@Override
public void addThruPipe(Instance inst) {
Object data = inst.getData();
List<String> tokens = Collections.emptyList();
if (data instanceof String) {
tokens = Arrays.asList(((String) data).split("\\s+"));
} else if (data instanceof List) {
tokens = (List<String>) data;
}
ArrayList<String> list = new ArrayList<String>();
StringBuffer buf = new StringBuffer();
for (int j = 0; j < gramSizes.length; j++) {
int len = gramSizes[j];
if (len <= 0 || len > tokens.size()) continue;
for (int i = 0; i < tokens.size() - len + 1; i++) {
buf.delete(0, buf.length());
int k = 0;
for (; k < len - 1; ++k) {
buf.append(tokens.get(i + k));
buf.append(' ');
}
buf.append(tokens.get(i + k));
list.add(buf.toString().intern());
}
}
inst.setData(list);
}
/**
* 构造并初始化网格
*
* @param carrier 样本实例
* @return 推理网格
*/
protected Node[][] initialLattice(Instance carrier) {
int[][] data = (int[][]) carrier.getData();
int length = carrier.length();
Node[][] lattice = new Node[length][];
for (int l = 0; l < length; l++) {
lattice[l] = new Node[ysize];
for (int c = 0; c < ysize; c++) {
lattice[l][c] = new Node(ysize);
for (int i = 0; i < orders.length; i++) {
if (data[l][i] == -1 || data[l][i] >= weights.length) // TODO: xpqiu 2013.2.1
continue;
if (orders[i] == 0) {
lattice[l][c].score += weights[data[l][i] + c];
} else if (orders[i] == 1) {
int offset = c;
for (int p = 0; p < ysize; p++) {
// weights对应trans(c,p)的按行展开
lattice[l][c].trans[p] += weights[data[l][i] + offset];
offset += ysize;
}
}
}
}
}
return lattice;
}
@Override
public void addThruPipe(Instance inst) throws Exception {
String str = (String) inst.getSource();
BinarySparseVector sv = (BinarySparseVector) inst.getData();
List<RETemplate> templates = new ArrayList<RETemplate>();
for (int i = 0; i < group.size(); i++) {
RETemplate qt = group.get(i);
float w = qt.matches(str);
if (w > 0) {
// System.out.println(qt.comment);
int id = features.lookupIndex("template:" + qt.comment);
sv.put(id);
}
}
}
/** @return 预测序列和对照序列之间不同的Clique数量 */
@Override
protected int diff(Instance inst, float[] weights, Object targets, Object predicts) {
data = (int[][]) inst.getData();
if (targets == null) golds = (int[]) inst.getTarget();
else golds = (int[]) targets;
preds = (int[]) predicts;
int diff = 0;
if (golds[0] != preds[0]) {
diff++;
diffClique(weights, 0);
}
for (int p = 1; p < data.length; p++) {
if (golds[p - 1] != preds[p - 1] || golds[p] != preds[p]) {
diff++;
diffClique(weights, p);
}
}
return diff;
}
public void addThruPipe(Instance instance) throws Exception {
String[][] data = (String[][]) instance.getData();
int length = data[0].length;
int[][] dicData = new int[length][labels.size()];
int indexLen = dict.getIndexLen();
for (int i = 0; i < length; i++) {
if (i + indexLen <= length) {
WordInfo s = getNextN(data[0], i, indexLen);
int[] index = dict.getIndex(s.word);
if (index != null) {
for (int k = 0; k < index.length; k++) {
int n = index[k];
if (n == indexLen) { // 下面那个check函数的特殊情况，只为了加速
label(i, s.len, dicData);
if (!mutiple) {
i = i + s.len;
break;
}
}
int len = check(i, n, length, data[0], dicData);
if (len > 0 && !mutiple) {
i = i + len;
break;
}
}
}
}
}
for (int i = 0; i < length; i++)
if (hasWay(dicData[i])) for (int j = 0; j < dicData[i].length; j++) dicData[i][j]++;
instance.setDicData(dicData);
}