rescal_01: transformation of the real-number-RESCAL-decompositions to...

In RDFTensor: Different Tensor Factorization (Decomposition) Techniques for RDF Tensors (Three-Mode-Tensors)

Description

applies different thresholds to get a number of triples from A and R (result of RESCAL) decomposition as scale_fact*the_number_of_triples_in_original_tensor.

Usage

rescal_01(X, A, R, scale_fact = 1)

Arguments

X	is a sparse tensor as set of sparse matrices, one for every relation (predicate). (a LIST of SparseMatrix )
A	the A matrix returned by RESCAL factorization
R	the R LIST returned by RESCAL factorization
scale_fact	scale of the number of triples to be considered in the result. When it is 1 then a threshold will be taken to get the same number of triples in each slice as the original tensor.

Value

a LIST

X_	The reconstructed tensor as a set of frontal slices.
tp	the number of true positives
fp	the number of false positives
fn	the number of false negatives
sr	the details of each slice i.e the number of tp, fn, fp, etc

Author(s)

Abdelmoneim Amer Desouki

References

-Maximilian Nickel, Volker Tresp, Hans-Peter-Kriegel, "A Three-Way Model for Collective Learning on Multi-Relational Data", ICML 2011, Bellevue, WA, USA

See Also

rescal CP_01

Examples

    X1=matrix(c(1,0,0,0,0, 0,1,0,0,0, 0,0,1,1,0, 0,0,0,0,1, 1,0,0,0,0),byrow=TRUE,nrow=5,ncol=5)
    X2=matrix(c(0,1,0,1,1, 1,0,0,1,0, 0,1,0,1,1, 0,0,0,0,1, 0,0,1,0,0),byrow=TRUE,nrow=5,ncol=5)
    X2_=matrix(c(0,1,0,1,1, 1,0,0,1,0, 0,0,0,0,0, 0,0,0,0,1, 0,0,1,0,0),byrow=TRUE,nrow=5,ncol=5)
    X=list(t(X1),t(X2),t(X2_))

    N=nrow(X1)
    Xs=list()
    for(s in 1:length(X)){
      aa=which(X[[s]]==1,arr.ind=TRUE)
      Xs[[s]]=Matrix::sparseMatrix(x=rep(1,nrow(aa)),i=aa[,1],j=aa[,2],dims=c(N,N))
    }

    print(Xs)

    rf=rescal(Xs,2)
    A=rf$A
    R=rf$R
    tmp=rescal_01(Xs,A,R,scale_fact=1.5)#generate 1.5*original number of triples
    print(sprintf('Precision:%.4f, Recall:%.4f',tmp$tp/(tmp$tp+tmp$fp),tmp$tp/(tmp$tp+tmp$fn)))

RDFTensor documentation built on Jan. 16, 2021, 5:19 p.m.