collection.h

使用具有增量学习的监控式学习方法。包括几个不同的分类算法。

#ifndef COLLECTION_H
#define COLLECTION_H

#include "box.h"
#include "document.h"

/**
 * Determines how to traverse classes when training.
 */
enum csm {
    LINEAR,		/**< One class at a time */
    CROSS,		/**< Different class every time, if possible */
    RANDOM,		/**< Random class */
    TIME		/**< Class with oldest message first (should be used
			     with LINEAR_SEQ to have that meaning) */
};

/**
 * Determines how to traverse messages when training.
 */
enum msm {
    LINEAR_SEQ,		/**< From beginning to end */
    RANDOM_SEQ		/**< Random order */
};

/**
 * Determines how many messages should be used for training.
 */
enum trm {
    ALL,		/**< Train with all messages, then test all messages */
    ALL_INTERLEAVED,	/**< For all messages, test it then train with it */
    CLASS_PERCENTAGE,	/**< Same specific percentage for each class */
    TOTAL_PERCENTAGE	/**< Specific percentage for whole collection */
};

int
collection_init(enum csm c, enum msm m, enum trm t, enum document_type dt);

int
collection_add_box(box *b);

int
collection_get_nod(void);

int
collection_get_notd(void);

document *
collection_get_document(void);

void
collection_drop_document(document *doc);

int
collection_get_class(void);

int
collection_get_msg(void);

int
collection_get_class_nod(int class);

int
collection_get_class_notd(int class);

int
collection_next_document(void);

int
collection_next_test_document(void);

#endif