System and method for management of call data using a vector based model and relational data structure

1. A method for representing call content in a searchable database, comprising the steps of:

transcribing call content of a call to text;

projecting the call content to vector space, wherein projecting comprises creating a call vector representative of the call content by indexing the call content using a content index dictionary;

determining a similarity measure for the call vector in relation to classes of an atomic-class dictionary which provides classification hierarchy descriptions of call data stored in a relational database; and

classifying the call content in the relational database in accordance with similarity measure for the call vector.

2. The method as recited in claim 1 , wherein indexing the call based on the content includes indexing the content based upon at least one of n-grams, unigrams and word/phrase frequency when compared with a content index dictionary.

3. The method as recited in claim 1 , wherein the step of determining a similarity includes computing a distance measurement between centers of atomic classes and the call to determine the similarity to one of the classes.

4. The method as recited in claim 3 , wherein the step of classifying includes comparing the distance measurement to a threshold, if the threshold is exceeded, creating a new atomic-class for the call content.

5. The method as recited in claim 3 , wherein the step of classifying includes comparing the distance measurement to a threshold, if the threshold is not exceeded, determining whether a call destination match exists.

6. The method as recited in claim 5 , wherein if a call destination match exists, taking the call destination class as a parent class and classifying the call in a nearby atomic-class of this parent class.

7. The method as recited in claim 1 , wherein the step of classifying includes placing a recording of the call in an atomic class having a best similarity.

8. The method as recited in claim 1 , further comprising the step of searching the relational database to satisfy a query.

9. The method as recited in claim 1 , further comprising the step of associating call control data with the call content vector.

10. A program storage device readable by machine, tangibly embodying a program of instructions executable by the machine to perform method steps for representing call content in a searchable database as recited in claim 1 .

11. A method for representing call content in a searchable database, comprising the steps of:

collecting incoming calls from source channels;

transcribing content of the incoming calls;

classifying the call content using a vector space model which represents the call content in vector space; and

storing the call content represented by a vector in a relational data set structure to a target database,

wherein classifying the call content comprises:

projecting the call content to vector space, wherein projecting comprises creating a call vector representative of the call content for each call by indexing associated call content using a content index dictionary; and

determining a similarity measure for each call vector in relation to classes of an atomic-class dictionary, which provides classification hierarchy descriptions of call data stored in a relational database.

12. The method as recited in claim 11 , wherein the relational data set structure employs database call records and the content index dictionary and the atomic-class dictionary for representing a call by using the vector space model.

13. The method as recited in claim 11 , wherein the classifying includes indexing the content based upon at least one of n-grams, unigrams and word/phrase frequency when compared with a content index dictionary.

14. The method as recited in claim 11 , wherein the classifying includes computing a distance measurement between centers of atomic classes and the content to determine a similarity to one of the classes.

15. The method as recited in claim 14 , wherein the step of classifying includes comparing the distance measurement to a threshold, if the threshold is exceeded, creating a new atomic-class for the call content.

16. The method as recited in claim 14 , wherein the step of classifying includes comparing the distance measurement to a threshold, if the threshold is not exceeded, determining whether a call destination match exists.

17. The method as recited in claim 16 , wherein if a call destination match exists, taking the call destination class as a parent class and classifying the call in a nearby atomic-class of this parent class.

18. The method as recited in claim 11 , further comprising the step of searching the target database to satisfy a query.

19. A program storage device readable by machine, tangibly embodying a program of instructions executable by the machine to perform method steps for representing call content in a searchable database as recited in claim 11 .

20. A system for representing call content in a searchable database, comprising:

a management module, which receives, call content from a source, the management module farther comprising:

a relational data structure employed for indexing call content and classifying call content for converting the call content into a vector;

a content index dictionary which provides information for creating the call content indexing; and

an atomic-class dictionary which is employed to classify the call content; and

a database which stores the converted vectors to provide searchable call content.

21. The system as recited in claim 20 , further comprising a transcription means which converts the call content to text.