从词性标签中提取名词词和原句

Question

我想从句子中提取名词，从词性标签中取回原句

 //Extract the words before _NNP & _NN from below  and also how to get back the original sentence from the Pos TAG. 
 Original Sentence:Hi. How are you? This is Mike·
 POSTag: Hi._NNP How_WRB are_VBP you?_JJ This_DT is_VBZ Mike._NN

我试过这样的

    String txt = "Hi._NNP How_WRB are_VBP you?_JJ This_DT is_VBZ Mike._NN";


    String re1 = "((?:[a-z][a-z0-9_]*))";   // Variable Name 1
    String re2 = ".*?"; // Non-greedy match on filler
    String re3 = "(_)"; // Any Single Character 1
    String re4 = "(NNP)";   // Word 1

    Pattern p = Pattern.compile(re1 + re2 + re3 + re4, Pattern.CASE_INSENSITIVE | Pattern.DOTALL);
    Matcher m = p.matcher(txt);
    if (m.find()) {
        String var1 = m.group(1);
        System.out.print(  var1.toString()  );
    }
}

输出：嗨 但我需要一个句子中所有名词的列表。

Answer 1

要提取名词，您可以这样做：

public static String[] extractNouns(String sentenceWithTags) {
    // Split String into array of Strings whenever there is a tag that starts with "._NN"
    // followed by zero, one or two more letters (like "_NNP", "_NNPS", or "_NNS")
    String[] nouns = sentenceWithTags.split("_NN\\w?\\w?\\b");
    // remove all but last word (which is the noun) in every String in the array
    for(int index = 0; index < nouns.length; index++) {
        nouns[index] = nouns[index].substring(nouns[index].lastIndexOf(" ") + 1)
        // Remove all non-word characters from extracted Nouns
        .replaceAll("[^\\p{L}\\p{Nd}]", "");
    }
    return nouns;
}

要提取原句，可以这样做：

public static String extractOriginal(String sentenceWithTags) {
    return sentenceWithTags.replaceAll("_([A-Z]*)\\b", "");
}

证明它有效：

public static void main(String[] args) {
    String sentence = "Hi._NNP How_WRB are_VBP you?_JJ This_DT is_VBZ Mike._NN";
    System.out.println(java.util.Arrays.toString(extractNouns(sentence)));
    System.out.println(extractOriginal(sentence));
}

输出：

[Hi, Mike]
Hi. How are you? This is Mike.

注意：对于从提取的名词中删除所有非单词字符（如标点符号）的正则表达式，我使用了 this Stack Overflow question/answer。

Answer 2

使用while (m.find()) 而不是if (m.find()) 来遍历所有匹配项。

此外，您的正则表达式可以真正简化：

• 如果您不需要捕获数据，请不要加括号（通常）
• 您使用的是((?:...))，这很奇怪：直接嵌套在捕获组中的非捕获组没有意义。
• 我不确定.*? 部分是否符合您的预期。如果要匹配点，请改用[.]。

因此，请改用([a-z][a-z0-9_]*)[.]_NNP。

或者甚至使用正向前瞻：[a-z][a-z0-9_]*(?=[.]_NNP)。使用m.group() 访问捕获的数据。

Answer 3

这个应该可以的

import java.util.ArrayList;
public class Test {

public static final String NOUN_REGEX = "[a-zA-Z]*_NN\\w?\\w?\\b";

public static ArrayList<String> extractNounsByRegex(String sentenceWithTags) {
    ArrayList<String> nouns = new ArrayList<String>();
    String[] words = sentenceWithTags.split("\\s+");
    for (int i = 0; i < words.length; i++) {
        if(words[i].matches(NOUN_REGEX)) {
            System.out.println(" Matched ");
            //remove the suffix _NN* and retain  [a-zA-Z]*
                nouns.add(words[i].replaceAll("_NN\\w?\\w?\\b", ""));
            }
        }
        return nouns;
    }

    public static String extractOriginal(String word) {
                return word.replaceAll("_NN\\w?\\w?\\b", "");
    }

    public static void main(String[] args) {
        //        String sentence = "Hi._NNP How_WRB are_VBP you?_JJ This_DT is_VBZ Mike._NN";
        String sentence = "Eiffel_NNP tower_NN is_VBZ in_IN paris_NN Hi_NNP How_WRB are_VBP you_PRP This_DT is_VBZ Mike_NNP Barrack_NNP Obama_NNP is_VBZ a_DT president_NN this_VBZ";
        System.out.println(extractNounsByRegex(sentence).toString());
        System.out.println(sentence);
    }
}