Probabilistic SHACL implementation

corese-core/src/main/java/fr/inria/corese/core/extension/Core.java

@@ -70,6 +70,8 @@ static void define(String pref, String ns) {
     static void defNamespace() {
         prefix = new HashMap<>();
         define("sh", NSManager.SHAPE);
+        // Probabilistic SHACL prefix
+        define("psh", NSManager.PROBSHACL);
         define("msh", MSH);
     }
 

corese-core/src/main/java/fr/inria/corese/core/extension/Extension.java

@@ -337,6 +337,18 @@ int fib(int n) {
         }
     }
 
-
+    /**
+     * @return the result of exp(dt), dt is a number
+     */
+    public IDatatype exponential(IDatatype dt) {
+        return DatatypeMap.newInstance(Math.exp(dt.doubleValue()));
+    }
+
+    /**
+     * @return PI value
+     */
+    public IDatatype pi() {
+        return DatatypeMap.newInstance(Math.PI);
+    }
 
 }

corese-core/src/main/java/fr/inria/corese/core/shacl/Shacl.java

@@ -42,6 +42,12 @@ public class Shacl {
     static final String PARSE = SH + "funparse";
 
     static final String SHACL = SH + "shacl";
+    // Probabilistic SHACL
+    static final String EXTENDED_SHACL = SH + "extendedshacl";
+    // modss
+    public static final int PROBABILISTIC_MODE = 1;
+    public static final int POSSIBILISTIC_MODE = 2; // not avalaible
+
     static final String SHEX = SH + "shex";
     static final String SHAPE = SH + "shaclshape";
     static final String NODE = SH + "shaclnode";
@@ -209,6 +215,35 @@ public Graph eval(Graph shacl) throws EngineException {
         return eval(SHACL, shacl);
     }
 
+    /**
+     * probabilistic SHACL Evaluation (usage for Corese GUI)
+     * @return
+     * @throws EngineException
+     */
+    public Graph eval(int mode) throws EngineException {
+        return eval(EXTENDED_SHACL, getShacl(), mode, DatatypeMap.createLiteral(String.valueOf(0.1), fr.inria.corese.sparql.datatype.RDF.xsddouble));
+    }
+
+    /**
+     * probabilistic SHACL Evaluation (usage for Corese server) with p-value
+     * @return
+     * @throws EngineException
+     */
+    public Graph eval(Graph shacl, int mode, IDatatype p) throws EngineException {
+        setShacl(shacl);
+        return eval(EXTENDED_SHACL, shacl, mode, p);
+    }
+
+    /**
+     * probabilistic SHACL Evaluation (usage for Corese server) with p-value and the number of considered triples
+     * @return
+     * @throws EngineException
+     */
+    public Graph eval(Graph shacl, int mode, IDatatype p, IDatatype nTriples) throws EngineException {
+        setShacl(shacl);
+        return eval(EXTENDED_SHACL, shacl, mode, p, nTriples);
+    }
+
     /**
      * Evaluate shape/node
      */

corese-core/src/main/resources/function/datashape/core.rq

@@ -35,6 +35,7 @@ function sh:core(report, sh, vis, nodeList, present) {
                 sh:pop()
             }
         } ;
+        sh:log("core.rq", "sh:core", concat(sh, " is conform ? ", res));
         return (res)
     }
 }
@@ -323,18 +324,43 @@ function sh:coreboolean(shape, cst) {
         }
 }
 
+# Probabilistic SHACL 
+# Binomial formulas
+function sh:normDensity(x, mu, sigma) {
+    set(normDensity_exponent = -0.5 * power(((?x - ?mu) / ?sigma), 2));
+    set(normDensity_coefficient = 1 / (?sigma * power(2 * fun:pi(), 0.5)));
+    set(normDensity_res = normDensity_coefficient * fun:exponential(normDensity_exponent));
+    return (normDensity_res)
+}
 
+# Optimized combinaison (n k)
+function sh:comb(n, k) {
+    # apply properties to optimize the computation
+    if(?n = ?k || ?k = 0) {
+        return (1)
+    };
+    # not scalable :
+    # return (sh:fac(?n) / (sh:fac(?k) * sh:fac(?n - ?k)))
+    if(?k>?n-?k) {
+        ?k=?n - ?k;
+    };
+    set(coef = 1);
+    for (i in xt:iota(0, ?k - 1)) {
+        set(coef=coef*(?n - i));
+        set(coef=floor(coef/(i+1)));
+    };
+    return (coef)
+}
 
+# P(X = k)
+function sh:binomial(n, p, k) {
+    if(?n>=30 && ?n*?p>5 && ?n*(1-?p)>5) {
+        return (sh:normDensity(?k, ?n*?p, power(?n*?p*(1-?p), 0.5)))
+    };
+    return (sh:comb(?n, ?k) * power(?p, ?k) * power(1 - ?p, ?n - ?k))
+}
 
-
-
-
-
-
-
-
-
-
-
-
-
+# P(X >= k)
+function sh:binomialEqualOrGreater(n, p, k) {
+    if(k = n, sh:binomial(?n, ?p, ?n), sh:binomial(?n, ?p, ?k) + sh:binomialEqualOrGreater(?n, ?p, ?k + 1))
+}

corese-core/src/main/resources/function/datashape/main.rq

@@ -6,7 +6,9 @@
 prefix sh:   <http://www.w3.org/ns/shacl#> 
 prefix xsh:  <http://www.w3.org/ns/shacl#> 
 prefix shex: <http://ns.inria.fr/shex/shacl#>
+prefix dct: <http://purl.org/dc/terms/>
 prefix fun:  <function://fr.inria.corese.core.extension.Extension>
+prefix psh: <http://ns.inria.fr/probabilistic-shacl/>
 
 @import  <api/define.rq> <focus.rq> <ppathparser.rq> <ppath.rq> <operator.rq> <operatorext.rq> <core.rq>  <path.rq> <constraint.rq>  <target.rq> <report.rq>
 
@@ -18,12 +20,14 @@ prefix fun:  <function://fr.inria.corese.core.extension.Extension>
 #
 function dt:graph sh:shacl() {
    if (sh:trace(), xt:print("shacl:"), true);
+   set(extended = false);
    sh:shacl(xt:graph(), sh:focus(xt:graph()))
 }
 
 # shape is shacl graph
 function dt:graph sh:shacl(dt:graph shape) {
    if (sh:trace(), xt:print("shacl:"), true);
+   set(extended = false);
    sh:shacl(shape, sh:focus(shape))
 }
 
@@ -37,6 +41,7 @@ function dt:graph sh:shaclshape(sh) {
 # Current graph is RDF and shacl graph
 function dt:graph sh:shaclShapeGraph(shacl, sh) {
    if (sh:trace(), xt:print("shaclshape:", sh), true);
+   set(extended = false);
    sh:shacl(shacl, sh:focus(shacl, sh))
 }
 
@@ -50,6 +55,7 @@ function dt:graph sh:shaclshape(sh, node) {
 
 function dt:graph sh:shaclShapeGraph(shacl, sh, node) {
    if (sh:trace(), xt:print("shaclshape:", sh, node), true);
+   set(extended = false);
    sh:shacl(shacl, xt:list(xt:list(sh, xt:list(node))))
 }
 
@@ -62,9 +68,52 @@ function dt:graph sh:shaclnode(node) {
 
 function dt:graph sh:shaclNodeGraph(shacl, node) {
    if (sh:trace(), xt:print("shaclnode:", node), true);
+   set(extended = false);
    sh:shacl(shacl, sh:focusnode(shacl, node))
 }
 
+#
+# extended shacl function : probabilistic or possibilistic mode
+#
+function dt:graph sh:extendedshacl(dt:graph shape, m, p) {
+    if (sh:trace(), xt:print("shacl:"), true);
+    set(extended = true);
+    # define mod
+    set(mode = xsd:integer(m));
+    # define p-value
+    set(prob = p);
+    # set default value and enable COUNT query on the RDF data graph
+    set(numInstances = 0);
+    # check the used mod
+    if (mode == 1, 
+        sh:log("main.rq", "sh:extendedshacl", concat("Probabilistic mode used with p=", prob)),
+        if(mode == 2, 
+            sh:log("main.rq", "sh:extendedshacl", "Possibilistic mode"), 
+            sh:log("main.rq", "sh:extendedshacl", "ERROR this ID is not found !")
+        )
+    );
+    sh:shacl(shape, sh:focus(shape));
+}
+
+function dt:graph sh:extendedshacl(dt:graph shape, m, p, nTriples) {
+    if (sh:trace(), xt:print("shacl:"), true);
+    set(extended = true);
+    # define mod
+    set(mode = xsd:integer(m));
+    # define p-value
+    set(prob = p);
+    # set the number of instances provided by user
+    set(numInstances = nTriples);
+    # check the used mod
+    if (mode == 1, 
+        sh:log("main.rq", "sh:extendedshacl", concat("Probabilistic mode used with p= ", prob, " and n-triples= ", numInstances)),
+        if(mode == 2, 
+            sh:log("main.rq", "sh:extendedshacl", "Possibilistic mode"), 
+            sh:log("main.rq", "sh:extendedshacl", "ERROR this ID is not found !")
+        )
+    );
+    sh:shacl(shape, sh:focus(shape));
+}
 
 
 function dt:graph sh:shex() {
@@ -115,20 +164,54 @@ function dt:graph sh:shacl(dt:graph shape, focus) {
 # focus = mappings or list(sh=nodeShape; list=targetNodeList)
 #
 function sh:shacleval(shape, focus) {
+    # compute the total number of instances in the graph if numInstances == 0
+    if (extended && numInstances == 0) {
+        set(numInstances = sh:getNumInstances());
+    } ;
     let (suc = true) {
         for ((sh list) in focus) {
             if (sh:trace(), 
                 xt:print("target:", coalesce(sh, "undef"), xt:size(list), list), true);
+            # set the final report as global variable
+            set(rep = sh:validationReport());
+            set(abnode = sh:bnodeid());
             if (bound(sh) && xt:size(list) > 0) {
-                let (res = sh:core(sh:validationReport(), sh, true, list)) {
+                sh:log("main.rq", "sh:shacleval", concat("xt:size(list)= ", xt:size(list)));
+                if(extended) {
+                    # set default value for reference cardinality of current shape
+                    set(referenceCardinality = 0);
+                    # set the name of the given shape
+                    set(shapeName = sh);
+                    # For each shape, we will create a list of exceptions
+                    set(exceptions = xt:list());
+                } ;
+                # start core function
+                let (res = sh:core(rep, sh, true, list)) {
+                    set(result = res);
                     if (res, true, set(suc = false))
-                }
+                } ;
+                # write the additionnal component (Extended SHACL)
+                if(extended) {
+                    sh:addExtendedShaclReport(rep, exceptions);
+                } ;
             }
+            else {
+                sh:log("main.rq", "sh:shacleval", "The current shape is not considered: no nodes to be tested");
+            } ;
         } ;
         return (suc)
     }
 }
 
+# Get the total number of instances in the dataset
+function sh:getNumInstances() {
+    let(select (count(*) as ?n) where {
+        ?s ?p ?o .
+    }) {
+        return (n)
+    }
+}
+
 function sh:focus() {
     sh:focuslist(xt:graph())
 }
@@ -731,4 +814,7 @@ function sh:checkinit() {
     if (bound(mapmap), true, sh:start(xt:graph()))
 }
 
-
+# Log a specific msg according to the parameters
+function sh:log(file, m, msg) {
+    xt:print(now()," - [",file,"] - ",m," - ",msg);
+}