Can I use the data of Friends TV show transcript in my thesis for analysis and publish my paper/findings. Does the usage of TV show transcript data for analysis and construction of new script based on the original script come under fair use of exception in US copyright law ? Can you please advise me on how to proceed further.
The description of your activities that you have provided, using a 'script' from a popular TV show for research purposes in a machine learning algorithm is allowed within the 'fair use' description of copyright. It is my understanding that the following descriptions place the usage described within 'fair use':
Amount and substantiality of the portion used in relation to the copyrighted work as a whole.
Since you are using the script to train a machine algorithm, and not the video, audio, nor are you reproducing it for your audience, it seems to be well within fair use.
Effect of the use upon the potential market for or value of the copyrighted work.
Using a script to train a machine algorithm, it seems, would have virtually no effect on the 'potential market or value of copyrighted work'.
Highlighted portions from "More Information on Fair Use by Copyright.gov"
Another possible solution you could consider is to train your machine learning algorithm on a different humorous show whose copyright has expired and is therefore in the Public Domain. Here is a list of humorous televeision shows whose copyright has expired, including "The Beverly Hillbillies".
I wouldn't. Fair use is a defense in court to a claim of copyright infringement brought against you by the copyright holder. That means that you have to get sued, hire a lawyer who fails to convince them to drop the suit due to your use under the fair use principles, and then take your defense to trial before a judge or jury. This, while unlikely to happen, is likely to cost you more than $10k and maybe as much as $100k if you get sued. It's not worth it.
I'm not a lawyer, nor am I your lawyer, but I'd find a play or radio play that is now in the public domain and use that instead. The idioms and colloquialisms are likely to be dated and may mess up your algorithm, but that's better than getting sued and being out the money.