Welcome to weblogs.com.pk Sign in | Join | Help

Call Centres

In call centre applications, usually the most important desire from the call centre operator is automatic caller identification and to provide related details. For instance if it’s a mobile phone provider, they will wish that the system instantly provide the subscriber name, account status, availed services etc.


With increase in processing power and reduction in data storage cost, the need of “artificial intelligent suggestions” is increasing, considering the same above example, it might be desired that the system suggest things from the recently launched services/offer, e-g the system analyze the subscriber usage pattern, and if it detects that the subscriber calls more during night time and if the provider has recently introduced some package that is comparatively cheaper during that timings than the current package of the subscriber the system either automatically inform this suggestion while caller is holding or this information is displayed to the call centre agent that convey this information during the session.


The next thing call centre operator asks logging of the calls, preferably in searchable way. Logging who called, who attended and when is easy, as these variables can easily be calculated, but what if the operator asks to log what’s communicated in the sessions. Storing audio recording might be the first approach, unfortunately we don’t have enough processing power and techniques that we can search them, e-g how to search the calls during which the caller asks for the “value added services”.


The problems in audio recording approach are detecting who is caller and who is agent (smart voice recognition) and voice to text conversion and then methods of searching patterns for “value added services”. Things are emerging, but nothing so far with acceptable results. Transcriptions can solve this problem, but it has its own inherent cost and only companies with reasonable finances can make its arrangements.


Things get more complex if there is some kind of translation is involved or we are targeting the international languages. E-g a local Urdu call centre operator who wishes to identify all the calls during which caller pin point any problem with say, invoice delivery system. Now should we transcribe the calls into UNICODE text and then search them? How to solve language grammar issues?


Interesting research area I guess J

Published Monday, January 3, 2005 9:00 PM by khurram
Filed under:


No Comments

New Comments to this post are disabled