如何在Matlab中形成邻接矩阵

Question

I am working on this code to find the adjacency matrix from the following matrix, mapr:

   'mad'     []       []      []

   'sister'  []       []      []

   'dog'     'inter'  'mad'   'said'

For the above matrix, based on the code I have written, this is the output I get which is not the desired one:

       0   1   1
       1   0   1
       1   1   0

The following is my code:

for i=1:no_of_rows
   for j=1:no_of_cols
      for m=i+1:no_of_rows
          for k=1:no_of_cols
             if(~isempty(mapr(i,j)))
               if(strcmp(mapr(i,j),mapr(m,k))==0)
                   Adjmatr(i,m)=1;
                   Adjmatr(m,i)=1;
                end
              end
          end
      end
    end
end

Can somebody help me out.Thanks in advance.

Answer 1

I think the following is what you were looking for:

mapr={   'mad',      [],    [],     [];
      'sister',      [],    [],     [];
         'dog', 'inter', 'mad', 'said'};

s1 = size(mapr,1);
s2 = size(mapr,2);
result = zeros(s1, s1);
for i = 1 : s1
  for j = 1 : s2 - 1
    if ~isempty(mapr{i,j})
        for k = i+1:s1
            for l = j+1:s2
                 if strcmp(mapr{i,j}, mapr{k,l})
                    result(i,k) = 1;
                    result(k,i) = 1;
                 end
            end
        end
     end
  end
end

The resulting matrix is

0 0 1
0 0 0
1 0 0

I think the key was to move the ~isempty out by one more loop, and not to test elements against themselves (adjacency matrix diagonal is zero)...

Answer 2

This code is a little more compact. It uses ndgrid to generate all combinations of rows, and ismember to test adjancency between rows. [] (empty matrix) needs to be converted to '' (empty string) so that ismember can be applied. Empty strings are explicitly taken care of so that they don't count for adjacency.

mapr = {'mad'    []       []      []
       'sister'  []       []      []
       'dog'     'inter'  'mad'   'said'}; %// example data

N = size(mapr,1);
mapr = cellfun(@(x) num2str(x), mapr, 'uni', 0); %// convert [] to ''
[ii,jj] = ndgrid(1:N); %// generate all combinations of rows
adjMatr = NaN(N,N); %// pre-shape result matrix
adjMatr(:) = arrayfun( @(n) ...
  any(ismember(mapr(ii(n),:), mapr(jj(n),:)) & ... %// check adjancency
  ~cellfun('isempty', mapr(ii(n),:))), 1:numel(ii) ); %// rule out empty strings 
adjMatr = adjMatr - eye(N); %// remove self-coincidences