NLP Assignment #2: Sundararajan Srinivasan

All FSAs in this assignment were drawn using isip_network_builder. Since I was not able to draw curved arcs using this, the inputs that should have been on the arcs, were drawn as nodes. The distinction between between the actual nodes and these nodes can be understood from the context.

(1) Possible telephone numbers:
    7 digits  : Without area code, should not start with 0 or 1.
    10 digits : With a 3-digit area code, with the area code's first digit not equal to 0 or 1
    11 digits : With International code 1 for US and also area code
    
    FSA can be seen by opening "isip_network_builder assignment2_1.text".
    The 3 branches at the start lead to the option of including international code and the area code.
    D represents 0-9 while N represents 2-9. Dn, D7_n and Dnn represent states.

(2) Open using "isip_network_builder assignment2_2.text".
    This FSA has 2 levels:
    At the first level, the inputs are the spoken digits ZERO/OH to NINE, along with appropriate transitions for silence and speech.
    At the second level, each word is modelled as HMM with N states (Sn). Though the number of levels is shown as 11,
    digits like SEVEN have 18 states as they cannot be represented adequately with 11 states. Since the path from start to stop
    cannot be known beforehand, it is a non-deterministic FSA.

(3) Open using "isip_network_builder assignment2_3.text".
    This FSA has 3 levels:
    At the first level, the inputs are the spoken digits ZERO/OH to NINE, along with appropriate transitions for silence and speech.
    At the second level, each word is broken down into it's constituent morphemes.
    At the final level, each morpheme is modelled as HMM with 3 states (Sn). Since the path from start to stop cannot be known
    beforehand, it is a non-deterministic FSA.

(4) Open using "isip_network_builder assignment2_4.text".
    
    In the NFSA, the non-deterministic nature can be seen when the machine is in state S2 and the input is "b". The machine finds itself
    with a decision to make: to move forward to S3 or to go back to S1.
    There is no such non-determinacy in the DFSA shown.

(5) The type of language represented in our XML and JSGF grammar formats is "Regular Language". These languages can be represented using
    NFSAs. Any FSA, whether deterministic or not, can be described by a regular expression. Hence, the set of all NFSAs that can be generated/
    recognized using these grammar formats, is a regular language.