1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
|
%wsj_lengths = load([ 'wsj_lengths.dat']);
%save([ 'wsj_lengths.mat'],'wsj_lengths');
load wsj
load wsj_lengths
figure(1)
clf
subplot(1,3,1);
hold on
for i = 2:6
b = 10^(i-1)
disp(['Loading results for b = ' num2str(b) ]);
%%% uncomment these lines if .mat file is not yet generated. %%%
%typecountrecord= load([ 'outputs/typecountrecordwsjflat0.0.' num2str(b) '.0.dat']);
%typecountrecordmean = mean(typecountrecord(500:999,:));
%save([ 'outputs/typecountrecordmeanwsjflat0.0.' num2str(b) '.0.mat'],'typecountrecordmean');
load([ 'outputs/typecountrecordmeanwsjflat0.0.' num2str(b) '.0.mat']);
% plot lines for CRP exact prediction using summation
[logbins predicted dummy] = logbinmean(counts, crppred(counts,b),20,20);
ph = plot(log10(logbins),log10(predicted),'r');
set(ph,'color',[0.7 0.7 0.7],'linewidth',1.5)
% plot lines for CRP Antoniak prediction
[logbins predicted dummy] = logbinmean(counts, antoniakpred(counts,b),20,20);
ph = plot(log10(logbins),log10(predicted),'r');
set(ph,'color',[0.7 0.7 0.7],'linewidth',1.5,'linestyle','--')
% plot lines for incorrect CRP Antoniak prediction (ACL07)
%[logbins predicted dummy] = logbinmean(counts, noP0pred(counts,b),20,20);
%ph = plot(log10(logbins),log10(predicted),'r');
%set(ph,'color',[0.7 0.7 0.7],'linewidth',1.5,'linestyle','-.')
% plot lines for CRP Cohn prediction
%[logbins predicted dummy] = logbinmean(counts, cohnpred(counts,b),20,20);
%ph = plot(log10(logbins),log10(predicted),'r');
%set(ph,'color',[0.2 0.2 1],'linewidth',1.5,'linestyle','.')
%plot emprical counts with error bars
[logbins meanval seval] = logbinmean(counts,typecountrecordmean,20,20);
errorbar(log10(logbins),log10(meanval),log10(meanval+seval)-log10(meanval),log10(meanval-seval)-log10(meanval),'k.');
end
set(gca,'xtick',log10([1:10 20:10:100 200:100:1000 2000:1000:5000]))
set(gca,'ytick',log10([.1:.1:1 2:10 20:10:100 200:100:1000 2000:1000:5000]))
set(gca,'xlim',[-0.1 3.5])
set(gca,'ylim',[-1.1 1.5])
set(gca,'FontSize',14)
set(gca,'xticklabel', {'1',' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ...
'10',' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', '100', ...
' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', '1000', ...
' ', ' ', ' ', ' '});
set(gca,'yticklabel', {'0.1',' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ...
'1',' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ...
'10',' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', '100', ...
' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', '1000', ...
' ', ' ', ' ', ' '});
%title('Chinese restaurant process adaptor')
ylabel('Mean number of lexical entries')
xlabel('Word frequency (n_w)')
legend('Expectation','Antoniak approx.','Empirical','Location','NorthWest')
box on
subplot(1,3,2);
hold on
for i =2:6
b = 10^(i-1)
disp(['Loading results for b = ' num2str(b) ]);
%%% uncomment these lines if .mat file is not yet generated. %%%
%typecountrecord= load([ 'outputs/typecountrecordwsjpeak0.0.' num2str(b) '.0.dat']);
%typecountrecordmean = mean(typecountrecord(500:999,:));
%save([ 'outputs/typecountrecordmeanwsjpeak0.0.' num2str(b) '.0.mat'],'typecountrecordmean');
load([ 'outputs/typecountrecordmeanwsjpeak0.0.' num2str(b) '.0.mat']);
% plot lines for CRP exact prediction using summation
[logbins predicted dummy] = logbinmean(counts, crppred(counts,b),20,20);
ph = plot(log10(logbins),log10(predicted),'r');
set(ph,'color',[0.7 0.7 0.7],'linewidth',1.5)
%plot emprical counts with error bars
[logbins meanval seval] = logbinmean(counts,typecountrecordmean,20,20);
errorbar(log10(logbins),log10(meanval),log10(meanval+seval)-log10(meanval),log10(meanval-seval)-log10(meanval),'k.');
end
set(gca,'xtick',log10([1:10 20:10:100 200:100:1000 2000:1000:5000]))
set(gca,'ytick',log10([1:10 20:10:100 200:100:1000 2000:1000:5000]))
set(gca,'xlim',[-0.1 3.5])
set(gca,'ylim',[-.1 2.5])
set(gca,'FontSize',14)
set(gca,'xticklabel', {'1',' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ...
'10',' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', '100', ...
' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', '1000', ...
' ', ' ', ' ', ' '});
set(gca,'yticklabel', {...%'0.1',' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ...
'1',' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ...
'10',' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', '100', ...
' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', '1000', ...
' ', ' ', ' ', ' '});
%title('Chinese restaurant process adaptor')
ylabel('Mean number of lexical entries')
xlabel('Word frequency (n_w)')
legend('Expectation','Location','NorthWest')
box on
%axis square
subplot(1,3,3);
hold on
for i =2:6
b = 10^(i-1)
disp(['Loading results for b = ' num2str(b) ]);
%%% uncomment these lines if .mat file is not yet generated. %%%
%typecountrecord= load([ 'outputs/typecountrecordwsjgeom0.0.' num2str(b) '.0.dat']);
%typecountrecordmean = mean(typecountrecord(500:999,:));
%save([ 'outputs/typecountrecordmeanwsjgeom0.0.' num2str(b) '.0.mat'],'typecountrecordmean');
load([ 'outputs/typecountrecordmeanwsjgeom0.0.' num2str(b) '.0.mat']);
% plot lines for CRP exact prediction using summation
% [logbins meaneval seval] = logbinmean(counts, crppred_geom(counts,wsj_lengths,b),20,20)
[logbins meaneval seval] = logbinmean(counts, crppred(counts,b),20,20)
plot(log10(logbins),log10(meaneval),'r.');
%errorbar(log10(logbins),log10(meanval),log10(meanval+seval)-log10(meanval),log10(meanval-seval)-log10(meanval),'r.');
% ph = plot(log10(logbins),log10(meaneval),'r');
% set(ph,'color',[0.7 0.7 0.7],'linewidth',1.5)
%plot emprical counts with error bars
[logbins meanval seval] = logbinmean(counts,typecountrecordmean,20,20);
errorbar(log10(logbins),log10(meanval),log10(meanval+seval)-log10(meanval),log10(meanval-seval)-log10(meanval),'k.');
end
set(gca,'xtick',log10([1:10 20:10:100 200:100:1000 2000:1000:5000]))
set(gca,'ytick',log10([1:10 20:10:100 200:100:1000 2000:1000:5000]))
set(gca,'xlim',[-0.1 3.5])
set(gca,'ylim',[-.1 2.5])
set(gca,'FontSize',14)
set(gca,'xticklabel', {'1',' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ...
'10',' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', '100', ...
' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', '1000', ...
' ', ' ', ' ', ' '});
set(gca,'yticklabel', {...%'0.1',' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ...
'1',' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ...
'10',' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', '100', ...
' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', '1000', ...
' ', ' ', ' ', ' '});
%title('Chinese restaurant process adaptor')
ylabel('Mean number of lexical entries')
xlabel('Word frequency (n_w)')
legend('Expectation','Location','NorthWest')
box on
hold off
%axis square
|
