Facing the challenge of human-agent negotiations via effective general opponent modeling