found a bug in w-formula, 1/P_s should be outside sums, starting to think we should...
[dmvccm.git] / DMVCCM.html.~1~
blob8405caa82b976f155fa5fad8ed98945f85e8b7f4
1 <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN"
2                "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
3 <html xmlns="http://www.w3.org/1999/xhtml"
4 lang="nn" xml:lang="nn">
5 <head>
6 <title>DMV/CCM</title>
7 <meta http-equiv="Content-Type" content="text/html;charset=utf-8"/>
8 <meta name="generator" content="Org-mode"/>
9 <meta name="generated" content="2008/05/25 17:09:24"/>
10 <meta name="author" content="Kevin Brubeck Unhammer"/>
11 <link rel="stylesheet" type="text/css" href="org.css">
12 </head><body>
13 <h1 class="title">DMV/CCM</h1>
14 <div id="table-of-contents">
15 <h2>Innhold</h2>
16 <ul>
17 <li><a href="#sec-1">1 dmvccm</a>
18 <ul>
19 <li><a href="#sec-2">1.1 [#A] DMV-probabilities   </a></li>
20 <li><a href="#sec-3">1.2 Initialization   </a></li>
21 <li><a href="#sec-4">1.3 Meet Yoav again about dmvccm</a>
22 <ul>
23 <li><a href="#sec-5">1.3.1 Initialization</a></li>
24 <li><a href="#sec-6">1.3.2 Corpus access?</a></li>
25 <li><a href="#sec-7">1.3.3 How do we interpret DMV as an inside/outside process?</a></li>
26 <li><a href="#sec-8">1.3.4 How do we know whether we are 'adjacent' or not? </a></li>
27 <li><a href="#sec-9">1.3.5 What are the formulas for P<sub>CHOOSE</sub> etc?</a></li>
28 </ul></li>
29 </ul>
30 </li>
31 <li><a href="#sec-10">2 Python-stuff</a></li>
32 </ul>
33 </div>
35 <div class="outline-2">
36 <h2 id="sec-1">1 dmvccm</h2>
38 <p><span class="timestamp-kwd">DEADLINE: </span> <span class="timestamp">2008-06-30 Mon</span><br/>
39 </p>
40 <p>
41 (But absolute, extended, really-quite-dead-now deadline: August 31&hellip;)
42 <a href="dmvccm/src/dmv.py">dmv.py</a>
43 <a href="dmvccm/src/io.py">io.py</a>
44 </p>
45 <div class="outline-3">
46 <h3 id="sec-2">1.1 <span class="todo">TODO</span> [#A] DMV-probabilities   </h3>
48 <p><a href="/Home/stud6/kun041/Documents/Skole/V08/Probability/dmvccm/src/dmv.py">dmv.py</a>
49 </p></div>
51 <div class="outline-3">
52 <h3 id="sec-3">1.2 <span class="todo">TOGROK</span> Initialization   </h3>
56 </div>
58 <div class="outline-3">
59 <h3 id="sec-4">1.3 Meet Yoav again about dmvccm</h3>
61 <p><span class="timestamp-kwd">SCHEDULED: </span> <span class="timestamp">2008-05-26 Mon</span><br/>
62 13:30, P3.21.
63 </p>
64 <p>
65 Questions:
66 </p>
67 <div class="outline-4">
68 <h4 id="sec-5">1.3.1 Initialization</h4>
70 </div>
72 <div class="outline-4">
73 <h4 id="sec-6">1.3.2 Corpus access?</h4>
75 </div>
77 <div class="outline-4">
78 <h4 id="sec-7">1.3.3 How do we interpret DMV as an inside/outside process?</h4>
80 <p>The upside-down P<sub>STOP</sub> formula (left-to-right also)
81 c<sub>s</sub>(x : i, j) is "the expected fraction of parses of s" with x from
82 i to j; expectation then uses the probabilities gotten from
83 initialization and previously gained probabilities, but these are of
84 the form P<sub>STOP</sub> and P<sub>CHOOSE</sub>, how do we translate this to inside
85 outside, which just uses the probabilities of CFG-rules?
86 </p></div>
88 <div class="outline-4">
89 <h4 id="sec-8">1.3.4 How do we know whether we are 'adjacent' or not? </h4>
91 <p>Can we even know that without the full tree?
92 </p>
93 <p>
94 if we have 
95 \Tree [<sub>b</sub> [<sub>b</sub> b <u>c</u> ] <u>d</u> ] 
96 then the lower tree [<sub>b</sub> b <u>c</u> ] is adjacent since, working your way up
97 the tree, no argument has been created to the right "yet"; while the
98 outer tree [<sub>b</sub> [<sub>b</sub> &hellip; ] <u>d</u> ] is non-adjacent, since there is something in
99 between&hellip; Is it thus always adjacent to the right if the the distance
100 is 1?
101 </p></div>
103 <div class="outline-4">
104 <h4 id="sec-9">1.3.5 What are the formulas for P<sub>CHOOSE</sub> etc?</h4>
106 <p>Is this the regular "outer" summation?
107 </p>
112 </div>
113 </div>
114 </div>
116 <div class="outline-2">
117 <h2 id="sec-10">2 Python-stuff</h2>
119 <p><a href="dmvccm/src/pseudo.py">pseudo.py</a>
120 </p>
122 <a href="http://nltk.org/doc/en/structured-programming.html">http://nltk.org/doc/en/structured-programming.html</a> recursive dynamic
123 <a href="http://nltk.org/doc/en/advanced-parsing.html">http://nltk.org/doc/en/advanced-parsing.html</a> 
124 </p>
126 </div>
127 <div id="postamble"><p class="author"> Forfattar: Kevin Brubeck Unhammer
128 <a href="mailto:K.BrubeckUnhammer at student uva nl ">&lt;K.BrubeckUnhammer at student uva nl &gt;</a>
129 </p>
130 <p class="date"> Dato: 2008/05/25 17:09:24</p>
131 </div><p class="postamble">Skrive vha. emacs + <a href='http://orgmode.org/'>org-mode</a></p></body>
132 </html>