CLIAMS:1. A method for providing context based conversations, the method comprising:
receiving, by a processor (202), user data pertaining to a user, wherein the user data comprises registration information of the user and metadata associated with the user;
determining, by the processor (202), a role of the user, wherein the role of the user is pre-defined based on the registration information;
providing, by the processor (202), restricted access to a users’ data repository to the user, based on the role of the user;
obtaining, by the processor (202), a text input from the user, wherein the text input pertains to a conversation;
generating, by the processor (202), an expression, based on the text input;
invoking, by the processor (202), one of a discussion service, a learning service, and an unlearning service, based on the expression and the metadata associated with the user; and
retrieving a response, by the processor (202), based on at least one of the invoked services, and the metadata associated with the user, wherein the response is shared with the user.

2. The method as claimed in claim 1, wherein the expression is generated by the processor (202) by parsing, to convert a natural language into a functional language.
3. The method as claimed in claim 1 further comprising classifying, by the processor (202), the expression into one of a discussion, learning, and unlearning.
4. The method as claimed in claim 3, wherein the expression is classified as the discussion when the expression is an execute expression, and wherein the classification of the expression as discussion comprises:
obtaining nodes corresponding to the text input and a relationship between the nodes from the execute expression;
determining a query from at least the obtained nodes and relationship, based on a baseform of the relationship;
identifying a domain of the query based on at least the user graph, factstore and metadata associated with the user, wherein the user graph (152) comprises historical conversation data corresponding to a user, factstore (156) comprises Enterprise factual data, and the metadata comprises information regarding the temporal data and the user persona data;
based on the domain, retrieving at least a response for the text input;
computing the confidence score for the at least one response, wherein the computation is based on a plurality of parameters, determining the context of the discussion and the relevance of the response to the query; and
providing the response to the user when the confidence score of the at least one response is above a pre-defined threshold value.
5. The method as claimed in claim 4, wherein the user persona may be based on aggregation of a plurality of historic and current user conversations, further indicating the interests of the user.
6. The method as claimed in claim 3, wherein the expression is classified as learning when the expression is a Subject Verb Predicate (SVP) expression, and wherein the classification of the expression as learning comprises:
retrieving at least one fact from the SVP expression, comprising at least two nodes and a relationship between these nodes, wherein the fact is being taught by the user;
storing the fact in the user graph, wherein the user graph comprises a user node, and the at least two retrieved nodes and the retrieved relationship between these nodes;
computing a confidence score for the at least one fact, wherein the computation is based on a plurality of parameters determining the context of the learning and the validity of the fact thereof; and
storing the fact in a factstore (156) when the confidence score of the at least one fact is above a pre-defined threshold value.
7. The method as claimed in claim 6, wherein the plurality of parameters comprise a source of a fact, frequency of a fact, fact confidence, credibility of a user, and association with other entities previously discussed/taught by a user.
8. The method as claimed in claim 6, wherein the at least one fact is directly stored in the factstore (156) by employing one of a dynamic mode and a scheduled mode.
9. The method as claimed in claim 1, wherein the learning service is performed as bulk learning at pre-defined time intervals.
10. The method as claimed in claim 3, wherein the unlearning comprises:
obtaining a list of previously taught facts from a user graph (152);
receiving a selection of at least one fact for being deleted from the user graph (152), wherein the at least one fact is one of incorrect, redundant, and irrelevant; and
deleting the at least one fact from the user graph (152).

11. The method as claimed in claim 10, wherein the deleting comprises:
determining that the fact is pending for storage in the factstore;
upon determining, confirming whether the fact is taught by a single user; and
deleting at least a data node representing the fact and a relationship of the data node with the single user from the user graph (152).

12. The method as claimed in claim 10, wherein the deleting comprises determining whether the fact is taught by multiple users, and updating the properties associated with the relationship of the data nodes based on the determining.
13. The method as claimed in claim 1 further comprising generating, by the processor (202), a user graph (152) based on the user data, wherein the user graph (152) comprises a user node, previously discussed/taught data nodes, and corresponding relationships, pertaining to the users currently logged into the system.
14. A context based conversation system (102) comprising:
a processor (202);
an authorization module (112), executable by the processor (202), to,
receive user data from a user, wherein the user data comprises login credentials of the user and metadata associated with the user;
based on the user data, determine a role of the user, wherein the role of the user is pre-defined; and
provide restricted access to the user, based on the role of the user;
a parsing module (114), executable by the processor (202), to,
receive text input from the user, wherein the text input is in natural language;
parse the text input to convert the text input from natural language to functional language;
based on the parsing, generate an expression from the text input; and
a classification module (116), executable by the processor (202), to
determine the type of expression from one of a discussion, learning, unlearning; and
invoke, based on the determination, one of a discussion service, a learning service, and an unlearning service;
retrieve a response based on the expression and the provide the response to the user.
15. The context based conversation system (102) as claimed in claim 14 further comprises a graph builder module (208), executable by the processor (202), to generate a user graph (152) depicting currently active users and facts or data previously and currently discussed by such users.
16. The context based conversation system (102) as claimed in claim 15, wherein the user graph (152) comprises a user node, previously discussed/taught data nodes, and corresponding relationships, pertaining to the users currently logged into the system.
17. The context based conversation system (102) as claimed in claim 14 further comprises a clustering module (210), executable by the processor (202), to generate user persona based on aggregation of a plurality of historic and current user conversations.
18. The context based conversation system (102) as claimed in claim 14 further comprises a scoring module (216), executable by the processor (202), to compute a confidence score to each fact being entered by a user, wherein the fact is entered by the user using a discussion service and a learning service, and wherein the computation is based on a plurality of parameters.
19. The context based conversation system (102) as claimed in claim 18, wherein, for learning service, the plurality of parameters may comprise a source of a fact, frequency of a fact, fact confidence, credibility of a user, and association with other entities taught/discussed recently by the user.
20. The context based conversation system (102) as claimed in claim 18, wherein the facts having the confidence score above a threshold value are stored to a factstore.
21. The context based conversation system (102) as claimed in claim 14 further comprises a maintenance module (214), executable by the processor (202), to,
move the facts from a user graph (152) or the context graph (154) to a factstore (156) over a period of time; and
cleanse a context graph (154) to remove irrelevant data from the context graph (154).

 
22. The context based conversation system (102) as claimed in claim 21, wherein the maintenance module (214) cleanses data from the context graph (154) in a scheduled mode and a demand-based mode.
,TagSPECI:As Attached