TW group meeting notes February 16 2010

From Semantic Portal Wiki

Jump to: navigation, search

Tim not allowed to touch conference phone.

Peter: announcements?

  • JimH: evan and shangguan: wireless is not working.
  • Evan: only 2 were ordered. we need 4.

JimH: Shuttle comes :15 after and :45 every hour. Used to have continuous shuttle, but we know when it comes now. Have not told us when we should be up hill. Use it, even if going up to lunch. If you pick one up at the horseshoe, tell them you're going to Winslow. Can get off at West Hall or Rickett (just past Troy building). Go in groups if possible. If you want more than 5/10 minutes, let professors know.

Patrick: Phone system bad, we need to speak up. Better phone system? We have digital-analog bridge from a digital line. The phone is analog. We need a digital conference phone. Upstairs phone has same issues. Follow tweets for new motto (TW never sleeps).

ACTION: look at phone issues. Patrick.

Peter: Heating issues, ventilation? Patrick: Computer room into 70s. Should be mid 60s. Peter: Global change, not global warming. Failure in heating system in building. Medha was tracking issues in heating within building. Tell Jacky of any issues. 68 is target temperature.

Patrick: instead of dimdim, webX? Peter: later.

JimH: Office spaces. Desks unclaimed. If you don't have one and want one, email JimH. First year does not have assigned desk.

Peter: This Friday is poster day in CS Dept. Student poster session. CSD holds a mixer in MRC/Lally102 (loc?). Few professors show posters and present. If TWC paid for you to go, you should be volunteering. Should we host? Facilities may work. Will try to have it here.

  • Jesse (but might be gone)
  • Jin
  • Jiao
  • Evan
  • Data-gov Dominic
  • Greg
  • Eric (eScience)
  • Shangguan/James (Provenance)

When doing things, think about reporting it on a poster.

JimH: Putting posters up on the walls, hallways. Everyone should put theirs up. Not on projector wall or "that wall".

Evan presenting "Reengineering the Tetherless World Website"

A Harrowing Tale of SPARQL, XSLT, and Content Management

combine them for a semantically-enabled system.

contrasting to web1.0

  • static links

dimdim'ers can get the file from Documents.

JimH: Our talks should be going onto the web site.

web2.0 uses dynamic content and social apps

  • scripts talk to DBs

web3.0

  • dynamic content, many data sources, machine understandable

more interactions among the data sources incorporated into a single web page production.

want to integrate these.

what if people have foaf files and use them to create a page.

Generating pages from RDF

  • perform sparql query
  • xslt the results (bindings2html)
  • apply css
  • ajax updates with limit/offset

flexibility:

  • change order, omit results
    • done with modularized xslt that is included
  • tw:hidden xsd:boolean

suggesting two styling predicates (a boolean or a URI)

flow diagram starting with agent's request

  • (look at headers, do they want RDF?)
  • creating php scripts
  • HTTP-Referer
  • creating on-the-fly stylesheet

How is stylesheet created? provide query string with param ?r to name the page that you are rendering. HTTP-Refer. perform sparql query to find out types of the object that page is describing.

JimH: useful for mobile agents as well, keep in mind.

User-agent string can be used as well.

"embedded content" abstracts data storage away from presentation.

sites are live.

http://wineagent.tw.rpi.edu/tw.rpi.edu/pages/Themes

query against endpoint, transformed with xslt. get list of TW topics.

showing html used. start with div for target. escaped javascript to make query with sparql.

e.g. lead professor is a URI value, need to grab string from that URI.

Deborah's page http://wineagent.tw.rpi.edu/tw.rpi.edu/pages/Themes

  • has rdfa

different sytlesheets for bio, affiiations, papers, name/image.

can change order of content using javascript (does not persist).

people can customize how content appears on page.

how to use same content in mediawiki? uses an extention named sparql, giving params will get content.

"embeddable stylesheet" strips header (?)

get same tools to move up and down content in mediawiki

questions?

JimH: make sure stylesheets. Kept static page and RDF-backed page. He wants static page to look how it looks so he can edit easily (with standing html). Should try to figure out how to make as cut-and-paste as possible ("do this to get 10 publications shown"). Need something flexible. Don't want to store everything in RDF to get onto page. "all photos I'm tagged in". Needs to integrate.

JimH: The group will be using this type of thing for TW's web page. Real info is on a static home page. Primary professional pages should be on TW site. Project pages, people pages will be on TW site. All publications, all students worked on a project. Should be aasy to add papers, easy to add events, show successes.

JimH: need to provide people their cut and paste. Can't scare them away with XSLT and SPARQL.

Alvaro: e.g. foaf maker

Jesse: set any endpoint? yes. any xslt? yes. results get transformed. If running same query, will xsl work? yes.

JimH: How to include a standard static header?

JimH: using rdfa? yes. xsl outputs rdfa from the sparql bindings.

TimL: what are the requirements of one of these xsl? process sparql bindings and output rdfa? Evan: take RDF/XML as results of describe or construct (not select). produce rdfa.

TimL: handle multiple forms of RDF/XML. Evan: uses pipe operator to get around

Li: ... Evan: redirects to (php)

Greg: based on content of http refer seems concerning. content will change based on referrer field is not in line with good web practice. hyperlinks in document will change on what page you saw last. If copy/paste, it will bring you back to the original (bad). Working with 3 different systems (drupal, smw, etc.). Needs to be independent of any system. Peter: using context referral. Greg: xsl done on server side vs doing with in browser (should move to server side, since accessibility will be issues on user browsers).

Li: how to decompose the solution, e.g. drupal, where people can plug in features so we can collaboratively create system. Evan: generating xslt is tough b/c browsers do it differently. Firefox most lenient in applying xslt. IE working pretty well. WebKit does not perform xsl includes (must update DOM by retrieving it for it). Li: visualization templates to make RDF data (own template language?) about a month ago http://less.aksw.org/browse. If you let people add xsls, you can make a library of people contributing them.

Peter: how to modularize the xsl? Evan: break down by class, author-publication.xsl takes papers/publications and renders them. if person an author, then show it. person-affiliations.xsl.

Li: Linking. using xsl to make link. follow linked data? if see a link to the publication, go to original source via html browser or rdf browser. if requesting rdf/xml, then just return. if requesting html, redirect to actual web site with actual page.

Li: if parse the rdfa, same as rdf/xml? Evan: most of it. some things are not rendered.

provide back rdf/xml with stylesheet, will produce rdfa

Li: Evan: XPATH looks for any node has attribute rdf:about Li: plan for normalization of RDF/XML Evan: Assumes that all descriptions are under the same element (will break of rdf/xml describes them in multiple places within file)

can continue discussion at web project meeting mondays 2:30

Li: http://data-gov.tw.rpi.edu/ws/xslt.html applies an xsl

query for what vis tempaltes can be used to render, e.g., foaf:Person.

Retrieved from "http://tw.rpi.edu/portal/TW_group_meeting_notes_February_16_2010"
Semantic Web Community
Tetherless World constellation
maintenance