and the Structured Web as a step towards the Semantic Web and as a tool to form knowledge from information.

Cheers,
Gilʻad

The Arc in the Center launches

gilad.goren@gmail.com — Mon, 06 Aug 2007 23:55:00 GMT

The Arc in the Center, powered by Terapad.com (http://www.terapad.com/) was launched today featuring blog, forums, image gallery, online shop, event calendar and more.

The Arc in the Center can be accessed at http://giladgoren.terapad.com/.