Merge pull request #18 from achen12/origin/develop

Completing Entity Analysis

examples/2Apps/1EntityAnalysis/EA1_ReadTEST.jl

@@ -0,0 +1,24 @@
+
+
+file_dir = Pkg.dir("D4M")*"/examples/2Apps/1EntityAnalysis/Entity.csv";
+E = ReadCSV(file_dir);
+print(E[1:5,:]);
+
+~,~,doc_val      = find(E[:,"doc,"]);
+~,~,entity_val   = find(E[:,"entity,"]);
+~,~,position_val = find(E[:,"position,"]);
+~,~,type_val     = find(E[:,"type,"]);
+
+typeEntity_val = CatStr(type_val, "/" , entity_val); 
+
+E = Assoc(doc_val,typeEntity_val,position_val);
+
+using JLD
+save(Pkg.dir("D4M")*"/examples/2Apps/1EntityAnalysis/Entity.jld" , "E",E );
+
+
+print(E[1,:]);
+
+
+using PyPlot 
+spy(E');

examples/2Apps/1EntityAnalysis/EA2_StatTEST.jl

@@ -0,0 +1,16 @@
+file_dir = Pkg.dir("D4M")*"/examples/2Apps/1EntityAnalysis/Entity.jld";
+
+using JLD
+
+E = load(file_dir)["E"]
+
+sum(logical(col2type(E,"/")),1);
+
+En = sum(logical(E),1);
+
+~,entity,count = find(En);
+
+An = Assoc(count,entity,1);
+
+using PyPlot
+loglog(full(sum(Adj(An[:,StartsWith("LOCATION/,")]),2)) ,"o"); 

examples/2Apps/1EntityAnalysis/EA3_FacetTEST.jl

@@ -0,0 +1,16 @@
+file_dir = Pkg.dir("D4M")*"/examples/2Apps/1EntityAnalysis/Entity.jld";
+
+using JLD
+
+E = load(file_dir)["E"]
+
+E = logical(E)
+
+x = "LOCATION/new york,";
+p = "PERSON/michael chang,";
+F = ( nocol(E[:,x]) & nocol(E[:,p]))' * E;
+print(F' > 1 )
+
+Fn = F./ sum(E,1)
+print((Fn' > 0.02))
+

examples/2Apps/1EntityAnalysis/EA4_GraphTEST.jl

@@ -0,0 +1,17 @@
+file_dir = Pkg.dir("D4M")*"/examples/2Apps/1EntityAnalysis/Entity.jld";
+
+using JLD
+
+E = load(file_dir)["E"]
+
+
+Es = E;
+E = logical(E);
+
+Ae = sqIn(E);
+Ae
+using PyPlot
+figure;
+spy(Ae);
+
+

examples/2Apps/1EntityAnalysis/EA5_GraphQueryTEST.jl

@@ -0,0 +1,42 @@
+file_dir = Pkg.dir("D4M")*"/examples/2Apps/1EntityAnalysis/Entity.jld";
+
+using JLD
+
+E = load(file_dir)["E"]
+
+E = logical(E);
+
+A = sqIn(E);
+
+d = diag(Adj(A))
+
+A = putAdj(A,Adj(A)-diagm(d));
+
+
+####
+
+i,j,v = findnz(Adj(A));
+
+An = putAdj(A, sparse(i,j,v ./ min(d[i],d[j])));
+
+####
+
+x = "LOCATION/new york,";
+
+p = StartsWith("PERSON/,");
+
+print( (A[p,x] > 4) & (An[p,x] > 0.3));
+
+####
+
+p0 = "PERSON/john kennedy,";
+
+p1 = Row(A[p,p0] + A[p0,p]);
+
+using PyPlot
+
+spy(A[p1,p1])
+
+p2 = Row( A[p1,p1] - (A[p,p0]+ A[p0,p]));
+
+A[p2,p2] > 1