The Research And Desigh Of User Behavior Analyze System Based On IVR Self-help System

The call-center system takes a great responsibility to link the companies and theircustomers and it’s still the best choice for large companies, organizations because of itstemporality of communication.With the increscent of voice menus,the self-help voice menu system gets complicatewhich makes customers getting lost in the maze of menus. The main reason for this awkwardsituation is badly composed voice menu structure.How to improve the behavior of self-help voice menu system is one of the mostconcerned question for all operating personnels.The application of theory of user behaveresearch in human-machine interactive UI is one of the most frequently usage. In IVR system,most frequently used method includes user behavior analyses and user attribute analyze.This thesis aim to use the theory of data mining, apply the method and flow path ofhuman behavior analyze to the self-help voice menu system, and design a self-adjusted IVRsystem which base on the structure of data-mining system framework and organize the voicemenu automatically. The system collects interactive key-press data,applies to data miningwork flow for analyzing the relation of key presses and use the result in IVR runningenvironment. The main data processing algorithm is Correlation algorithm. During the mainprocess of generating results data, the system filter the noise from the original key-press data,apply the data mining algorithm, generate the correlation results of the IVR menu data andorganize the voice menu automatically and dynamically. In order to improve the efficiency,accuracy and stability of the predict menu data, some parameter in the system and Correlationalgorithm are installed with some experiment method.In the end, the last chapter will compare the normal static voice menu system and the one applies the human interactive results. Two indicators are lead into the estimate: theaverage key press in IVR system and the average operation time in IVR system. In bothestimates of the two indicators, the results verified the system could improve the efficiency ofIVR menu operation.