Merge remote-tracking branch 'origin/pubmed' into wikipedia_AND

Conflicts:
	GetDistances.m
	NWD.m
	results.txt

.gitignore

@@ -28,3 +28,4 @@ obj/
 _ReSharper*/
 [Tt]est[Rr]esult*
 *.asv
+~$*

GetCount.m

@@ -6,7 +6,7 @@ if isempty(strServer)
     strServer='wikipedia'
 end
 
-term=[words{1}];
+term=[ words{1} ];
 for i=2:length(words)
     term=[ term '+' words{i}];
 end
@@ -85,28 +85,15 @@ for i=1:5
     str = urlread(URL);
     try
         %amazon
-        idx=strfind(str,'results for <span class="');
+        idx=strfind(str,'results for <span>');
         idx=idx(1)-2;
         idxStart=idx;
-        while(str(idxStart)~=' ')
+        chPrev = str(idxStart-1);
+        while(isNumeric(chPrev))
             idxStart=idxStart-1;
+            chPrev = str(idxStart-1);
         end
         nx=str(idxStart:idx);
-        
-        
-        % WIKIPEDIA%
-        %     idx=strfind(str,'div class="results-info"');
-        %     idx=idx(1);
-        %     str2=str(idx:end);
-        %
-        %     idx2=strfind(str2,'of <b>');
-        %     str3=str2(idx2+6:end);
-        %
-        %     idx3=strfind(str3,'</b>');
-        %     idx3=idx3(1);
-        %
-        %     nx=str3(1:idx3-1);
-        % END WIKIPEDIA
         f=str2double(nx);
         break;
     catch
@@ -117,4 +104,18 @@ end
 
 end
 
+function bNumber = isNumeric(ch)
+
+bNumber=1;
+if ~isnan(str2double(ch))
+    return
+end
+
+if ch==','
+    return
+end
+
+bNumber=0;
+end
+
 

GetDistances.m

@@ -54,10 +54,10 @@ for i=1:length(words)
 %     NWD2(i)=NWD(w2);
 
     d1(i)=(NWD1x(i)-NWD1(i));
-    d2(i)=(NWD2x(i)-NWD2(i));
-    if isnan(d1(i)) % f(X)=0 --> -log2(0)==Inf , Inf-Inf=NaN
-        d1(i)=Inf;
+    if isnan(d1(i))
+        d1(i)=inf;
     end
+    d2(i)=(NWD2x(i)-NWD2(i));
     if isnan(d2(i)) % f(X)=0 --> -log2(0)==Inf 
         d2(i)=Inf;
     end

NWD.m

@@ -26,5 +26,5 @@ else
     nwd = (log2(max(fw))-log2(fX)) / (log2(N)-log2(min(fw))); % III.3 from arxiv
 end
 % normalize for cardinality
-nwd = nwd/(length(words)-1);
-
+ nwd = nwd/(length(words)-1);
+% nwd=(log2(max(fw))-log2(fX))/( log2(N)-log2(max(fexclude)) );

getDistancesPairs.m

@@ -19,6 +19,7 @@ for i=1:length(words)
 end
 
 d(length(words),length(words))=0;
+d(find(isinf(d)))=10;
 d=Regularize(d);
 [kGap Gap S idx] = GapSpectral(d,6,1);
 fprintf('kGap=%d\n',kGap);

goAmazon.m

+
+tic
+% goNWD
+%
+%
+global strServer
+strServer='amazon'
+global QueryCache
+if isempty(QueryCache)
+QueryCache.Queries={};
+QueryCache.Count=[];
+end
+global nQueryCount nCacheCount
+nQueryCount=0;
+nCacheCount=0;
+
+words1={'Macbeth','The Tempest','Othello','King Lear','Hamlet'...
+ 'The Merchant of Venice','A Midsummer Nights Dream',...
+ 'Much Ado About Nothing', 'Taming of the Shrew','Twelfth Night' }
+ 
+words2={'Carrie','Salems Lot','The Shining','The Stand','The Dead Zone',...
+    'Firestarter','Cujo'}
+
+GetCount(words1)
+GetCount(words2)
+    
+% 
+[d1,d2]=GetDistances(words1,words2);
+idx=getDistancesPairs(words1,words2);

goPubMed.m

+global strServer
+
+strServer='pubmed'
 
 %http://www.nature.com/nature/journal/v511/n7510/full/nature13595.html
 % supp table 2, in ranked order
@@ -37,7 +40,11 @@ neuroblastoma = {'rs6939340','rs4712653','rs9295536','rs3790171','rs7272481'};
 % parkinsons={'rs356219','rs10847864','rs1491942','rs947211','rs2390669'}
 parkinsons={'rs356219','rs10847864','rs2942168','rs11724635'}
 
-words={ alzheimers;parkinsons;shizophrenia;leukemia;obesity;neuroblastoma}
+% http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3839234/
+% http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3839234/table/T1/
+als = {'rs2303565','rs1344642','rs2814707','rs3849942','rs2453556', 'rs1971791', 'rs8056742'};
+
+words={ alzheimers;parkinsons;als;shizophrenia;leukemia;obesity;neuroblastoma}
 
 global QueryCache
 if isempty(QueryCache)

results.txt

+<<<<<<< HEAD
 
 
 -------------------------
 GetDistances::(multiples) NW=5 
 2014              9              5             22             18         10.438
 red,orange,yellow,green,blue,indigo,lion,tiger,bear,monkey,zebra,elephant,aardvark,lamb,fox,ape,dog,nQueryCount=0, nCacheCount=1292
+=======
+% no normalization! 
+
+-------------------------
+GetDistances::(multiples) NW=5 
+2014              8             13             12             18         14.692
+red,orange,yellow,green,blue,indigo,lion,tiger,bear,monkey,zebra,elephant,aardvark,lamb,fox,ape,dog,nQueryCount=394, nCacheCount=898
+>>>>>>> origin/pubmed
 
 [6 0;0 11]
 
 
 -------------------------
+<<<<<<< HEAD
 GetDistances::(multiples) NW=5 
 2014              9              5             22             21          31.57
 red,orange,yellow,green,blue,indigo,violet,purple,cyan,white,square,circle,rectangle,ellipse,triangle,rhombus,nQueryCount=320, nCacheCount=832
+=======
+GetDistances::(pairs) 
+2014              8             13             12             19         27.056
+red,orange,yellow,green,blue,indigo,lion,tiger,bear,monkey,zebra,elephant,aardvark,lamb,fox,ape,dog,kGap=2
+nQueryCount=136, nCacheCount=680
+nCorrect=17
+
+
+-------------------------
+GetDistances::(multiples) NW=5 
+2014              8             13             12             22         38.066
+red,orange,yellow,green,blue,indigo,violet,purple,cyan,white,square,circle,rectangle,ellipse,triangle,rhombus,nQueryCount=342, nCacheCount=810
+>>>>>>> origin/pubmed
 
 [10 0;0 6]
 
 
 -------------------------
+<<<<<<< HEAD
 GetDistances::(multiples) NW=5 
 2014              9              5             22             23         50.569
+=======
+GetDistances::(pairs) 
+2014              8             13             12             23         32.038
+red,orange,yellow,green,blue,indigo,violet,purple,cyan,white,square,circle,rectangle,ellipse,triangle,rhombus,kGap=1
+nQueryCount=105, nCacheCount=615
+nCorrect=16
+
+
+-------------------------
+GetDistances::(multiples) NW=5 
+2014              8             13             12             25         37.418
+>>>>>>> origin/pubmed
 Barack Obama,Hillary Clinton,John Edwards,Joe Biden,Chris Dodd,Mike Gravel,John McCain,Mitt Romney,Mike Huckabee,Ron Paul,Fred Thompson,Alan Keyes,nQueryCount=198, nCacheCount=474
 
 [6 0;0 6]
@@ -26,23 +62,41 @@ Barack Obama,Hillary Clinton,John Edwards,Joe Biden,Chris Dodd,Mike Gravel,John
 
 -------------------------
 GetDistances::(pairs) 
+<<<<<<< HEAD
 2014              9              5             22             24         38.426
+=======
+2014              8             13             12             26         19.956
+>>>>>>> origin/pubmed
 Barack Obama,Hillary Clinton,John Edwards,Joe Biden,Chris Dodd,Mike Gravel,John McCain,Mitt Romney,Mike Huckabee,Ron Paul,Fred Thompson,Alan Keyes,kGap=2
 nQueryCount=66, nCacheCount=330
 nCorrect=7
 
+<<<<<<< HEAD
 
 -------------------------
 GetDistances::(multiples) NW=5 
 2014              9              5             22             29         12.913
 red,orange,yellow,green,blue,indigo,lion,tiger,bear,monkey,zebra,elephant,aardvark,lamb,fox,ape,dog,nQueryCount=0, nCacheCount=1292
+=======
+% normalization!
+
+
+-------------------------
+GetDistances::(multiples) NW=5 
+2014              8             13             12             37         22.423
+red,orange,yellow,green,blue,indigo,lion,tiger,bear,monkey,zebra,elephant,aardvark,lamb,fox,ape,dog,nQueryCount=394, nCacheCount=898
+>>>>>>> origin/pubmed
 
 [6 0;0 11]
 
 
 -------------------------
 GetDistances::(pairs) 
+<<<<<<< HEAD
 2014              9              5             22             30         36.213
+=======
+2014              8             13             12             38         38.851
+>>>>>>> origin/pubmed
 red,orange,yellow,green,blue,indigo,lion,tiger,bear,monkey,zebra,elephant,aardvark,lamb,fox,ape,dog,kGap=2
 nQueryCount=136, nCacheCount=680
 nCorrect=17
@@ -50,15 +104,24 @@ nCorrect=17
 
 -------------------------
 GetDistances::(multiples) NW=5 
+<<<<<<< HEAD
 2014              9              5             22             30         36.384
 red,orange,yellow,green,blue,indigo,violet,purple,cyan,white,square,circle,rectangle,ellipse,triangle,rhombus,nQueryCount=0, nCacheCount=1152
+=======
+2014              8             13             12             41         55.419
+red,orange,yellow,green,blue,indigo,violet,purple,cyan,white,square,circle,rectangle,ellipse,triangle,rhombus,nQueryCount=342, nCacheCount=810
+>>>>>>> origin/pubmed
 
 [10 0;0 6]
 
 
 -------------------------
 GetDistances::(pairs) 
+<<<<<<< HEAD
 2014              9              5             22             31         38.331
+=======
+2014              8             13             12             42         49.628
+>>>>>>> origin/pubmed
 red,orange,yellow,green,blue,indigo,violet,purple,cyan,white,square,circle,rectangle,ellipse,triangle,rhombus,kGap=1
 nQueryCount=105, nCacheCount=615
 nCorrect=16
@@ -66,15 +129,56 @@ nCorrect=16
 
 -------------------------
 GetDistances::(multiples) NW=5 
+<<<<<<< HEAD
 2014              9              5             22             31         38.411
 Barack Obama,Hillary Clinton,John Edwards,Joe Biden,Chris Dodd,Mike Gravel,John McCain,Mitt Romney,Mike Huckabee,Ron Paul,Fred Thompson,Alan Keyes,nQueryCount=0, nCacheCount=672
+=======
+2014              8             13             12             44          55.56
+Barack Obama,Hillary Clinton,John Edwards,Joe Biden,Chris Dodd,Mike Gravel,John McCain,Mitt Romney,Mike Huckabee,Ron Paul,Fred Thompson,Alan Keyes,nQueryCount=198, nCacheCount=474
+>>>>>>> origin/pubmed
 
 [6 0;0 6]
 
 
 -------------------------
 GetDistances::(pairs) 
+<<<<<<< HEAD
 2014              9              5             22             31         40.923
+=======
+2014              8             13             12             45         38.156
+>>>>>>> origin/pubmed
 Barack Obama,Hillary Clinton,John Edwards,Joe Biden,Chris Dodd,Mike Gravel,John McCain,Mitt Romney,Mike Huckabee,Ron Paul,Fred Thompson,Alan Keyes,kGap=2
 nQueryCount=0, nCacheCount=396
 nCorrect=7
+
+
+-------------------------
+GetDistances::(multiples) NW=5 
+2014              9             19             10             33         53.142
+Alls Well That Ends Well,As You Like It,The Comedy of Errors,Measure for Measure,The Merchant of Venice,A Midsummer Nights Dream,Much Ado About Nothing,Taming of the Shrew,Twelfth Night,Carrie,Salems Lot,The Shining,The Stand,The Dead Zone,Firestarter,Cujo,nQueryCount=344, nCacheCount=808
+
+[8 1;0 7]
+
+
+-------------------------
+GetDistances::(multiples) NW=5 
+2014              9             19             14             25         25.227
+Alls Well That Ends Well,As You Like It,The Comedy of Errors,Measure for Measure,The Merchant of Venice,A Midsummer Nights Dream,Much Ado About Nothing,Taming of the Shrew,Twelfth Night,Carrie,Salems Lot,The Shining,The Stand,The Dead Zone,Firestarter,Cujo,nQueryCount=153, nCacheCount=999
+
+[8 1;0 7]
+
+
+-------------------------
+GetDistances::(multiples) NW=5 
+2014              9             19             16             13         15.418
+Macbeth,The Tempest,Othello,King Lear,Hamlet,The Merchant of Venice,A Midsummer Nights Dream,Much Ado About Nothing,Taming of the Shrew,Twelfth Night,Carrie,Salems Lot,The Shining,The Stand,The Dead Zone,Firestarter,Cujo,nQueryCount=37, nCacheCount=1255
+
+[10 0;0 7]
+
+
+-------------------------
+GetDistances::(pairs) 
+2014              9             19             16             21         31.009
+Macbeth,The Tempest,Othello,King Lear,Hamlet,The Merchant of Venice,A Midsummer Nights Dream,Much Ado About Nothing,Taming of the Shrew,Twelfth Night,Carrie,Salems Lot,The Shining,The Stand,The Dead Zone,Firestarter,Cujo,kGap=2
+nQueryCount=136, nCacheCount=680
+nCorrect=17