DESIRR. Semantic platform for organizing, sharing and reusing R functions

Transcription

1 DESIRR Semantic platform for organizing, sharing and reusing R functions Pascal Neveu a, Yuan Lin a, Caroline Domerg a, Juliette Fabre a, Alexandre Granier a, Anne Tireau a, Vincent Negre a, Isabelle Mirbel b, Olivier Corby c, Catherine Faron-Zucker b pascal.neveu@supagro.inra.fr a INRA, Montpellier, France b I3S, University of Nice, Sophia Antipolis, France c INRIA, Sophia Antipolis Méditerrane, Sophia Antipolis, France

2 Context 2! Overview! R functions! Great success in different scientific domains! High production of R functions! Authors from different sites, with authors turn over! Challenge: Open data (function)! Share and reuse

3 Context 3! Objectives! Store and organize available R functions! Give an easy and long-term access! How! Create formal description for R functions based on ontological approach - Define concepts and relations between concepts (Domain expertise) - Controlled vocabulary for management and search facility - Build a function repository with powerful reasoning and search capabilities! Technologies! Methods and tools of Semantic Web (W3C) (Knowledge organization and management)

4 Ontology (Definition) 4 Leaf_Growth_Analysis <- function(para1, para2 ) { } hasauthor

5 Ontology (Definition) 4 Leaf_Growth_Analysis <- function(para1, para2 ) { } hasauthor Triplet representation Subject Predicate Object <#Leaf_Growth_Analysis.R> <#hasauthor> <#Vera>

6 Ontology (Definition) 4 Leaf_Growth_Analysis <- function(para1, para2 ) { } hasauthor Triplet representation Subject Predicate Object <#Leaf_Growth_Analysis.R> <#hasauthor> <#Vera> Graph representation <#Leaf_Growth_Analysis.R> <#hasauthor> <#Vera>

7 Ontology (Definition) 5 dataframe xxxxx <#defaultvalue> <#datatype> <#RFunction> <#para1> <#Person> vera@inra.fr <#type> <#hasargument> <#type> <#has > <#Leaf_Growth_Analysis.R> <#hasauthor> <#Vera>

8 Ontology (Definition) 6

9 Ontology (Model-Data) 7 Model <#RFunction> <#hasauthor> <#Person>

10 Ontology (Model-Data) 7 Model <#RFunction> <#hasauthor> <#Person> Data <#LeafGrowth Analysis.R> <#hasauthor> <#Vera>

11 R function ontology (overview) 9! General desc, Detailed info, Relationship between R functions #intention Argument Has type Has description Has default value #type #description #defaultvalue #Rpackage Requires R package Has Intention Has argument Value Has type Has description Has value #documentation Has documentation R Function Calls R core function #RcoreFunction Has creation date Has repository date #date Has name Has description Has author Has contact Has audience Calls local R function Could be used after Could be used before Is adapted from Is a new version of Looks like #name #description Person Belongs to Group

12 R function ontology 10! General description name, creationdate, desc, author, contact, audience, documentation #description Person #name Has description Has author Has contact Has name Has creation date R Function Has documentation Has audience Belongs to Group #date #documentation

13 R function ontology 11! Detailed information requirespackage, hasargument, hasvalue, hasintention, etc. #intention Value Has Intention Has value Has type Has description R Function Requires R package #Rpackage Has argument Argum ent #type #description #defaultvalue Has type Has description Has default value

14 R function ontology 12! Relationship between R functions callsrcorefunc, callslocalfunc, couldbeusedbefore/after, lookslike, isadaptedfrom, isnewversionof. R Function transitivity Calls R core function Could be used after Could be used before Looks like Is adapted from Is a new version of Calls local R function inverseof symmetry transitivity #RcoreFunction

15 R function ontology 12! Relationship between R functions callsrcorefunc, callslocalfunc, couldbeusedbefore/after, lookslike, isadaptedfrom, isnewversionof. R Function transitivity Calls R core function Could be used after Could be used before Looks like Is adapted from Is a new version of Calls local R function inverseof symmetry transitivity #RcoreFunction Creation of the call graph

16 Application (architecture) 13! Application web (works with a web browser)! R terminal (parser and validate the uploaded R files)! Description storage in the server and in a specific data base! Version and concurrency control Web browser Web application for R Annotations (PHP/Apache Server) RDF Annotations R Functions Triple store (CORESE INRIA) Datasets, doc... Versioning (Subversion)

17 14! New description creation

18 14! New description creation Upload a R func file Parse & validate by R client Fill the desc form

19 14! New description creation Upload a R func file Parse & validate by R client Fill the desc form

20 14! New description creation Upload a R func file Parse & validate by R client Fill the desc form

21 14! New description creation Upload a R func file Parse & validate by R client Fill the desc form

22 14! New description creation Upload a R func file Parse & validate by R client Fill the desc form Save the R func & its desc

23 15! Function consultation

24 15! Function consultation Input search conditions Example: isdedicatedto 'Phenodyn' and hasintention of Visualization

25 15! Function consultation Input search conditions Generalize and run SPARQL query Example: isdedicatedto 'Phenodyn' and hasintention of Visualization

26 15! Function consultation Input search conditions Generalize and run SPARQL query Show result list

27 15! Function consultation Input search conditions Generalize and run SPARQL query Show result list Consult & modify descriptions

28 15! Function consultation Input search conditions Generalize and run SPARQL query Show result list Consult & modify descriptions Call graph of the current function Relations with other functions

29 Application (inference) 16 Leaf_Growth_Computation.R Could be used after Growth_Curve_Fitting.R Calls local R function Leaf_Growth_Analysis.R Calls local R function Leaf_Growth_Modeling.R

30 Application (inference) 16 Leaf_Growth_Computation.R INVERSE OF Could be used after Could be used before Growth_Curve_Fitting.R Calls local R function Leaf_Growth_Analysis.R Calls local R function Leaf_Growth_Modeling.R

31 Application (inference) 16 Leaf_Growth_Computation.R INVERSE OF Could be used after Could be used before TRANSITIVITY Growth_Curve_Fitting.R Calls local R function Calls local R function Leaf_Growth_Analysis.R Calls local R function Leaf_Growth_Modeling.R

32 17! Inference example 1 - Upload Add.R - Upload Add2.R Add2.R adaptedfrom Add.R - Upload Add3.R Add3.R adaptedfrom Add2.R Add.R adaptedfrom Add2.R adaptedfrom Add3.R

33 17! Inference example 1 - Upload Add.R - Upload Add2.R Add2.R adaptedfrom Add.R - Upload Add3.R Add3.R adaptedfrom Add2.R Add.R adaptedfrom Add2.R adaptedfrom Add3.R Add3.R is adapted from which function(s)?

34 17! Inference example 1 - Upload Add.R - Upload Add2.R Add2.R adaptedfrom Add.R - Upload Add3.R Add3.R adaptedfrom Add2.R Add.R adaptedfrom Add2.R adaptedfrom Add3.R Add3.R is adapted from which function(s)?

35 17! Inference example 1 - Upload Add.R - Upload Add2.R Add2.R adaptedfrom Add.R Add.R adaptedfrom Add2.R adaptedfrom - Upload Add3.R Add3.R adaptedfrom Add2.R adaptedfrom Add3.R Add3.R is adapted from which function(s)?

36 17! Inference example 1 - Upload Add.R - Upload Add2.R Add2.R adaptedfrom Add.R Add.R adaptedfrom Add2.R adaptedfrom - Upload Add3.R Add3.R adaptedfrom Add2.R adaptedfrom Add3.R Add3.R is adapted from which function(s)?

37 18! Inference example 2 - Upload Func1 Func1 usedbefore - Upload Func2 Func2 couldbeusedbefore Func1 - Upload Func3 Func3 couldbeusedafter Func2 Func2 usedafter Func3

38 18! Inference example 2 - Upload Func1 Func1 usedbefore - Upload Func2 Func2 couldbeusedbefore Func1 - Upload Func3 Func3 couldbeusedafter Func2 Func2 usedafter Func3 What are functions that could be used after Func2

39 18! Inference example 2 - Upload Func1 Func1 usedbefore - Upload Func2 Func2 couldbeusedbefore Func1 - Upload Func3 Func3 couldbeusedafter Func2 Func2 usedafter Func3 What are functions that could be used after Func2 Func3

40 18! Inference example 2 - Upload Func1 Func1 usedbefore usedafter - Upload Func2 Func2 couldbeusedbefore Func1 - Upload Func3 Func3 couldbeusedafter Func2 Func2 usedafter Func3 What are functions that could be used after Func2 Func3

41 18! Inference example 2 - Upload Func1 Func1 usedbefore usedafter - Upload Func2 Func2 couldbeusedbefore Func1 - Upload Func3 Func3 couldbeusedafter Func2 Func2 usedafter Func3 What are functions that could be used after Func2 Func3 Func1

42 Conclusion 19! Users find this repository relevant (efficient search, easy annotating)! Semantic Web tools allow reasoning for an intelligent repository! Models and software are easy to adapt :! for other research fields! for other programming languages! for mathematical models?

43 THE END Thank you