QString MostEnglishName::getMostEnglishName(const Tags& tags)
{
if (tags.contains("name:en") && tags.get("name:en").isEmpty() == false)
{
return tags.get("name:en");
}
QStringList names = tags.getNames();
double bestScore = -numeric_limits<double>::max();
QString bestName;
for (int i = 0; i < names.size(); i++)
{
double score = scoreName(names[i]);
if (score > bestScore)
{
bestScore = score;
bestName = names[i];
}
}
return bestName;
}
void TagComparator::compareNames(const Tags& t1, const Tags& t2, double& score, double& weight,
bool strict)
{
//double score = LevenshteinDistance::score();
// Check out picard's coefficient
// or sum(top scores) / min(t1.count, t2.count)
// score for the same and score for different
score = 0;
weight = 0;
QStringList n1 = t1.getNames();
QStringList n2 = t2.getNames();
for (int i = 0; i < n1.size(); i++)
{
n1[i] = Translator::getInstance().translateStreet(n1[i]);
}
for (int i = 0; i < n2.size(); i++)
{
n2[i] = Translator::getInstance().translateStreet(n2[i]);
}
priority_queue<Entry, deque<Entry>, Entry> heap;
// create a n x m matrix of scores
vector< vector<double> > scores;
scores.resize(n1.size());
Entry e;
for (int i = 0; i < n1.size(); i++)
{
scores[i].resize(n2.size());
e.i = i;
for (int j = 0; j < n2.size(); j++)
{
e.j = j;
e.score = LevenshteinDistance::score(n1[i], n2[j]);
//LOG_INFO("n1: " << n1[i].toStdString() << " n2: " << n2[j].toStdString() << " " << e.score);
heap.push(e);
}
}
int scoreCount = (int)((double)std::min(n1.size(), n2.size()) / 2.0 + 0.5);
weight = scoreCount;
set<int> used1;
set<int> used2;
while (scoreCount > 0)
{
assert(heap.size() > 0);
e = heap.top();
heap.pop();
if (used1.find(e.i) == used1.end() && used2.find(e.j) == used2.end())
{
//LOG_DEBUG(" " << n1[e.i].toStdString() << ", " << n2[e.j].toStdString() << " " << e.score);
score += e.score;
used1.insert(e.i);
used2.insert(e.j);
scoreCount--;
}
}
if (weight > 0)
{
score /= weight;
}
// if this is strict checking and one entry doesn't have a name.
else if (strict && (n1.size() > 0) != (n2.size() > 0))
{
score = 0.2;
}
else
{
score = 1;
}
}
