DesignPatternCampParis2011

Design Pattern Camp, Paris, France, 28-29 October 2011

The 1st event dedicated to Design Pattern France , will take place in IRI's conference room in Paris the 28th and 29th October 2011.

The hashtag for the vocamp is #DPC

Outcomes

This VoCamp on Flickr : photos tagged with #dpc

Follow on Twitter : twits with the tag #dpc

Follow on Identi.ca : notices with the tag #dpc

Follow on FriendFeed : feeds with the tag #dpc

...to be completed after the Camp...

Call for participation/Appel à participation

English version:

For an English description: http://ontologydesignpatterns.org/wiki/WOP:2010

French version:

A mesure que l'intérêt pour le Web Sémantique croît, et que les technologies développées pour transformer cette vision en réalité acquièrent une certaine maturité, le besoin d’ontologies de bonne qualité, réutilisables, se fait de plus en plus sentir. Pour saisir à bras le corps les question de la qualité et de la réutilisabilité, différents types de design patterns1 d’ontologie (Ontology Design Patterns, ODP) ont vu le jour. Les patterns offrent aux designers des bénéfices variés, allant de l’analyse des exigences liées à la réutilisabilité en passant par clarification des lignes directrices ou l’amélioration de la communication. Les ODP sont en passe d’apporter des bénéfices identiques. Quelques-uns ont été soumis par le W3C lui-même et sont actuellement rassemblés dans divers entrepôts, à l’instar du catalogue maintenu par l’université de Manchester ou du portail [ontologydesignpatterns.org]. Cependant, les catalogues de patterns demeurent encore relativement modestes à l’heure actuelle, ne couvrant qu’un nombre restreint de domaines. Les applications du Web Sémantique tireraient donc fortement parti de l’introduction de nouveaux types de patterns, tels les patterns relatifs à la connaissance ou les patterns plus spécifiquement liés à ces applications. En outre, une bonne communication suppose que tous ces patterns soient partagés au sein d’un communauté disposant d’un langage commun, ceci afin d’être en mesure de débattre des (nombreuses) difficultés que soulève l’activité de modélisation. Ce Design Pattern Camp entend tirer parti de l’activité du portail [ontologydesignpatterns.org] pour disposer d’un moyen de communication (autorisant aussi bien la soumission de nouveaux patterns que leur évaluation) susceptible d’héberger les discussions qui ne manqueront d’excéder les limites d’un événement ponctuel tel que celui-ci.

La réutilisabilité constitue depuis de nombreuses années un sujet de choix, aussi bien dans le domaine de l’ingénierie des ontologies que du Web Sémantique. L’utilisation des patterns fournit une méthode qui a démontré ses bienfaits dans de nombreux domaines, à commencer par le génie logiciel et la modélisation des données. Au cours des dernières années écoulées, des patterns pour les ontologies du Web Sémantique et les applications à base d’ontologies ont été introduits. Nous croyons donc venu le temps, pour la communauté, d’en récolter les bénéfices, y compris par un examen des méthodes permettant d’extraire de tels patterns.

De nombreuses initiatives ont d’ores et déjà été conduites dans ce but. Si ce “Design Pattern Camp” constitue une première en France, des précédents existent néanmoins au plan international avec l’organisation en 2005 du workshop “Ontology Patterns for the Semantic Web”, lors de la conférence ISWC 2005. A l’époque, la communauté mobilisée autour de ces questions était néanmoins nettement plus restreinte qu’aujourd’hui. L’accent était alors mis sur la réutilisation d’ontologies RDF ou OWL pour résoudre des problème généraux, relativement ouverts. La séries des WOP (Workshops on Ontology Patterns) en a élargi la portée jusqu’à inclure tous les patterns liés au design d’ontologies et à l’ingénierie des connaissance pour le Web Sémantique. Ceci s’inscrivant dans la continuité de la conférence EKAW2008 (portant le sous-titre “Knowledge Patterns”). Parmi ses thématiques, on recensait, outre les patterns du génie logiciel, les patterns destinés à la refonte des processus, le traitement des connaissances ou encore les aspects sociaux et cognitifs de la sémantiques. Le premier atelier WOP se tint en marge d’ISWC 2009, et le second l’an passé.

Un atelier, et à plus forte raison un “Camp” tel que celui-ci, se veut riche en interactions et échanges pratiques. Le format des VoCamps se prêtait donc particulièrement bien à l’exercice, propice à la fois à l’initiation de celles et ceux désireux d’en apprendre plus, comme à la réunion de praticiens souhaitant s'aguerrir, échanger et participer au développement de nouveaux patterns, à l’évaluation de patterns existants, comme au développement de méthode d’extraction de patterns. Au-delà de ces quelques considérations, l’occasion est offerte à la communauté française du Web Sémantique, dans la droite lignée du Meetup des Rencontres du Web de données du 16 mai dernier ([1]), de bénéficier de la meilleure des introductions qui soit à ces problématiques, grâce à la présence et à l’expertise de deux des chercheurs les plus reconnus en la matière, Aldo Gangemi et Valentina Presutti.

1Les patterns sont utilisés dans de nombreux domaines tels des patrons, des gabarits (templates) ou des descriptions abstraites encodant les meilleurs pratiques d’un domaine. L’introduction des patterns en informatique est généralement associée à l’oeuvre de l’architecte Christopher Alexander. Les patterns logiciels sont sans doute mieux connus sous le vocable “Design Patterns”, comme en atteste l’ouvrage du Gang of four de 1995, Design Patterns: Elements of Reusable Object-Oriented Software. Parmi les autres patterns, on recense les patterns associés aux modèles de données, les patterns linguistiques, les patterns de connaissance, et bien d’autres.

When

Friday 28th and Saturday 29th October 2011.

Program

The following are outline timings for the event which are unlikely to change. Apart from these timings, exactly how we spend the time will largely be determined before and during the event.

Day 1 (Friday 28th October)

09:00 - 9:30 Presentation of IRI(Alexandre Monnin)

09:30 - 9:40 Planning of VoCamp time (Alexandre Monnin)

09:40 - 10:05 A Hitchhiker's Guide to Ontology Design Pattern

10:05 - 10:30 One minute introductions of participants over coffee and croissants.

10:30 - 11:00 Presentation of Topics.

11:00 - 12:00 Work in Groups

12:00 - 14:00 Lunch

14:00 - 17:15 Work in Groups

17:15 - 18:00 Groups present their progress

Day 2 (Saturday 29th October)

09:30 - 12:00 patterns, patterns, patterns 12:00 - 12:30 wrap up by groups

12:30 - 14:00 Lunch

Where

Venue (meeting place)

Centre George Pompidou, Salle Piazza (near the red cube), Paris:

Travel and how to arrive

Paris is well connected by air, and train lines. You may get instructions on how to get there on WikiTravel. Once you reach the Métro system aim for lign 11 and get out on the station "Rambuteau" which leads to the Centre Pompidou. Once you get out, turn to the right, then to the left to reach the Place Pompidou. Go down the stairs, go ahead and the entrance to the Salle Piazza will be near a giant red cube(!).

You may find the map of the Métro lines on the RATP website. This website provides the timetables for all the means of transportation in Paris.

You may also use buses to get around Paris. They tend to be much slower, but can be much more comfortable than the Métro. In the springtime they provide a marvelous way to enjoy the beautiful streets of Paris.

Accommodation

We have contacted a number of Hotels and Hostels in an effort to provide a discount for participants, but have had no success, as the number of participants in not interesting for them! Therefore, we decided not to recommend any of the hotels/hostels. You may find an accommodation easily using Google, Trip Advisor or similar service.

Hotels

Hotels near Pompidou on Google on TripAdvisor

Hostels

Hostels near Pompidou on Google

Topics

Here people can suggest vocabularies and other things to work on. Just put a topic in the list and other people can put their names behind it to indicate interest in that topic.

Patterns I want to work on

(list topics of interest and name(s) of people interested in each topic)

• Valentina proposes to have two main groups, one working on top-down modeling, the other working on bottom-up extraction of patterns from lod and web ontologies.

• Tagging patterns (I'd like to do some work on the NiceTag ontology developed during the VoCamp in Nice). (Alexandre Monnin)

• Design Patterns and philosophy: is the concept also useful in philosophy? (Alexandre Monnin)

• Information realization (Alexandre Monnin)

• Modelling of 'context' - for complex objects as well as on the triple/quad-level (Thomas Lörtsch)

Working Groups

We have 5 groups working on defining patterns on 5 different domains. They are applying a mixed approach i.e. top-down/bottom-up.

Useful links:

• Content ontology design patterns
• Wikilink Patterns
• Aemoo: Exploratory search based on wikilink patterns

Geo-Politics

We're trying to examine the intricate links between organizations and territories, in particular in the context of data publication.

Overlap, relevance, imbrication... these patterns seem to apply both to territories and institutions. Our goal is to articulate the two domains and identify differences (like the lack of relevance of mereological structures for institutions in contradistinction with territories or the relevance of dependence relations - and a substitution one - for institutional objects). We also encountered a publication pattern that has a lot to do with trust, provenance, etc.

First day: We worked mainly from a top-down perspective and ended up with a schema.

Second day: From a bottom-up perspective, we compared the schema to existing patterns (regionoverlap.owl, place.owl, partof.owl, parameterregion.owl, componency.owl, dependence.owl, trust.owl, actingfor.owl on http://ontologydesignpatterns.org/cp/owl/).

See also: Parts and Places: The Structures of Spatial Representation, Roberto Casati & Achille C. Varzi.

Medicine

Team members: Andrea Nuzzolese, Gaoussou Camara and Rim Djedidi

Work outline:

First step: Bottom-up approach: extracting medical patterns from resources in the web

Second step: Top-down approach: modelling patterns for a specific task (disease control)

Third Step: general evaluation of defined and extracted patterns

1) Bottom-up step

Sources: diseasome database connecting disease, gene and drug resources. Dataset: http://thedatahub.org/dataset/fu-berlin-diseasome List of all diseases: http://www4.wiwiss.fu-berlin.de/diseasome/directory/diseases Disease resource example (Alzheimer disease): http://www4.wiwiss.fu-berlin.de/diseasome/page/diseases/74 Related drug dataset (examples): http://www4.wiwiss.fu-berlin.de/dailymed/page/drugs/1075 http://www4.wiwiss.fu-berlin.de/drugbank/page/drugs/DB00155 List of all genes: http://www4.wiwiss.fu-berlin.de/diseasome/directory/genes

DrugBank dataset describing a list of drugs Dataset: http://www4.wiwiss.fu-berlin.de/drugbank/directory/drugs

Aemoo application research engine aggregating knowledge from diverse sources such as linked data, Wikipedia, dbpedia, Twitter, and Google News. Web application: www.aemoo.net Examples: http://wit.istc.cnr.it/aemoo/search?entityUri=http://dbpedia.org/resource/Malaria# http://wit.istc.cnr.it/aemoo/search?entityUri=http://dbpedia.org/resource/Schistosomiasis

Existing patterns in ODP Portal (in ekp repository) Disease: http://ontologydesignpatterns.org/ekp/owl/Disease.owl Drug: http://ontologydesignpatterns.org/ekp/owl/Drug.owl

Description: First, we have built an ontology containing all extracted patterns: disease, drug, gene, chromosomal location, etc. with their associated data and object properties. We discussed reusability of existing patterns (drug and disease existing odp). The purpose of the bottom-up step is to extract patterns that could be reused in the context of disease modelling task. We decided to build 3 distinguished patterns describing disease, drug and gene. For user comprehension, we associate a label to classes. And for reuse, we put entity URI of disease (from diseasome dataset), drug (from drugbank dataset) and gene (from diseasome dataset) as defined in existing datasets and relate them as equivalent classes.

2) Top-down

We apply a top-down approach to model disease pattern for disease control task. We identify concepts and relations involved in this task and build a preliminary disease pattern.
Then, we reuse and specialize disease and drug knowledge patterns of ODP (ekp repository).

Architecture

The problem:

For several years architects have been using parametric modelling (like generative components, digital project, or grasshopper). Modelling can be difficult because it requires specific skills in geometry and programming, but there are a lot of online resources that can help. The community is proactive and shares a lot of solutions, pieces or models etc. To help parametric modelling Robert Woodbury proposes to use a pattern approach. For the specific case of Grasshopper several catalogues of patterns are being developed. The problems with this approach are:

• How the user is going to find the right patterns for his task.
• How can the patterns in the catalogue be described.
• How do the patterns actually correspond to what users are doing, i.e. what they are posting online.

Bottom-up approach:

Attempting to address the third bullet above, in order to discover potential new patterns or common solutions that correspond to patterns of the catalogue. Idea for a process:

• Collect online postings of models.
• Parse and extract relevant part of the XML-file.
• Model the pattern in a form for comparing it to other patterns, e.g. a graph or possibly a table of features.
• Represent the pattern on different levels of abstraction, i.e. some components may be useful to generalize using a "taxonomy of components".
• Analyze all the pattern models to find clusters or occurrences of common substructures, i.e. patterns.

Concerning the data collection, we discovered that the texts describing the posted models were very diverse and not always very useful to describe the solution. This is why we chose to focus on the files themselves. The XML of the file needs to be parsed and cleaned from the irrelevant data, e.g. information on how to display the pattern graphically in the tool. This was difficult to do by hand, hence more information about the schema of the XML file would be needed and a tool should be created. Instead we derived a model of the pattern directly from its graphical representation in the tool. We tried to model it as a graph, where nodes are components or parameters and edges are their input/output connections. For the analysis we had the idea to use a graph clustering tool, to analyze similarities of the solutions, or a "graph pattern mining" tool to find common substructures inside the patterns. We did not manage to find appropriate tools now, so we did a first manual comparison of the graphs. It turns out that there seems to be common substructures in the files belonging to the same (or similar) patterns, but a better analysis should be the next step.

Top-down approach:

We would like to see how the above approach can be mixed with a top-down one. We will first investigate the top-down approach and then see if they can be combined.

Music

Jamendo and John Peel Sessions patterns

How Much

The VoCamp event itself is free, although participants will need to pay for their own travel and accommodation.

Who

Hosted by

IRI & Centre Pompidou

Organisers

• Aldo Gangemi
• Valentina Presutti
• Alexandre Monnin

Participants (there are 20 places)

please put a link to your page, give an e-mail or send an e-mail to alexandre.monnin [at] iri.centrepompidou.fr

• Florent André.
• Olivier Rovellotti
• Hondjack Dehainsala
• Amina Chniti
• Hacene Cherfi
• Aurélie de Boissieu
• François Pelé
• Nouha Omrane
• Peter Livaudais
• Sarra Ben Abbès
• Hélène Rabault
• Eva Blomqvist
• Gaoussou Camara
• Rim Djedidi
• Emmanuel Chateau
• Joaquim Silvestre
• Andrea Nuzzolese
• François Hardel

Would like to, but can't

• Laurent Lefort - May I suggest one topic? Graphical representation for ODPs - other than UML-based!
• Thomas Lörtsch - C'est la m...aladie :( I most probably won't be able to make it. So sorry!
• Sylvie Despres : some unexpected constraints prevent me from participating.

Call for Participation

Subject :

Design Pattern Camp, creating, reviewing patterns, Paris, France, 28th and 29th October 2011

Body:

The 1st Design Patterns Camp in France, will take place in Paris the 28th and 29th October 2011.
VoCamp is a series of free informal events where people can spend some time creating and maintaining lightweight
vocabularies/ontologies/thesauruses for the Semantic Web/Web of Data/Linked Open Data. (see http://vocamp.org/ )
The VoCamp idea is influenced by BarCamp but is oriented to hands-on technical work and practical outputs to publish
new vocabularies. The emphasis of the events is not on creating the perfect ontology in a particular domain, but
on creating vocabularies that are good enough for people to start using for publishing data on the Web.
VoCamps are free for participants. Sign up by going to the website (attendance is limited to 25 people at the
moment so be quick)http://vocamp.org/wiki/DesignPatternsCampParis2011