Sixth Meeting

From Kolflow Project

Meeting description

Name : non-Kick-off meeting (the project has already been kicked off)

Description :

Start Date : 2013-11-07T09:30:00

End Date : 2013-AA-08T15:30:00

Laboratory : LIRIS

Location : Lyon

General information

Slides and talks can be made either in French or in English

Attending:

• Nantes: Pascal Molli, Luis Ibanez, Diego Torres (by visio)
• Nancy: Alice Hermann, Jean Lieber, Yannick and My Thao by visio
• Lyon: Amélie Cordier, Pierre-Antoine Champin,
• Nice: Olivier Corby, Rakeb Hasan, Fabien on 8

Locations

• Meeting Address : Bâtiment Nautibus. Salle C1 le jeudi, Salle C2 le vendredi.

Plan d'accès : http://fst-informatique.univ-lyon1.fr/departement/acces-et-plans/

Detailed Schedule

Thurday 7 :

• 9:00-10:00 Task1 Coordination, GDD (Project coordination)

• Tour des délivrables
• préparation de l'évaluation du 25 novembre

• 10:00-10:30 Coffee Break

• 10:30-11:30 Task3 Inferences and Interactions for dealing with Semantic Inconsistencies, Orpailleur (Design and experiment strategies for dealing with Semantic Inconsistencies)

• 1. Discussion on status of deliverables (maximum : 15 minutes) :

• Summary of task 3, by Alice Hermann (30 minutes)
  • Belief merging
  • Belief revision
  • Integrity constraint belief merging

• 12:30-14:30 Lunch

• 14:30-16:00 Task6 Scenarios and Evaluations, GDD (project evaluation)
  1. Discussion on status of deliverables:

• Présentation Diego Torres par Skype sur "BlueFinder"

• 16:00-16:30 Coffee Break

• 16:30-18:00 Task2 Continuous extraction, edition, and annotation, Orpailleur (extracting knowledge units from texts using data mining)

• 1. Discussion on status of deliverables:

• Présentation Yannick Toussaint/Mi thao.

• 19h30 Dinner " ?

Friday 8

• 09:00-10:30 Task5 Social semantic space, Silex (Design and experiment of a social semantic space)

• 1. Discussion on status of deliverables:

• Presentation: Luis Ibanez "Writable Linked Data"

• 10:30-10:45 Coffee Break

• 10:45-12:00 Task4 Traces and explanations: documenting inferences, query solving and interactions, Wimmics (Design and experiment tracing and explanation approaches)

1. Discussion on status of deliverables:

• Presentation Rakeb

Title: Predicting SPARQL query execution time In this talk we discuss the problem of predicting SPARQL query execution time. Accurately predicting query execution time is central to effective query scheduling, resource management, and query optimization. We use a machine learning approach to predict query execution time. We generate the training dataset from real queries collected from DBPedia 3.8 query logs. In our experiments, we use Apache Jena Fuseki with Jena TDB as a triple store as the system configuration to generate the query execution time for training queries. As features of a query, we use the SPARQL query algebra operators and different query types that we generate by clustering the training SPARQL queries. We cluster the training SPARQL queries into different types based on the basic graph patterns that appear in the queries. We use k-medoid clustering algorithm and approximate graph edit distance algorithm as a distance function for clustering. We train a regression model using the training the dataset with SPARQL query algebra and query types features for predicting SPARQL query execution time.

• 12:30-14:30 Lunch

• 14:30-15:30 Discussion

• 15:30 Coffee Break & End of the day