Given the task definition, example input & output, solve the new input case.
You are provided with a list of converations between a "USER" and "ASSISTANT". Figure out if the conversation provided as "classify_text" is authored by "USER" or "ASSISTANT". There are maximum of 7 conversations between "USER" and "ASSISTANT" in each input
Example: ASSISTANT: generally speaking what type of movies do you watch , USER: I like thrillers a lot. , ASSISTANT: thrillers? for example? , USER: Zodiac's one of my favorite movies. , USER: Zodiac the movie about a serial killer from the '60s or '70s, around there. , ASSISTANT: Zodiac? oh wow ok, what do you like about that movie, classify_text: And I just think serial killers, in general, are interesting, so the movie was really good. And it was just neat to see that world. Like it went really in-depth. It was like almost 3 hours long, so you got to really feel like you were a part of that world and time period. And see what the detectives and the police did and the investigators.
Output: USER
Based on last conversation in which the ASSISTANT ask's a question for which it waits for a reply from the "USER". We can classify this conversation as "USER" as it needs to respond for the given question.

New input case for you: ASSISTANT: what kind of movies do u like watching? , USER: I like movies that, you know, have a great story line, that are memorable, that have great acting, that you can't, you know, that make you cry or are very emotional movies. , ASSISTANT: can u name a movie that fits what you've mentioned? , ASSISTANT: what about that movie did u like? , USER: It's just very interesting overall. , ASSISTANT: what about a movie u didn't like?, classify_text: Movies I didn't like there are so many movies that are bad, you know, the the acting is terrible, it doesn't keep your interest. You just want to turn it off. Just poor poor acting, very cheaply image movies
Output:
USER