Seminar at Speech, Music and Hearing:
Automatic Methods for Dialogue Classification and Prediction
Opponent: Daniel Åkerberg
AbstractThe problem presented in this thesis is to specify and implement a solution
which could be used to automatically classify previously completed calls to
a call routing service, and furthermore, to specify and implement a solution
which could be used to automatically predict the outcome of calls to this call
routing service on-line, i.e. as they happen. The specic service on which
to investigate this problem is a residential customer care service designed for
a triple-play provider, providing TV, telephone, and broadband solutions to
end consumers. The customer care service is a call routing system, to which
end consumers can call and be routed to the appropriate customer service (for
example, the customer service for broadband issues, or the customer service
for invoice issues). As the triple-play provider has large-scale operations,
and oers a wide variety of products, their need for an easy-to-use, dynamic,
maintainable and inexpensive call routing system is obvious.
The method used to classify the calls to the service yields a classier
with an accuracy of 86%, which is a vast improvement on the majority-guess
baseline. The method used to predict problematic dialogues in calls to the
service yields a predictor with an accuracy of 96% after three dialogue turns,
which is an improvement on the majority-guess baseline by over 35%. The
method is also used to successfully predict transaction success after three
dialogue turns, with an accuracy of 82%, which is an improvement on the
majority-guess baseline by over 5%.
Although the work presented in this thesis was carried out on data collected
from this specic service, it is the author\'s claim that the results and
conclusions found in chapter 11 are largely applicable to any similar voice
controlled call routing service.
15:15 - 17:00
Tuesday December 16, 2008
The seminar is held in Fantum.
| Show complete seminar list