Mercurial > hg > duomoOWLProject
annotate src/de/mpiwg/dwinter/duomo/stanford/Analyse.java @ 8:919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
alle has_readable_labe Datatype properties durch rdfs:label ersetzt.
author | dwinter |
---|---|
date | Thu, 21 Jun 2012 17:08:22 +0200 |
parents | |
children |
rev | line source |
---|---|
8
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
1 package de.mpiwg.dwinter.duomo.stanford; |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
2 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
3 import java.io.BufferedReader; |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
4 import java.io.DataInputStream; |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
5 import java.io.FileInputStream; |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
6 import java.io.FileWriter; |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
7 import java.io.IOException; |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
8 import java.io.InputStreamReader; |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
9 import java.io.Reader; |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
10 import java.util.Collection; |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
11 import java.util.HashMap; |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
12 import java.util.HashSet; |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
13 import java.util.List; |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
14 import java.util.Map; |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
15 import java.util.Set; |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
16 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
17 import edu.stanford.nlp.io.EncodingPrintWriter.out; |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
18 import edu.stanford.nlp.ling.CyclicCoreLabel; |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
19 import edu.stanford.nlp.ling.DocumentReader; |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
20 import edu.stanford.nlp.ling.HasWord; |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
21 import edu.stanford.nlp.ling.Word; |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
22 import edu.stanford.nlp.parser.lexparser.LexicalizedParser; |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
23 import edu.stanford.nlp.process.DocumentPreprocessor; |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
24 import edu.stanford.nlp.trees.GrammaticalRelation; |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
25 import edu.stanford.nlp.trees.GrammaticalStructure; |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
26 import edu.stanford.nlp.trees.GrammaticalStructureFactory; |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
27 import edu.stanford.nlp.trees.PennTreebankLanguagePack; |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
28 import edu.stanford.nlp.trees.Tree; |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
29 import edu.stanford.nlp.trees.TreebankLanguagePack; |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
30 import edu.stanford.nlp.trees.TypedDependency; |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
31 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
32 public class Analyse { |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
33 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
34 public void analyse(String filename) throws IOException { |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
35 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
36 LexicalizedParser lp = LexicalizedParser.loadModel("edu/stanford/nlp/models/lexparser/englishPCFG.ser.gz"); |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
37 // This option shows loading and sentence-segment and tokenizing |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
38 // a file using DocumentPreprocessor |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
39 TreebankLanguagePack tlp = new PennTreebankLanguagePack(); |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
40 GrammaticalStructureFactory gsf = tlp.grammaticalStructureFactory(); |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
41 // You could also create a tokenier here (as below) and pass it |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
42 // to DocumentPreprocessor |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
43 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
44 int count=0; |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
45 Map<String,Integer> tuple = new HashMap<String,Integer>(); |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
46 Map<String,Integer> tupleLong = new HashMap<String,Integer>(); |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
47 Map<String,Integer> words = new HashMap<String,Integer>(); |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
48 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
49 FileInputStream fstream = new FileInputStream(filename); |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
50 // Get the object of DataInputStream |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
51 DataInputStream in = new DataInputStream(fstream); |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
52 BufferedReader br = new BufferedReader(new InputStreamReader(in)); |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
53 String strLine; |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
54 //Read File Line By Line |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
55 while ((strLine = br.readLine()) != null) { |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
56 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
57 // correct line needs to be completed to a sentence |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
58 strLine=strLine.replace("\"", ""); |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
59 strLine="This is a "+strLine; |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
60 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
61 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
62 Reader dr = DocumentReader.getReader(strLine); |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
63 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
64 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
65 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
66 for (List<HasWord> sentence : new DocumentPreprocessor(dr)) { |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
67 Tree parse = lp.apply(sentence); |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
68 //parse.pennPrint(); |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
69 //System.out.println(); |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
70 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
71 for (HasWord word: sentence) |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
72 { |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
73 Word wd = (Word)word; |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
74 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
75 String st= wd.value().toLowerCase(); |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
76 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
77 if (words.containsKey(st)){ |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
78 words.put(st, words.get(st)+1); |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
79 } else { |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
80 words.put(st, 1); |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
81 } |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
82 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
83 } |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
84 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
85 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
86 GrammaticalStructure gs = gsf.newGrammaticalStructure(parse); |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
87 Collection tdl = gs.typedDependenciesCCprocessed(true); |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
88 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
89 for (Object t: tdl){ |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
90 if (TypedDependency.class.isInstance(t)){ |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
91 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
92 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
93 TypedDependency td = (TypedDependency)t; |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
94 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
95 GrammaticalRelation reln = td.reln(); |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
96 if (reln.getShortName().equals("prep") || reln.getShortName().equals("conj") ){ |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
97 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
98 String st = reln.getShortName() |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
99 +"\t"; |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
100 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
101 st +=td.gov().label().value()+"\t"; |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
102 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
103 st+=td.dep().label().value(); |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
104 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
105 st=st.toLowerCase(); |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
106 if (tuple.containsKey(st)){ |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
107 tuple.put(st, tuple.get(st)+1); |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
108 } else { |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
109 tuple.put(st, 1); |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
110 } |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
111 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
112 st = reln.getShortName()+"\t"+reln.getSpecific()+"\t"; |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
113 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
114 st +=td.gov().label().value()+"\t"; |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
115 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
116 st+=td.dep().label().value(); |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
117 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
118 st=st.toLowerCase(); |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
119 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
120 if (tupleLong.containsKey(st)){ |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
121 tupleLong.put(st, tupleLong.get(st)+1); |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
122 } else { |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
123 tupleLong.put(st, 1); |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
124 } |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
125 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
126 } |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
127 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
128 } |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
129 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
130 } |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
131 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
132 //System.out.println(tdl); |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
133 //System.out.println(); |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
134 count++; |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
135 System.out.println(count); |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
136 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
137 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
138 } |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
139 //if (count > 5) |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
140 // break; |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
141 } |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
142 System.out.println(tuple); |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
143 System.out.println(tupleLong); |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
144 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
145 FileWriter fw = new FileWriter("/tmp/tuple"); |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
146 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
147 for (String key : tuple.keySet()){ |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
148 fw.write(key+"\t"+String.valueOf(tuple.get(key))+"\n"); |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
149 } |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
150 fw.close(); |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
151 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
152 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
153 fw = new FileWriter("/tmp/tupleLong"); |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
154 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
155 for (String key : tupleLong.keySet()){ |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
156 fw.write(key+"\t"+String.valueOf(tupleLong.get(key))+"\n"); |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
157 } |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
158 fw.close(); |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
159 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
160 fw = new FileWriter("/tmp/words"); |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
161 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
162 for (String key : words.keySet()){ |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
163 fw.write(key+"\t"+String.valueOf(words.get(key))+"\n"); |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
164 } |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
165 fw.close(); |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
166 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
167 } |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
168 /** |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
169 * @param args |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
170 */ |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
171 public static void main(String[] args) { |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
172 Analyse a = new Analyse(); |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
173 try { |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
174 a.analyse("/tmp/reges.csv"); |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
175 } catch (IOException e) { |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
176 // TODO Auto-generated catch block |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
177 e.printStackTrace(); |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
178 } |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
179 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
180 } |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
181 |
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
diff
changeset
|
182 } |