Function main

examples/sequence_segmenter_ex.cpp:169–235

// ----------------------------------------------------------------------------------------

int main()
{
    // Finally we make it into the main program body.  So the first thing we do is get our
    // training data.
    std::vector<std::vector<std::string> > samples;
    std::vector<std::vector<std::pair<unsigned long, unsigned long> > > segments;
    make_training_examples(samples, segments);


    // Next we use the structural_sequence_segmentation_trainer to learn our segmentation
    // model based on just the samples and segments.  But first we setup some of its
    // parameters.
    structural_sequence_segmentation_trainer<feature_extractor> trainer;
    // This is the common SVM C parameter.  Larger values encourage the trainer to attempt
    // to fit the data exactly but might overfit.  In general, you determine this parameter
    // by cross-validation.
    trainer.set_c(10);
    // This trainer can use multiple CPU cores to speed up the training.  So set this to
    // the number of available CPU cores. 
    trainer.set_num_threads(4);


    // Learn to do sequence segmentation from the dataset
    sequence_segmenter<feature_extractor> segmenter = trainer.train(samples, segments);


    // Let's print out all the segments our segmenter detects.
    for (unsigned long i = 0; i < samples.size(); ++i)
    {
        // get all the detected segments in samples[i]
        std::vector<std::pair<unsigned long,unsigned long> > seg = segmenter(samples[i]);
        // Print each of them
        for (unsigned long j = 0; j < seg.size(); ++j)
        {
            print_segment(samples[i], seg[j]);
        }
    }


    // Now let's test it on a new sentence and see what it detects.  
    std::vector<std::string> sentence(split("There once was a man from Nantucket whose name rhymed with Bob Bucket"));
    std::vector<std::pair<unsigned long,unsigned long> > seg = segmenter(sentence);
    for (unsigned long j = 0; j < seg.size(); ++j)
    {
        print_segment(sentence, seg[j]);
    }



    // We can also test the accuracy of the segmenter on a dataset.  This statement simply
    // tests on the training data.  In this case we will see that it predicts everything
    // correctly.
    cout << "\nprecision, recall, f1-score: " << test_sequence_segmenter(segmenter, samples, segments);
    // Similarly, we can do 5-fold cross-validation and print the results.  Just as before,
    // we see everything is predicted correctly.
    cout << "precision, recall, f1-score: " << cross_validate_sequence_segmenter(trainer, samples, segments, 5);