eCulture and Europeana Semantic Search

MultimediaN E-Culture NL project: CWI Amsterdam; Vrije Universiteit, Amsterdam; Universiteit van Amsterdam; Digital Heritage Netherlands DEN; Dutch Institute for Cultural Heritage ICN

• won the Semantic Web Challenge at ISWC 2006. Oldie but goldie!
• featured in Europeana ThoughtLab Semantic Search category
• I think this is a seminal paper: Ossenbruggen, J. et al., Searching and Annotating Virtual Heritage Collections with Semantic-Web Techniques. In J. Trant and D. Bearman (eds). Museums and the Web 2007: Proceedings. Toronto: Archives & Museum Informatics. Published March 31, 2007
  • in particular sec 6 "Keyword Search With Semantic Clustering" presents an interesting approach based on graph navigation. But their DB is fairly small...
  • the source is available as the "slashfacet" (/facet) browser, http://slashfacet.semanticweb.org/. However, it's in SWI Prolog we won't use it directly
• current demo (implemented in SWI Prolog)
  • nice auto-completion of concepts, similar to what we have in LinkedLifeData
  • eg type "Rembrandt", expand Persons. There are 90, since some of the Rembrandts are not aligned (identified as sameAs)
  • select the "Rembrandt van Rijn", which is an entry from Thesauri Tools#Getty Thesauri ULAN resulting in this search:

    keyword:http://e-culture.multimedian.nl/ns/getty/ulan#500011051

Shots from the paper and demo follow

Lightbox with Author and Painting Timeline

Search by style and display lightbox with author life-lines, with paintings represented as dots:

Hierarchical faceted search

Search in thesaurus and then disambiguate from context

Painting detailed view

Shows metadata thesaurus terms in a very readable way, yet identifying the source

Clustered results

Works are grouped by relation to the search term ("created by/showing/related to" Rembrandt), the groups are collapsible. Nice animated scroll (left/right). Few pages are represented by dots, many pages by "Go to page N" control

Chart results

Chart results by various criteria, in this case by Iconography:

Object Views

Object popup view: Links to full view, original page, annotate (after login with OpenID), search by any concept appearing in the object

Full object view. Nice display options.

Exhibit

Exhibit is a Publishing Framework for Large Scale Data-Rich Interactive Web Pages.
Created as part of MIT Simile, its development is now taken over by Zepheira on contract with the Library of Congress and others.

• "Any data that can be modeled as an RDF graph and encoded in simple JSON (i.e., anything)". Which may nicely answer the question "What should be the communication between our backend and frontend layers?"
• Display of data collections: faceted, timeline, geographical.
• See these excellent demos
  

Versions:

• Exhibit 2 is the original
  • The current in-browser option is called Scripted and has limit of 1k items
    I think this is the reason for Europeana's limitation: "Currently the timeline only shows up to 1000 items relevant to your query"
  • Still, it may be adopted in RS because of concerns with Exhibit 3 maturity/stability, and speed of transferring and storing the results
• Exhibit 3 is still in development.
  • Blog post Part1
    Main benefits:
  • Staged server-side option for publishing large collections (100k items, 20k facet values)
    Involves an Exhibit server that stores the items
    • But how fast can this update? Eg to transfer 100k results from OWLIM to Exhibit, times all concurrent users... will be slow and resource-consuming
  • permalink (bookmarking) to capture state (search terms, facets selected, timeline/geomap view). Important for sharing insights with others. Similar idea to 20 Data Basket
  • options to pass data to Google Refine and other processing
  • provides HTTP interface

Recollection

• Recollection is a free platform for generating and customizing views, (interactive maps, timelines, facets, tag clouds) that allow users to experience your digital collections. Generate distinct interactive visual interfaces to your digital collections, including maps and timelines, and sophisticated faceted navigation. Just copy-paste to embed your interface in any webpage."
• Recollection is created by (part of) the same people who do Exhibit
• Builds on Exhibit and several other OS frameworks: http://recollection.zepheira.com/about/community/
• Includes nice and simple config/layout tools
• MarinD writes: this tool will be open sourced soon, it looks very handy for creating simple browsing/visualisation over Linked Data (even though it seem they don't have RDF adapters yet, only DB/Excel)
• For each collection there are 4 views - Map, List, Timeline and Charts

Present Visual Collections

• Map View
  

• List View
  

• Timeline View
  

• Charts View
  

Map Collections

Browse collection timelines

Chart relationships in collection data

Pachyderm

A tool for creating galleries or exhibitions
Video intro, registry" of sample exhibitions, User manual
V.Fol: "the video and presentations go very slow" (maybe she had bad internet connection?)