Logo AHome
Logo BIndex
Logo CACM Copy

shortpapTable of Contents


The Telephony Customer Interface:

Five Perspectives on Problems and Solutions

John Chin*, Gregory Tatchell**

*AT&T Bell Labs

200 Laurel Avenue

Middletown, NJ 07748

**Stentor Resource Center, Inc.

4535 Canada Way

Nurnaby, B.C., Canada, V5G 1J0


INTRODUCTION

Collectively, this suite of five papers illustrate the diverse set of issues that should be addressed to formulate a strategic path towards resolving existing and growing usability problems found in telecommunications. Specifically, there are three main thrusts applied in developing profitable and competitive new services:

1) identifying sources of usability problems

2 exploring potential design solutions that address current and future user needs

3) implementing prototypes to assess the feasibility of design concepts.

UNUSABLE, CONFUSING TELEPHONY INTERFACES

Typically, customers subscribe to only a few of the many different telephony features available in the network. The lack of subscription to these features is attributed to the myriad confusing codes, rings, tones and announcements associated with the usage of these network services. Over 30 different usability problems have been identified in existing telephony customer interface associated with these services [1]. The complex interactions between these services create unintended side effects. These service interactions become more pronounced when greater intelligence and functionality is introduced into network services [2]. The ultimate challenge for future intelligent network services is to resolve these interactions in a manner natural to users of communication services.

FUTURE AGENT BASED TELEPHONY INTERFACES

We believe that a fifth generation telephony customer interface is the solution to creating a usable local access network. As we replace the archaic touch-tone interface of today with a desirable, intuitive, interactive, intelligent agent based, voice activated customer interface, we believe it can stimulate the same exponential growth in local access services as Graphical User Interfaces (GUIs) have done for computer based services.

INTERACTIONS IN A SOCIAL CONTEXT

A conversational interaction with an agent must be defined by the social context in which telephony applications will be used. Market research has been conducted to determine users needs and the types of tasks that may be delegated to agents. The function of agents may be viewed in terms of occupations or professions (i.e. butlers). Delegation of tasks must begin with proper job description for the agent application [3]. The nature of the agent-user interaction is defined by user expectations and perceptions of the personality or image of the system. Before future systems are designed, user perceptions of current systems (i.e. voice mail) should be assessed. Once desirable characteristics or personality traits are identified, they should be incorporated into future agent-based products. Future anthropomorphic agents need to be designed with personalities that are appropriate for novice and expert users [4].

IMPLEMENTATION OF A PROTOTYPE

A concept prototype was built to support scenario based user interactions that addressed specific customer needs, including: a) security with speaker identity verification, b) call origination and screening with Automatic Speech Recognition (ASR) and c) seamless integration of features across wireline and wireless services. The prototype utilized a platform that emulates the potential future configuration and network architecture of these agent-based services [5].

LIST OF RELATED SHORT PAPERS

  1. Problems with the Existing Telephony Customer Interface: The Pending Eclipse of Touch-Tone and Dial-Tone (Stentor).
  2. Making a Simple Interface Complex: Interactions among Telephone Features (Bellcore).
  3. From Interactions to Interfaces: Butlers, Job Descriptions and Personal Agents (MPR).
  4. Personality Trait Attributions to Voice Mail User Interfaces (AT&T)
  5. Technical Considerations in the Design of an Intelligent Agent Using ASR: (MPR)