void MlSldaState::InitializeAssignments(bool random_init) {
InitializeResponse();
InitializeLength();
LdawnState::InitializeAssignments(random_init);
if (FLAGS_num_seed_docs > 0) {
const gsl_vector* y = static_cast<lib_corpora::ReviewCorpus*>
(corpus_.get())->train_ratings();
boost::shared_ptr<gsl_permutation> sorted(gsl_permutation_alloc(y->size),
gsl_permutation_free);
boost::shared_ptr<gsl_permutation> rank(gsl_permutation_alloc(y->size),
gsl_permutation_free);
std::vector< std::vector<int> > num_seeds_used;
num_seeds_used.resize(corpus_->num_languages());
for (int ii = 0; ii < corpus_->num_languages(); ++ii) {
num_seeds_used[ii].resize(num_topics_);
}
gsl_sort_vector_index(sorted.get(), y);
gsl_permutation_inverse(rank.get(), sorted.get());
// We add one for padding so we don't try to set a document to be equal to
// the number of topics.
double num_train = corpus_->num_train() + 1.0;
int train_seen = 0;
int num_docs = corpus_->num_docs();
for (int dd = 0; dd < num_docs; ++dd) {
MlSeqDoc* doc = corpus_->seq_doc(dd);
int lang = doc->language();
if (!corpus_->doc(dd)->is_test()) {
// We don't assign to topic zero, so it can be stopwordy
int val = (int) floor((num_topics_ - 1) *
rank->data[train_seen] / num_train) + 1;
// Stop once we've used our limit of seed docs (too many leads to an
// overfit initial state)
if (num_seeds_used[lang][val] < FLAGS_num_seed_docs) {
cout << "Initializing doc " << lang << " " << dd << " to " << val <<
" score=" << truth_[dd] << endl;
for (int jj = 0; jj < (int)topic_assignments_[dd].size(); ++jj) {
int term = (*doc)[jj];
const topicmod_projects_ldawn::WordPaths word =
wordnet_->word(lang, term);
int num_paths = word.size();
if (num_paths > 0) {
ChangePath(dd, jj, val, rand() % num_paths);
} else {
if (use_aux_topics())
ChangeTopic(dd, jj, val);
}
}
++num_seeds_used[lang][val];
}
++train_seen;
}
}
}
}
int BotClient::ChatBotFunctions(string buffer)
{
MessageHandler data(buffer);
int pos=0;
if (logger)
{
stringstream buf;
for(int i=0;i<(int)data.message.size();i++)
buf << data.message[i] << " ";
if (data.message.size() > 0)
LogChat(data.sender, buf.str());
}
if( (pos=data.atPosition("-change_nick")) != -1 && (int)data.message.size() > pos+1 )
{
Nick(data.message[pos+1]);
conn->Send("PRIVMSG #" + conn->GetChannel() + " :name changed\r\n");
}
if( (pos=data.atPosition("-join")) != -1 && (int)data.message.size() > pos+1 )
{
Join( data.message[pos+1] );
}
if( (pos=data.atPosition("-change_topic")) != -1 && (int)data.message.size() > pos+1)
{
ChangeTopic(data.message[pos+1]);
conn->Send("PRIVMSG #" + conn->GetChannel() + " :topic changed\r\n");
}
if( (pos=data.atPosition("-leave")) != -1 && (int)data.message.size() > pos+1 )
{
conn->Send("PRIVMSG #" + conn->GetChannel() + " :ciao!\r\n");
Leave( data.message[pos+1] );
}
if((pos=data.atPosition("-show_chatlog")) != -1 )
{
ShowLog();
}
if ((pos=data.atPosition("-last_seen_user")) != -1 && data.message.size() > pos+1)
{
ShowLastSeen(data.message[pos+1]);
}
if ((pos=data.atPosition("-quit")) != -1 )
{
conn->Send("PRIVMSG #" + conn->GetChannel() + " :cya!\r\n");
conn->Send("QUIT cya\r\n");
return -1;
}
if ((pos=data.atPosition("-log")) != -1 && (int)data.message.size() > pos+1 ) {
if (data.message[pos+1] == "on" )
{
logger = true;
conn->Send("PRIVMSG #" + conn->GetChannel() + " :logging on\r\n");
} else if( data.message[pos+1] == "off" )
{
logger = false;
conn->Send("PRIVMSG #" + conn->GetChannel() + " :logging off\r\n");
}
}
return 0;
}
