During the chosen-plaintext attack, a cryptanalyst can choose arbitrary plaintext data to be encrypted and then he receives the corresponding ciphertext. He tries to acquire the secret encryption key or alternatively to create an algorithm which would allow him to decrypt any ciphertext messages encrypted using this key (but without actually knowing the secret key).
This is a rather comfortable situation for the attacker. He can obtain more information about the secret key and about the whole attacked system, because he is able to choose any text to be processed by the cipher. He can analyse the system behaviour and output ciphertext, based on any kind of input data.
During breaking deterministic ciphers with the public key, the intruder can easily create a database with popular ciphertexts, for example with popular queries to the server. After that he will be able to find the meaning of many intercepted encrypted messages, by simply comparing them with his own database entries.
The most known chosen-plaintext attacks were performed by the Allied cryptanalysts during World War II against the German Enigma ciphers.
In this kind of chosen-plaintext attack, the intruder has the capability to choose plaintext for encryption many times. Instead of using one big block of text, it can choose the smaller one, receive its encrypted ciphertext and then based on the answer, choose another one, and so on. This allows him to investigate the attacked system in much more details.