For the same reason, the algorithm can’t take the multiple meanings of words into consideration, so words such as “well” and “like” are often marked as positive, even when they’re used in neutral ways.