Trip Report WWW2004 P3: will the semantic web scale ?

panelists:

• prof. dr. jürgen angele, uni karlsruhe (ja)
• prof. dr. alon y. halevy, transformic (ah)
• prof. dr. ian horrocks, uni manchaster and OWL contributer (ih)
• dr. cathy marshall, microsoft (cm)

there were approximately 150-200 attendees including "tbl".

statements from the panelists:

cathy marshall:

• semantic web is meta data.
• libraries use meta data.
• it is created by trained people.
• on the web, semantic data is created by anybody - no rules applies.

alon y. halevy:

• structuring data is hard.
• people find it hard to share data.
• people need a clear return on their investment.
• • web: unstructure, easy, simple, meant to share
  • semantic data (databases): structured, difficult (requires education), not meant to be shared (privacy)

jürgen angele:

• OWL does not scale, neither conceptually nor technically.
• the semantic web is a great opportunity, but OWL is a step in the wrong direction.

Q1: what is the return of investments for annotations ?

• ja: deutsche telecom prooves there is a clear roi, at least within closed communities.
• ah: maintenance is expensive
• cm: who get's the cost, who get's the benefit ?
• ih: AI is successfull where information is expensive
• question from the floor: were annotation handmade ?
• ja: yes
• comment from the floor: depends on if the customer will pay for it.
• comment from the floor: same motivation as with open software: re-use existing information

Q2: will the scoial process behind the semantic web scale up ?

• ah: there will be many different ontologies, we need to find ways to cope with that
• cm: do we want everybody (the government) to see our personal data ?
• ah: using the same schema (= onthology) does not garantuee interoperability. many companies use the same SAP schema very differently.
• comment from the floor: if the technic is there, people (companies) will use it.
• comment from the floor: nobody claims that agreeing to a schema will solve all the worlds problem. it's just the next step.
• comment from the floor: just do it and don't spend so much time/energy thinking about what is the best way to do it.
• comment from the floor: it doesn't have to be centralized or distributed - it has to be both.
• comment from the floor: the web is chaotic - onthologies are complex. just let the semantic web evolve.

Q3: is the representation language rich enough ?

• ih: counter-question: are people willing to use it ?
• ja: language is important for providing semantic data
• ah: appling sem to a webpage is more like writing your first program a give it to somebody else to use it.
• comment from the floor: you don't need bright craftmanship to add sem to a webpage. let's start with lightweight sem data and free meta data.
• ih: can we make the semweb good enough to be useful ?

to the trip report main page