Skip to main content

Kynetx Impact Conference - notes - afternoon day 1

Posted by haroldcarr on November 19, 2009 at 6:40 PM PST

1:00-1:45 PM
Client-Side Endpoints and Context Automation
by Dave McNamee, Kynetx VP Client Engagement/VAR Services

Endpoint: thing on client that ties application to user.

- endpoints make KNS apps work on client
- many options
- endpoint selection

controller (e.g., card selector)
endpoint (browser extension)
client app <-> KNS

Endpoint Domains: Outlook, Thunderbird, IE, Firefox

Browser Endpoints:
- site tagging
- bookmarklets
- proxy server
- single-purpose browser extension
- OEM browser extension
- Info Cards

Site Tagging
- deploy method: manually tag site
- distribution channel: web server
- opt-in- process: user visits tagged site
- user action to fire rules: as defined in rules
- good choice for: website administrator

- deploy: download from AppBuilder
- distrib: App Directory or download sites
- opt-in: install bookmarklet
- fire: click on bookmarklet
- use-case: apps invoked by user

Proxy server
- install proxy server with Kynetx
- proprietary network
- user accesses network
- as defined in rules
- closed network environments

- select extension in AppBUilder
- App Directory or download sites
- install browser extension
- as defined in rules
- communities, affinity groups, loyalty programs

OEM Browser Extension
- integrate KNS into extension
- App Directory or download sites
- user installs extension
- as defined in rules
- companies with existing browser extensions

Info Cards (endpoint aggregator - other apps can use it)
- info card in AppBuilder
- App Directory, other download sites, card distribution systems
- user downloads selector and info card
- card turned on and defined in rules
- aggregating endpoint, personal info, viral distribution

2:00-2:45 PM
Using Information Cards
Paul Trevithick, Azigo CTO

Status quo:
- site to site: google, boa, bestbuy, borders, equifax
- you type in same data into each
- each site has cookies - a profile of you for that site
- gmail is selling you using its cookies

- easy for user
- crude context automation

- when people know how it works, they don't like
- no way to build deeper relationships
- improved relevance hard to do

Internet is missing identity layer

Open Source Higgins Project
- 2003

Card Metaphor
- digital version of cards you carry in your purse/wallet
- pioneered by Kim C and MS

Info Cards and Claims
- data portability (profile and social networking attributes)
- any kind of info: preferences, favorite songs, id numbers, affiliations, ...

- cards from multiple sites are managed in a local selector application
- hardened piece of infrastructure separate from browser but able to
  be integrated with it

Info Card Foundation
- 2008
- Oracle, Google, PayPal, Novell, Intel, MS, Equifax

Managed card: what somebody else says about you
Personal card: what you say about you

Federal Government User Case: First Pilot: NIH
(HC: Note: Metro is in this space)


Relationship cards: introduce a 4th party (besides user, IP and RP)
- points to data handled by 4th party

Kantara UMA Authorization Manager

Action Cards:
- Kick starting InfoCards
- Client-side mashup (e.g., Kynetx)
- Unmodified RPs
- Now you are in control of context
- Every website is a fresh canvas
- load whatever cards (apps) you like
- limitless cross-site context and personalization

Capturing your own data (instead of forms to each site)
- capture web form data and map to a common data model
- capture what you type
- is it OK to spy on yourself?
- depends on which self

Making data available to Kyntex (by click a card)

Azigo 3.0 roadmap
- universal login
  u/p; IMI card, OpenID
- Card Store
  2011 goal: install base of 1 million
  card issuers can offer card-apps in the store
- password mngr
- form filter
- roaming
  XDI personal data store
  All data attributes encrypted on client

See: mozilla weave (syncs bookmarks)

3:00-3:45 PM
Using Data to Enhance Contextual Experiences
Phil Windley, Kynetx CTO

Linking datasets to Kynetx Apps. Include ETL/DW and how Kynetx handles large volumes of data. Also information on data feeds and reporting.

data issues:
- caching
- format

KRL: why a new language:
- linguistic abstractions are powerful
- nothing was out there that was sufficient
- getting hugh leverage
- surprising what people can do with language not intended by lang designer

- fixed static set of data (e.g., where does AAA offer discounts)
- dataset foo <- "";
- cache data from that URL for 20 minutes
- dataset foo <- "" cachable;
- 24 hours and pushed to browser with cached data
- dataset foo <- "" cachable for 3 hours;
- just like it says

- in global: datasource foo ,_ "";
- in prelude: bar = datasouce:foo({"q":isbn});
- every query might be different - cached according to complete URL of query

intrinsics (built into KRL - ambient - as opposed to active - supplied by card)
- url; location; browser
- c = location:country();
- temp = weather:curr_temp();
- domain = page:url("domain");
- if weekend() then { notify("Time for play", temp) }

How to pick data (pick is xpath for data)
- books = datasource:lib({"q":isbn});
- url = books.pick("$[0].url");
- title = books.pick("$[0].title");
- if (books.pick("$..numFOund") > 0) then { notify() }

- language evaluation
- init; evaluate; callbacks
- each generates a log
- analyzer has to correlate logs
- get graph of ruleset evaluations and rules fired

4:00-4:45 PM
Purpose-Centric App Showcase
Chris Wells, VP Business Development

Learn about a variety of example Apps that are leveraging the idea of
the Purpose-based web using KRL, Information Cards, end points and/or
some combination of the above.