cmake ..

苯并芘
亚硝酸盐

雷人 v a
】 wp

g++ -o cws cws.cpp -I /path/to/your/ltp-project/include/ -I /path/to/your/ltp-project/thirdparty/boost/include -WL,-dn -L /path/to/your/ltp-project/lib/ -lsegmentor -lboost_regex -WL,-dy

#include <iostream>
#include <string>
#include "segment_dll.h"

int main(int argc, char * argv[]) {
    if (argc < 2) {
        std::cerr << "cws [model path]" << std::endl;
        return 1;
    }

    void * engine = segmentor_create_segmentor(argv[1]);
    if (!engine) {
        return -1;
    }
    std::vector<std::string> words;
    int len = segmentor_segment(engine,
            "爱上一匹野马，可我的家里没有草原。", words);
    for (int i = 0; i < len; ++ i) {
        std::cout << words[i] << "|";
    }
    std::cout << std::endl;
    segmentor_release_segmentor(engine);
    return 0;
}

#include <iostream>
#include <vector>
#include "postag_dll.h"

int main(int argc, char * argv[]) {
    if (argc < 1) {
        return -1;
    }

    void * engine = postagger_create_postagger(argv[1]);
    if (!engine) {
        return -1;
    }

    std::vector<std::string> words;

    words.push_back("我");
    words.push_back("是");
    words.push_back("中国人");

    std::vector<std::string> tags;

    postagger_postag(engine, words, tags);

    for (int i = 0; i < tags.size(); ++ i) {
        std::cout << words[i] << "/" << tags[i];
        if (i == tags.size() - 1) std::cout << std::endl;
        else std::cout << " ";

    }

    postagger_release_postagger(engine);
    return 0;
}

#include <iostream>
#include <vector>
#include "ner_dll.h"

int main(int argc, char * argv[]) {
    if (argc < 2) {
        std::cerr << "usage: ./ner [model_path]" << std::endl;
        return -1;
    }

    void * engine = ner_create_recognizer(argv[1]);
    if (!engine) {
        std::cerr << "failed to load model" << std::endl;
        return -1;
    }

    std::vector<std::string> words;
    std::vector<std::string> postags;

    words.push_back("中国");    postags.push_back("ns");
    words.push_back("国际");    postags.push_back("n");
    words.push_back("广播");    postags.push_back("n");
    words.push_back("电台");    postags.push_back("n");
    words.push_back("创办");    postags.push_back("v");
    words.push_back("于");      postags.push_back("p");
    words.push_back("1941年");  postags.push_back("m");
    words.push_back("12月");    postags.push_back("m");
    words.push_back("3日");     postags.push_back("m");
    words.push_back("。");      postags.push_back("wp");

    std::vector<std::string>    tags;

    ner_recognize(engine, words, postags, tags);

    for (int i = 0; i < tags.size(); ++ i) {
        std::cout << words[i] << "\t" << postags[i] << "\t" << tags[i] << std::endl;
    }

    ner_release_recognizer(engine);
    return 0;
}

#include <iostream>
#include <vector>
#include "parser_dll.h"

int main(int argc, char * argv[]) {
    if (argc < 2) {
        return -1;
    }

    void * engine = parser_create_parser(argv[1]);
    if (!engine) {
        return -1;
    }

    std::vector<std::string> words;
    std::vector<std::string> postags;

    words.push_back("一把手");      postags.push_back("n");
    words.push_back("亲自");        postags.push_back("d");
    words.push_back("过问");        postags.push_back("v");
    words.push_back("。");          postags.push_back("wp");

    std::vector<int>            heads;
    std::vector<std::string>    deprels;

    parser_parse(engine, words, postags, heads, deprels);

    for (int i = 0; i < heads.size(); ++ i) {
        std::cout << words[i] << "\t" << postags[i] << "\t"
            << heads[i] << "\t" << deprels[i] << std::endl;
    }

    parser_release_parser(engine);
    return 0;
}

<?xml version="1.0" encoding="utf-8" ?>
<xml4nlp>
    <note sent="y" word="y" pos="y" ne="y" parser="y" wsd="y" srl="y" />
    <doc>
        <para id="0">
            <sent id="0" cont="我们都是中国人">
                <word id="0" cont="我们" pos="r" ne="O" parent="2" relate="SBV" />
                <word id="1" cont="都" pos="d" ne="O" parent="2" relate="ADV" />
                <word id="2" cont="是"  pos="v" ne="O" parent="-1" relate="HED">
                    <arg id="0" type="A0" beg="0" end="0" />
                    <arg id="1" type="AM-ADV" beg="1" end="1" />
                </word>
                <word id="3" cont="中国" pos="ns" ne="S-Ns" parent="4" relate="ATT" />
                <word id="4" cont="人" pos="n" ne="O" parent="2" relate="VOB" />
            </sent>
        </para>
    </doc>
</xml4nlp>

# -*- coding: utf-8 -*-
#!/usr/bin/env python
import urllib, urllib2

uri_base = "http://127.0.0.1:12345/ltp"

data = {
    's': '我爱北京天安门',
    'x': 'n',
    't': 'all'}

request = urllib2.Request(uri_base)
params = urllib.urlencode(data)
response = urllib2.urlopen(request, params)
content = response.read().strip()
print content

$ ./tools/train/otcws learn
otcws(learn) in LTP 3.3.0 - (C) 2012-2015 HIT-SCIR
Training suite for Chinese word segmentation

usage: ./otcws learn <options>

options:
  --model arg                  The prefix of the model file, model will be
                               stored as model.$iter.
  --reference arg              The path to the reference file.
  --development arg            The path to the development file.
  --algorithm arg              The learning algorithm
                                - ap: averaged perceptron
                                - pa: passive aggressive [default]
  --max-iter arg               The number of iteration [default=10].
  --rare-feature-threshold arg The threshold for rare feature, used in model
                               truncation. [default=0]
  --dump-details arg           Save the detailed model, used in incremental
                               training. [default=false]
  -h [ --help ]                Show help information

$ ./tools/train/otcws test
otcws(test) in LTP 3.3.0 - (C) 2012-2015 HIT-SCIR
Testing suite for Chinese word segmentation

usage: ./otcws test <options>

options:
  --model arg           The path to the model file.
  --lexicon arg         The lexicon file, (optional, if configured, constrained
                        decoding will be performed).
  --input arg           The path to the reference file.
  --evaluate arg        if configured, perform evaluation, input words in
                        sentence should be separated by space.
  -h [ --help ]         Show help information

$ ./tools/train/nndepparser learn
nndepparser(learn) in ltp 3.3.0 - (c) 2012-2015 hit-scir
training suite for neural network parser
usage: ./nndepparser learn <options>

options:
  --model arg               the path to the model.
  --embedding arg           the path to the embedding file.
  --reference arg           the path to the reference file.
  --development arg         the path to the development file.

  --init-range arg          the initialization range. [default=0.01]
  --word-cutoff arg         the frequency of rare word. [default=1]
  --max-iter arg            the number of max iteration. [default=20000]
  --batch-size arg          the size of batch. [default=10000]
  --hidden-size arg         the size of hidden layer. [default=200]
  --embedding-size arg      the size of embedding. [default=50]
  --features-number arg     the number of features. [default=48]
  --precomputed-number arg  the number of precomputed. [default=100000]
  --evaluation-stops arg    evaluation on per-iteration. [default=100]
  --ada-eps arg             the eps in adagrad. [defautl=1e-6]
  --ada-alpha arg           the alpha in adagrad. [default=0.01]
  --lambda arg              the regularizer parameter. [default=1e-8]
  --dropout-probability arg the probability for dropout. [default=0.5]
  --oracle arg              the oracle type
                             - static: the static oracle [default]
                             - nondet: the non-deterministic oracle
                             - explore: the explore oracle.
  --save-intermediate arg   save the intermediate. [default=true]
  --fix-embeddings arg      fix the embeddings. [default=false]
  --use-distance arg        specify to use distance feature. [default=false]
  --use-valency arg         specify to use valency feature. [default=false]
  --use-cluster arg         specify to use cluster feature. [default=false]
  --cluster arg             specify the path to the cluster file.
  --root arg                the root tag. [default=root]
  --verbose                 logging more details.
  -h [ --help ]             show help information.

$ ./tools/train/otcws customized-learn
otcws(customized-learn) in LTP 3.3.0 - (C) 2012-2015 HIT-SCIR
Customized training suite for Chinese word segmentation

usage: ./otcws learn <options>

options:
  --baseline-model arg         The baseline model, which should be saved with
                               --dump-details options.
  --model arg                  The prefix of the model file, model will be
                               stored as model.$iter.
  --reference arg              The path to the reference file.
  --development arg            The path to the development file.
  --algorithm arg              The learning algorithm
                                - ap: averaged perceptron
                                - pa: passive aggressive [default]
  --max-iter arg               The number of iteration [default=10].
  --rare-feature-threshold arg The threshold for rare feature, used in model
                               truncation. [default=0]
  -h [ --help ]                Show help information

$ ./tools/train/otcws customized-test
otcws(customized-test) in LTP 3.3.0 - (C) 2012-2015 HIT-SCIR
Customized testing suite for Chinese word segmentation

usage: ./otcws test <options>

options:
  --baseline-model arg  The path to the baseline model.
  --model arg           The path to the model file.
  --lexicon arg         The lexicon file, (optional, if configured, constrained
                        decoding will be performed).
  --input arg           The path to the reference file.
  --evaluate arg        if configured, perform evaluation, input words in
                        sentence should be separated by space.
  -h [ --help ]         Show help information