From the definitions you might know about these two attacks!!

You need to construct your own alphabetical frequency analyzer to predict the cipher text from "chosen PT attack" and plain text from "chosen cipher text attack".. These two attacks used to find the key that has been used! That key will be used to reconstruct the original one!

Bottom line : Built your own analyzer and pass the "Chosen PT" and predict the key!!

I dont think there is a universal connection between the two that would be cryptosystem independent...The maths of RSA and AES are so different for instance...

Mr. Kanagaraj, I realize there are many more types of attacks in cryptanalysis. But is the Chosen PT and CT attacks that seem to be very similar in literature, I require a deeper understanding about these two attacks and their characteristic difference.

I answered for practicing the attacking types (especially for the mentioned attacks). Exp.: Sometimes, the hacker may get the (arbitrarily)plain text or cipher text through some source; that is said to be chosen PT/ chosen CT. The decrypted CT will be created using some key that is usually frequency analyzer. There after the same key is used to decrypt anything(further communications). This is the process of the both attacks. These attacks used by older days and it will work if and only if the keys is static but nowadays, the key becomes dynamic in nature. . Example : OTP - One Time Password. The communication channels also becomes complex by using SYMMETRIC and ASYMMETRIC i.e SSL.

Mr. Patrick ::: Chosen PT, Chosen CT, Known PT, Ciphertext-only attack are follows the same mechanism to predict. RSA, AES, DES, 3DES are falls under the Fiestel structure model sir. Everything correlates with each other sir.

Usually hackers scrutinise a cipher from both sides: the encryption procedure and from the decryption procedure. This helps the hacker in finding a better (encryption/decryption) side to simplify the algorithm. In a chosen plaintext (ciphertext) attack, the cryptanalyst has temporary access to the encryption (decryption) machinery, and so is able to construct the ciphertext (plaintext) corresponding to a chosen plaintext (ciphertext). In all such attacks, as correctly mentioned by Patrick, the inputs for attacks are deliberately chosen based on the characteristics of the encryption/decryption algorithms.

The major difference is that the chosen-cipher text attack is an unrealistic scenario which is rare to happen in practice. However, before being applied to practical applications, a cipher should be thoroughly analysed making use of all the well-known cryptanalysis procedures.

 In choosing plain text attack, after receiving the cipher text of a message, you need some of the plain text of the same encrypted message that was received. and then using a technique, depending of the cryptosystem, to match the cipher text with its identical plain text to get the part of key which help to find the hole key or break the algorithm.

in choosing cipher text attack, the cryptanalysis depend only on the cipher text of the received message to recover the key or break the algorithm by using for example statistical tests.

Simple example, to attack a message that encrypted using simple substitution .

The letter frequency of encrypted message is calculated. And depending on the previous information about English letters, letter E is high frequency. So the letter with most frequency with replace with letter E as plain text and so on.

You can use a mathematical representation of the original image. You decompose the face image into a number of characteristics feature image (called eigen face) such as eyes , mouth .. . . And construct the feature vector as a combination of the eigen vectors of the covariance matrix of a set face image .

The attacker who has the choice on plaintexts can mount even differential cryptanalysis.  He can deliberately pick plaintexts, that can be expected to reveal the structure of the key. 

This freedom is not with the attacker who has ciphertexts, as his choices (on patterns) are limited, especially if the cipher system has avalanche effect. 

Even though the requirements for both the attacks are similar, i.e, encryption algo, PT/CT and the corresponding CT/PT, a cipher system which is secure against chosen plaintext attack is secure against known plaintext attack also.

Good wishes.

The difference is in how the game is played between  the Oracle and the adversary. CPA and CCA  are indeed real life attacks. However the games represent frameworks for analyzing the cryptosystem in hand. The following are a brief description of these games (i hope i can make it simple to understand):

In CPA game, from the name, the adversary chooses two different plaintexts of her own (m1, m2), sends them to the encryption oracle, the encryption oracle flips a coin and encrypts one of the plaintexts returning the ciphertext back to the adversary. The adversary has to decide (without knowing the private key) whether this ciphertext is an encryption of m1 or m2. All cryptosystems that incorporates random coins in the encryption process are CPA secure (such as el-gamal cryptosystem) since the ciphertext is randomized, i.e. an encryption of the same message twice produces two different ciphertexts since the incorporated random coins are different for each encryption process (encryption is probabilistic).

CCA game (some times called midnight or lunch time attack) is somewhat similar to that of CPA. However, in addition to the  encryption oracle, the adversary is given access to a decryption oracle which decrypts arbitrary ciphertexts at the adversary's request, returning the plaintext.  The adversay is free to query decryptions of ciphertexts chosen by her own except the ciphertext in challenge.

There are also CCA-Type one and CCA-Type two.

Since in symmetric cryptosystems the encryption and decryption keys are the same and the encryption and decryption algorithms are almost the same, CCA security is similar to CPA security.

For more about these games as a frame work in the analysis of any cryptosystem i suggest you refer to the following valuable rich papers:

Non-malleable encryption. (Moni Naor)

Concrete security treatment of symmetric cryptosystems. (Phillip Rogaway)

Probabilistic Encryption. (Selvio Micali)