Zemanta - reblog

I bumped into this webcast from others referencing it as a ‘must watch’ from NextWeb 2008. So, while I had sworn off pondering semantics for awhile (too early for most clients) I decided I should check it out. I wasn’t disappointed. Nova Spivack (CEO of Twine) gave the best 101 on Semantic I have seen:

• Five approaches to add/infer semantic data
• How semantic enables JIT database models
• How semantic enables part of the “Web as an OS” dream
• Where we are now, where we are going
• Q&A
• and moreAs usual, the keys I saw are below plus my usual raw scribble I took while running/watching this.

• Nova Spivack
  

  • AI focus (I knew I liked him for a reason), Co-Founded Earthweb which created DICE.com

• Put aside that he and I both agree this is early adopter material … you need to watch this to begin to get your head around it. If what he is saying shakes out … the tooling to enable semantic frameworks around data is 2-3 years away from prime time. That is just a blink (especially if you have kids).

1. Queue the “Circle Of Life” Song … Web 3.0 Returns Us To The Backend via Semantic Tools

• He made some excellent points on the cyclical nature of development focus and how Web 2.0 is all about the UI. The next wave, however, will refocus us on the backend and the catalyst for that will be the value semantic structures can bring to businesses (he has some great slides on that). Primarily he pointed to how documents and data exploding on the web making algorithmic search models like Google begin to be inefficient.
  

• Perhaps a joke only a AI major would appreciate. His point was that AI can’t solve complex riddles like deriving semantic data for existing stuff. It can help but it is more of tool for the human mind to apply … than a machine you can delegate this task to.

• Another good word play. Here he was talking about how enriching data with semantics can enable the applications that consume it to become dumber about the specifics of data (relationships between it, what it means). Instead all of the application focus can be on how to dynamic leverage data that is presented to it based on its associated semantics. I think his venture Twine is something along that front.

I’m am going to ask to get an invitation to the beta of Twine. From that I hope to get a better sense of what he is seeing here.

*************************************************
*************************************************
****** START OF RAW SCRIBBLE TAKEN WHILE RUNNING ******
*************************************************
*************************************************

• 6/6/2008, 6:11 AM
• Radar Networks - CEO - Nova Spivack
,,,,,,,,,,,,,,○ Product called Twine
• Excellent article hitting topics -
,,,,,,,,,,,,,,○ http://www.techcrunch.com/2008/04/25/is-keyword-search-about-to-hit-its-breaking-point/
,,,,,,,,,,,,,,○ http://thenextweb.org/2008/04/03/nova-spivack-the-semantic-web-as-an-open-and-less-evil-web/
• 6/6/2008, 6:12 AM
• Overview
,,,,,,,,,,,,,,○ Stuff = Nouns
,,,,,,,,,,,,,,○ Connections = Verbs
,,,,,,,,,,,,,,○ Social Graph is subset of Semantic Graph
,,,,,,,,,,,,,,○ Improves collab, integration of data, personalization, etc
• 6/6/2008, 6:15 AM
• Web 3.0
,,,,,,,,,,,,,,○ Coming 3rd Decade of Web (till in 2.0)
,,,,,,,,,,,,,,○ Focus enrich structure of web
,,,,,,,,,,,,,,○ File Server to Database
,,,,,,,,,,,,,,○ Bauer Comment - nice slide showing evolution by web ‘waves’ (1, 2, 3, etc)
,,,,,,,,,,,,,,○ Pendulum from front end focus to a back end focus
,,,,,,,,,,,,,,○ 3.0 backend … 4.0 back to front end based on improved backend
• 6/6/2008, 6:18 AM
• Failure of search as data on web increases
,,,,,,,,,,,,,,○ Bauer Comment - another referenced graph that is solid (on his blog)
,,,,,,,,,,,,,,○ Solutions
,,,,,,,,,,,,,,,,,,,,,,,,,,,,1. Tagging
,,,,,,,,,,,,,,,,,,,,,,,,,,,,2. Natural language search
,,,,,,,,,,,,,,,,,,,,,,,,,,,,3. Semantic (goes into understanding meaning of natural search)
,,,,,,,,,,,,,,a. Keyword search is capped out
• 6/6/2008, 6:19 AM
• Five approaches to add / infer from data
,,,,,,,,,,,,,,,,,,,,,,,,,,,,1. Tagging
,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,1) Bad due to inconsistency between public taggers
,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,a) Bauer Comment - he talk about how large data set can analyze the tags and reach a common definition of tags from the crowd (UX card sorting in mass)
,,,,,,,,,,,,,,,,,,,,,,,,,,,,2. Statistical Approach
,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,1) Google
,,,,,,,,,,,,,,,,,,,,,,,,,,,,3. Linguistic Approach
,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,1) Understand meaning of text … rules, grammar, etc … read the text systemically
,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,2) Great but high CPU, hard to scale, language dependancies
,,,,,,,,,,,,,,,,,,,,,,,,,,,,4. Semantic web
,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,1) Set of standards (open) make the semantic web (W3)
,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,2) Semantic technologies
,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,a) Express, infer meaning of things
,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,3) Meta Data … put into the data that describes meaning of data
,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,4) Does w/ open standards
,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,5) Problems — Hard to scale, who makes all this meta-data on existing content
,,,,,,,,,,,,,,,,,,,,,,,,,,,,5. Artificial Intelligence
,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,1) Decades away … Cycorp in US … 15 years manually keying human knowledge
,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,2) Wikipedia - AI on top might do better keeping up w/ human consensus
,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,3) Huge in 4.0 …. But far away
• 6/6/2008, 6:24 AM
• How to compare add/infer data
,,,,,,,,,,,,,,○ Software smarter
,,,,,,,,,,,,,,○ Data smarter
,,,,,,,,,,,,,,○ AI is both
,,,,,,,,,,,,,,○ Semantic web is in the middle
• 6/6/2008, 6:26 AM
• Two paths
,,,,,,,,,,,,,,○ Manual - RDF, L … too hard
,,,,,,,,,,,,,,○ Top down - Automatically generating … from content .. More practical
• Twine
,,,,,,,,,,,,,,○ Use RDF … top down
• Make a higher resolution web as each piece of data carries more information about it
,,,,,,,,,,,,,,○ Types (person, city)
,,,,,,,,,,,,,,○ Relationship (like ‘lives in’)
,,,,,,,,,,,,,,○ Google is just pages / links
,,,,,,,,,,,,,,○ Bauer Comment - Key becomes normalizing, in effect, static content via verbs and nouns descriptions
• Bauer Comment - Between websites imply global nouns
• Smart data
,,,,,,,,,,,,,,○ Enables dumb software …. To become smart via the enriched data
• Applications evolve that read semantic data and are independent of vertical (health, etc)
• He wants to not use AI … use AS (Artificial Stupidity)
,,,,,,,,,,,,,,○ Bauer Comment - good word play … helps clarify role of machine in complex decision chain
• 6/6/2008, 6:31 AM
• JIT in time data
,,,,,,,,,,,,,,○ Semantic web pulls in just when needed
,,,,,,,,,,,,,,○ Problem in past way communicating schema (i.e. XML, DDL)
,,,,,,,,,,,,,,○ Semantic web … does it via an entology … a page on web … makes sense of data
,,,,,,,,,,,,,,,,,,,,,,,,,,,,§ Bauer Comment - sounds like XML to me
• 6/6/2008, 6:33 AM
• Making the web an OS
,,,,,,,,,,,,,,○ File system aspect could be based on semantic
,,,,,,,,,,,,,,○ Bauer Comment - How does MESH play into this … or similar
• Open - Ontologies, Rules, Data Records, Data Mappings, Query Interface,
,,,,,,,,,,,,,,○ Bauer Comment - Key break down of basic semantic framework
• Standards
,,,,,,,,,,,,,,○ OWL - Built on RDF … more expressive schemas
,,,,,,,,,,,,,,○ SWIRL - Rules Language
,,,,,,,,,,,,,,○ Griddle - Transform data
,,,,,,,,,,,,,,○ Sparkle - Another RDF solution
• 6/6/2008, 6:36 AM
• Basic Unit of Data in Semantic Web
,,,,,,,,,,,,,,○ Triple - Subject , Predicate, Object (susan, works for, IBM)
,,,,,,,,,,,,,,○ URI - Points to location on web with further information on each piece (for the triple) …
,,,,,,,,,,,,,,,,,,,,,,,,,,,,§ Bauer Comment - Linking database definitions or file definitions cross site
,,,,,,,,,,,,,,○ Then link data records - create open database
,,,,,,,,,,,,,,○ Better term “Data Web”
• 6/6/2008, 6:39 AM
• Table isn’t how we think
,,,,,,,,,,,,,,○ List of triples … but lists get really REALLY long … talking billions
,,,,,,,,,,,,,,○ Vomits a DB to stick triples in DB
,,,,,,,,,,,,,,○ Solve this via ‘Triple Store” … new DB designed for this
,,,,,,,,,,,,,,,,,,,,,,,,,,,,§ Easier to maintain
,,,,,,,,,,,,,,,,,,,,,,,,,,,,§ List A + List B = List C
,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,□ Bauer Comment - Hello VSAM
,,,,,,,,,,,,,,○ Form and Dissolve database relationship structure
• 6/6/2008, 6:41 AM
• Linked data universe
,,,,,,,,,,,,,,○ Number of ontologies cover various Ontologies
,,,,,,,,,,,,,,,,,,,,,,,,,,,,§ Reviews, Music, Communities, etc
,,,,,,,,,,,,,,,,,,,,,,,,,,,,§ Bauer Comment - Open XML standards are similar
• 6/6/2008, 6:42 AM
• Right now
,,,,,,,,,,,,,,○ Early adoption period
,,,,,,,,,,,,,,○ A lot of momentum in developers … early applications like Twine
• Focus today is how to become more open
,,,,,,,,,,,,,,○ Data portability project
,,,,,,,,,,,,,,,,,,,,,,,,,,,,§ Semantic web enables portability
• Very hard … no tools
,,,,,,,,,,,,,,○ Two examples
,,,,,,,,,,,,,,,,,,,,,,,,,,,,§ FOOF - Friend of Friend
,,,,,,,,,,,,,,,,,,,,,,,,,,,,§ Shock - Sharing forums
,,,,,,,,,,,,,,○ API’s get more open in next year
• Twine
,,,,,,,,,,,,,,○ TO get into beta — nova@radarnetworks.com nextweb2008 twine
• Q&A
,,,,,,,,,,,,,,○ 6/6/2008, 6:47 AM
,,,,,,,,,,,,,,○ Economic consequences of this? New business models? Theme.
,,,,,,,,,,,,,,,,,,,,,,,,,,,,§ Semantic web doesn’t create new business models. Just makes existing ones better.
,,,,,,,,,,,,,,,,,,,,,,,,,,,,§ Bauer Comment - Better can be disruption for existing players that don’t change
,,,,,,,,,,,,,,,,,,,,,,,,,,,,§ Example of new player - Singechay (sp?) … semantic commerce, ads, etc
,,,,,,,,,,,,,,○ Open standards … enable?
,,,,,,,,,,,,,,,,,,,,,,,,,,,,§ Semantic web … can you use it to lock in users?
,,,,,,,,,,,,,,,,,,,,,,,,,,,,§ Technically if you use the open standards … it is easier (in his view)
,,,,,,,,,,,,,,,,,,,,,,,,,,,,§ Focus more on value creation … not lock in
,,,,,,,,,,,,,,,,,,,,,,,,,,,,§ Trick to lock in … make your own Ontology … stuck in your system
,,,,,,,,,,,,,,○ Reason web succeeded was simple … Semantic web isn’t
,,,,,,,,,,,,,,,,,,,,,,,,,,,,§ Key question is how long did WWW take to get defined … 15 years
,,,,,,,,,,,,,,,,,,,,,,,,,,,,§ We typically think of explosion point in early 90’s
,,,,,,,,,,,,,,,,,,,,,,,,,,,,§ Get to point of ‘easier’ to use
,,,,,,,,,,,,,,,,,,,,,,,,,,,,§ Gap is tools for Semantic web … estimate 1st wave 12-18 months … twine is open up hosted development tools
,,,,,,,,,,,,,,○ Semantic requires definitions of objects and relationships… how get agreement
,,,,,,,,,,,,,,,,,,,,,,,,,,,,§ Bauer Comment - Whispering in background about ‘free beers at door’ … hehe
,,,,,,,,,,,,,,,,,,,,,,,,,,,,§ Semantic web designed for disagreement
,,,,,,,,,,,,,,,,,,,,,,,,,,,,§ Good but many definitions
,,,,,,,,,,,,,,,,,,,,,,,,,,,,§ Bauer Comment - Similar to MLS in real estate market
,,,,,,,,,,,,,,,,,,,,,,,,,,,,§ Built into standard you can map across Onotologies
,,,,,,,,,,,,,,,,,,,,,,,,,,,,§ Anyone can make associations … community driven …
,,,,,,,,,,,,,,,,,,,,,,,,,,,,§ Bauer Comment - so the walls come down via crowd
,,,,,,,,,,,,,,○ Amazed on how quickly google tagged images online
,,,,,,,,,,,,,,,,,,,,,,,,,,,,§ Twine is doing same .
,,,,,,,,,,,,,,,,,,,,,,,,,,,,§ Social network for sharing knowledge … discover around interest
,,,,,,,,,,,,,,,,,,,,,,,,,,,,§ Add information auto tags
,,,,,,,,,,,,,,,,,,,,,,,,,,,,§ Learn and make recommendations
,,,,,,,,,,,,,,,,,,,,,,,,,,,,§ Open more in summer / fall
,,,,,,,,,,,,,,,,,,,,,,,,,,,,§ Tries to automate RDF setup
** END RAW SCRIBBLE TAKEN WHILE RUNNING **

Tim Bauer’s Running Thoughts, Jun 2008

Publish to your blog ...

Your blog address:

Like yourblog.blogspot.com, yourblog.typepad.com, yourblog.wordpress.com

Log-in details:

• Username:
• Password:

We will not store your username or password—ever.

« Go back to edit Publish »

... or copy the code into your blog.

Post code

Tags