WP6 meeting in Lyon
Wednesday February 2nd, 11:30 CET
Participants:
- Cefriel: Emanuele, Irene
- Siemens: Volker, Yi, Florian
- VUA: Zhisheng, Gaston
- Softgress: Raluca
Discussion
- proposal for LB-SMA data model:
- Examples of questions we want the system to answer (ADDED on 8.3.2011)
- Deductive Stream Reasoning:
- how fast/wide the tweets of somebody get replied/retweeted?
- an H-Index for twitter users?
- Inductive Stream Reasoning
- who shall "I" tweet to get an opinion of some POI?
- who is "my" maven for a given POI?
- Inductive Reasoning (recommendation with probability)
- I'm here in this neighbourhood: where shall I go eating? where can I listen some "good" music?
- Deductive Reasoning - Inductive Reasoning
abstracting knowledge before converting it to matrix, e.g. X is in Milan -> X is in Italy and use Italy as a feature in building the matrix
- Deductive Stream Reasoning:
- plan for activities about location-based social media analytics:
- mid February: annotated (part of) dataset by Saltlux
- requirements from partners about the dataset:
- 1k POIs
- 1k users
- each user should have rated 10s of POIs
- each POI should have been rated 10s of times
- data distributed over 1 year
- structured format (RDF is fine), not a collection of files
- requirements from partners about the dataset:
- end of March: WP3-WP4 plug-ins release
- mid of May (next plenary): simple running LarKC 2.0 work-flows exercising single LarKC plug-ins
- end of June: server side running
- end of July: server side running with no more bugs
- end of September: mobile application running
- ACTION ITEM: Emanuele to propose potential publication together
- mid February: annotated (part of) dataset by Saltlux
- Raluca briefly introduces the topic of instrumentation and monitoring
- we have two chances: doing it on the existing demos built on LarKC 1.x (and not going to be ported to 2.0) or instrumenting the new workflows we are about to design and develop
- we will better discuss this in WP11 session later today
- Zhisheng presents the WP4-WP6 work on reasoning on noisy data (in collaboration with Stanley from Saltlux)
- he summarizes data and ontologies
problem of noisy data --> examples: incomplete/uncertain data, inconsistent data, duplicated data
- Gaston present the formal formulation of the problem in terms of definitions of:
- partial/incomplete data
- inconsistency (duplicated data can be a special case)
- clean knowledge base (not partial and consistent)
- noisy data
- Gaston and Zhisheng present the mathematical definition of the problem on which they want to build a plug-in/workflow to test the OSM dataset "fixing"
- ACTION ITEM: Gaston and Zhisheng to discuss with Saltlux about the feasibility of developing (or not) a UI to show the result of this WP4-WP6 collaboration at the review
- TBC: add Zhisheng/Gaston slides
- discussion on traffic prediction-related paper (not enough time... we will discuss it tomorrow)
