wolffd@0: % ---
wolffd@0: % class ClipSimGraphMD
wolffd@0: % Directed graph representing comparative clip similarities.
wolffd@0: % EACH pair of vertices has just ONE directed edge connecting them
wolffd@0: %
wolffd@0: % each node represents a pair of clips,
wolffd@0: % an edge (a,b) -> (a,c) is present if there is at least one
wolffd@0: % users judging clip a more similar to clip b than to clip c
wolffd@0: % 
wolffd@0: % The Edges thus represent the (are nearer to each othen than) 
wolffd@0: % expression
wolffd@0: % ---
wolffd@0: 
wolffd@0: classdef ClipSimGraphMD < ClipSimGraph & handle
wolffd@0: 
wolffd@0:     
wolffd@0: properties
wolffd@0: 
wolffd@0:     % ---
wolffd@0:     % History of edge weights, is a nodes x nodes cell.
wolffd@0:     %
wolffd@0:     % E_hist(i,j) = m x 2 Matrix containing
wolffd@0:     % Pair of information for each set [votes votes_complete]
wolffd@0:     hE; % History of edge weights
wolffd@0: end
wolffd@0: 
wolffd@0: % ---
wolffd@0: %  the methods
wolffd@0: % ---
wolffd@0: methods
wolffd@0:         
wolffd@0: % constructor of the graph
wolffd@0: function G = ClipSimGraphMD(comparison, comparison_ids)
wolffd@0:     
wolffd@0:     if nargin == 0
wolffd@0:         % ---
wolffd@0:         % empty constructor
wolffd@0:         % ---
wolffd@0:         
wolffd@0:     elseif nargin == 1
wolffd@0:         % todo: copy graph 
wolffd@0:         
wolffd@0:     elseif nargin == 2
wolffd@0: 
wolffd@0: 
wolffd@0:         % ---
wolffd@0:         % handle automatic or manual input
wolffd@0:         % ---
wolffd@0:         for i = 1:size(comparison,1)
wolffd@0: 
wolffd@0:             % get clips and votes
wolffd@0:             clips = comparison_ids(comparison(i,1:3));
wolffd@0:             votes = comparison(i,4:6);
wolffd@0: 
wolffd@0:             % for each triplet position create an edge reflecting the 
wolffd@0:             % absolute and relative voting for this position
wolffd@0: 
wolffd@0:             % ---
wolffd@0:             % NOTE: we index 0 - 2 to ease the mod
wolffd@0:             % calculaton for the remaining indices
wolffd@0:             % ---
wolffd@0:             for v = 0:2
wolffd@0: 
wolffd@0:                 % ---
wolffd@0:                 % has at least one user voted this clip out?
wolffd@0:                 % If so, treat it as an outlier and determine score
wolffd@0:                 % ---
wolffd@0:                 if votes(v+1) > 0
wolffd@0: 
wolffd@0:                     a = mod(v+1, 3)+1;
wolffd@0:                     b = mod(v+2, 3)+1;
wolffd@0:                     c = v+1;
wolffd@0: 
wolffd@0: 
wolffd@0:                     % ---
wolffd@0:                     % every outlier vote results in two
wolffd@0:                     % dissimilarity equations, favored by 
wolffd@0:                     % the people who voted for the outlier
wolffd@0:                     % ---
wolffd@0: 
wolffd@0:                     % edge 1     
wolffd@0:                     add_edge(G, clips(a), clips(b), clips(c), votes(v+1), sum(votes));
wolffd@0: 
wolffd@0:                     % edge 2
wolffd@0:                     add_edge(G, clips(b), clips(a), clips(c), votes(v+1), sum(votes));
wolffd@0:                 end
wolffd@0:             end
wolffd@0: 
wolffd@0:         end
wolffd@0:     end
wolffd@0:     
wolffd@0: % end constructor function
wolffd@0: end
wolffd@0: 
wolffd@0: % adds a node stating clip a is more near %
wolffd@0: % to clip b then clip c
wolffd@0: function Vid = add_node(G, a, b)
wolffd@0: 
wolffd@0:     Vid = add_node@ClipSimGraph(G, a, b);
wolffd@0:     
wolffd@0:     G.hE{Vid,Vid} = [];
wolffd@0: 
wolffd@0: end
wolffd@0: 
wolffd@0: function remove_node(G, a, b)
wolffd@0: 	
wolffd@0:     if nargin == 3
wolffd@0:         
wolffd@0:         V1 = G.node(a, b);
wolffd@0:     elseif nargin == 2
wolffd@0:         
wolffd@0:         V1 = a;
wolffd@0:     end
wolffd@0:     
wolffd@0:     if ~isempty(Vid)
wolffd@0: 
wolffd@0:         % ---
wolffd@0:         % look for edges connected to Vid
wolffd@0:         % and delete their histograms
wolffd@0:         % ---
wolffd@0:         G.hE(:,Vid) = [];
wolffd@0:         G.hE(Vid,:) = [];
wolffd@0: 
wolffd@0:         % ---
wolffd@0:         % TODO: Efficiently Remove the actual node from the
wolffd@0:         % GRAPH
wolffd@0:         % ---
wolffd@0:         G.Vclips(Vid,:) = [];
wolffd@0: 
wolffd@0:     end
wolffd@0: end
wolffd@0: 
wolffd@0: % ---
wolffd@0: % add an edge saying sim(a,b) > sim(a,c)
wolffd@0: % ---
wolffd@0: function add_edge(G, a, b, c, votes, allvotes)
wolffd@0:     
wolffd@0:     V1 = add_node(G, a, b);
wolffd@0:     V2 = add_node(G, a, c);
wolffd@0:  
wolffd@0:     % ---
wolffd@0:     % save new weight values into histogram
wolffd@0:     % ---
wolffd@0:     if isempty(G.hE(V1, V2))
wolffd@0: 
wolffd@0:         G.hE{V1,V2} = [votes allvotes];
wolffd@0:     else
wolffd@0:         G.hE{V1,V2} = [G.hE{V1,V2}; votes allvotes];
wolffd@0:     end
wolffd@0:     
wolffd@0:     % ---
wolffd@0:     % update Edges
wolffd@0:     % ---
wolffd@0:     G.update_E(a, b, c);
wolffd@0:     
wolffd@0: end
wolffd@0: 
wolffd@0: function remove_edge(G, a, b, c)
wolffd@0: 
wolffd@0:     if nargin == 4
wolffd@0:         
wolffd@0:         V1 = G.node(a, b);
wolffd@0:         V2 = G.node(a, c);
wolffd@0:     elseif nargin == 3
wolffd@0:         
wolffd@0:         V1 = a;
wolffd@0:         V2 = b;
wolffd@0:     end
wolffd@0:     
wolffd@0:     if ~isempty(V1) && ~isempty(V2)
wolffd@0: 
wolffd@0:         % set Edge to zero
wolffd@0:         G.hE{V1, V2} = [];
wolffd@0:     else
wolffd@0: 
wolffd@0:         error 'nodes not found';
wolffd@0:     end
wolffd@0: end 
wolffd@0: 
wolffd@0: % ---
wolffd@0: % updates the edge weight given 3 clip ids or 
wolffd@0: % two nodes, based on the edges history
wolffd@0: %
wolffd@0: % The specified (V1,V2) and the symmetric edges' (V2,V1) weights
wolffd@0: % are evaluated and the stronger edge will get 
wolffd@0: % the excessive weight while the loosing edge 
wolffd@0: % will be deleted
wolffd@0: % ---
wolffd@0: function update_E(G, a, b, c)
wolffd@0: %  update_E(G, a, b, c)
wolffd@0: %  update_E(G, V1, V2)
wolffd@0:     
wolffd@0:     % determine the type of input parameters
wolffd@0:     if nargin == 4
wolffd@0:         
wolffd@0:         V1 = G.node(a, b);
wolffd@0:         V2 = G.node(a, c);
wolffd@0:     elseif nargin == 3
wolffd@0:         
wolffd@0:         V1 = a;
wolffd@0:         V2 = b;
wolffd@0:     end
wolffd@0:     
wolffd@0:     % ---
wolffd@0:     % calculate weighted sum for specified edge
wolffd@0:     % and  for symmetric edge
wolffd@0:     % ---
wolffd@0:     thisw = unbal_edge_weight(G, V1, V2);
wolffd@0:     symmw = unbal_edge_weight(G, V2, V1);
wolffd@0: 
wolffd@0:     % ---
wolffd@0:     % the two //competing// weights are now balanced
wolffd@0:     % ---
wolffd@0:     w = thisw - symmw;
wolffd@0:     
wolffd@0:     % ---
wolffd@0:     % set both edges
wolffd@0:     % ---
wolffd@0:     G.E(V1,V2) = max(w,0);
wolffd@0:     G.E(V2,V1) = -min(w,0);
wolffd@0:     
wolffd@0: %     if w >= 0
wolffd@0: %         G.E(V1,V2) = w;
wolffd@0: %         G.E(V2,V1) = 0;
wolffd@0: %     elseif w < 0
wolffd@0: %         G.E(V1,V2) = 0;
wolffd@0: %         G.E(V2,V1) = -w;
wolffd@0: %     end
wolffd@0: end
wolffd@0: 
wolffd@0: end
wolffd@0: 
wolffd@0: methods (Hidden)
wolffd@0:     
wolffd@0:     % ---
wolffd@0:     % This is the core function of this Graph. 
wolffd@0:     % it allows for the calculation of a single Edge,
wolffd@0:     % before it is balanced with its symmetrical counteredge
wolffd@0:     %
wolffd@0:     % ---
wolffd@0:     function thisw = unbal_edge_weight(G, V1, V2)
wolffd@0:         % ---
wolffd@0:         % trivial cases
wolffd@0:         % ---
wolffd@0:         if isempty(G.hE(V1, V2)) || isempty(G.hE{V1, V2})
wolffd@0:             thisw = 0;
wolffd@0:             return;
wolffd@0:         end
wolffd@0:         
wolffd@0:         % ---
wolffd@0:         % Evaluate the single historical entries 
wolffd@0:         % and sum up
wolffd@0:         % ---
wolffd@0:         thisw = sum(G.hE{V1, V2},1);
wolffd@0:         
wolffd@0:         % ---
wolffd@0:         % now divide the single votes by the number of
wolffd@0:         % votes totally made with the option to improve 
wolffd@0:         % this edge
wolffd@0:         % ---
wolffd@0:         thisw = thisw(1) /  thisw(2); 
wolffd@0:         
wolffd@0:     end
wolffd@0: end
wolffd@0: end