Loading...
 
[Show/Hide Left Column]

Sujets Help

View Item

Domain Interaction
Domain - extra
Year 2014
Starting march-april 2015
Status Closed
Subject Getting into the Web of Data through the Time dimension
Thesis advisor PIETRIGA Emmanuel
Co-advisors APPERT Caroline
Laboratory INRIA-Saclay ILDA
Collaborations
Abstract In this internship, we want to investigate a novel interaction paradigm to make users able to gather knowledge from the Web of Data. As opposed to the usual web that links documents, the Web of Data links resources that are objects or concepts with multiple typed properties. The simple paradigm that consists in delivering a list of documents that match a given textual query does not provide a flexible enough solution for gathering data from this new web.

The concepts that can be easily mapped on meaningful graphical representations could act as entry points into the web of data. For instance, a timeline is an appropriate representation for temporal resources. This internship is about the design of an interactive timeline to select and filter temporal resources and gather data that are linked to them. Advanced multi-scale representations and gesture-based interactions should be investigated to design a flexible and powerful navigation tool.
Context The emergence and gradual adoption of novel technologies for making data – as opposed to documents – available on the Web is likely to trigger yet another revolution in data-driven activities. The underlying data models, based, e.g., on the Resource Description Framework (RDF) and Uniform Resource Identifiers (URIs), considerably ease the interlinking of (semi-)structured data originating from multiple independent sources. How to navigate and dig into the Web of Data remains a largely open question.
Objectives The usual web search engines are built around the simple paradigm that consists in delivering a list of documents that match a given textual query. On the Web of Data, resources are interlinked objects or concepts with multiple typed properties, meaning that data are described at a much finer grain than documents are. By adding structure and semantics to the data, these novel data models give us a tremendous opportunity to rethink how we navigate and dig into data at the scale of the World Wide Web, far beyond the queries that we typically make through simple text fields using conventional search engines. In particular, resources can be time instants and intervals that could be mapped on a timeline. With recent efforts for unifying time descriptions as URIs 1 and interlinking datasets 2, we believe that these time-related resources will offer powerful entry points to query and retrieve data that have a temporal descriptive dimension.
Work program We want to investigate a novel interaction paradigm to make users able to gather knowledge from the Web of Data. Advances in HCI about multi-scale representations and gesture-based interactions for communicating with a rich vocabulary of input are especially promising for manipulating such large structured datasets. In particular, we want to design an interactive timeline on which users can perform gestures to select resources that are linked to time URIs, filter them and arrange them on screen. Starting from a multi-scale timeline, users could zoom into a time period of interest and select specific instants and/or time intervals. With an appropriate set of gestures, they could compose time intervals (intersection, union, etc.) or depict temporal behaviors (periodical, ongoing, etc.) to specify a set of temporal URIs. Thanks to the semantics associated with the data, they could also specify different ways of aggregating them to get a representation that is as meaningful as possible.
Extra information The work plan consists of:
- step 1: literature review
* time representation and linked data
* interactive visualization of timelines
- step 2: design of an interactive URI-timeline
* multi-scale timeline visualizations
* appropriate sets of gestures to specify queries and aggregations
- step 3: validation
* demonstration on a few datasets (e.g., statistical datasets 3, DBPedia 4)
* collection of user feedback
Prerequisite
Details
Expected funding Institutional funding
Status of funding Confirmed
Candidates
user caroline.appert
Created Friday 05 of December, 2014 22:33:10 CET
LastModif Monday 08 of December, 2014 11:24:23 CET


Ecole Doctorale Informatique Paris-Sud


Directrice
Nicole Bidoit
Assistante
Stéphanie Druetta
Conseiller aux thèses
Dominique Gouyou-Beauchamps

ED 427 - Université Paris-Sud
UFR Sciences Orsay
Bat 650 - aile nord - 417
Tel : 01 69 15 63 19
Fax : 01 69 15 63 87
courriel: ed-info at lri.fr