Browse Prior Art Database

Method of intelligent selection against service types based on mass voice data analysis

IP.com Disclosure Number: IPCOM000249266D
Publication Date: 2017-Feb-15
Document File: 4 page(s) / 75K

Publishing Venue

The IP.com Prior Art Database

Abstract

Our solution is a cognitive based, client orient and self-learning telephone service system, the client is active to ask what he/she want to get, and system judge how to provide the service, if the system couldn't get the question clearly at the beginning, it will issue some questions to the client and after several loop, the customer could get the final served quickly. This method is more direct, good interactive and user friendly than current telephone service system.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 39% of the total text.

1

Method of intelligent selection against service types based on mass voice data analysis

Customer service is a important part of a corporation, nowadays, most corporations support telephone service to the clients, it's very convenient to people to call a number to contact the support representative, for new product information, after-sales service, consultant, or even complain and suggestions. Compare to message chat and e-mail, telephone have its advantages, it would be more direct, good interactive and user friendly. Current DTMF (Dual Tone Multi Frequency) technology is low efficiency. Our solution is a cognitive based, client orient and self-learning telephone service system, the client is active to ask what he/she want to get, and system judge how to provide the service, if the system couldn't get the question clearly at the beginning, it will issue some questions to the client and after several loop, the customer could get the final served quickly.

2

3

1) Voice recognition module use cognitive system to analyze the voice that explained their request to get key words. Then the cognitive system will use these key words to find correct customer service. R2 dictionary and a lexicon should be provided for cognitive system. The R2 dictionary and lexicon include special word and phrase in the field in which our system is used. cognitive system can easily translate the voice that describe the customer’s problem to a set of key words. Then these key words are transported to cognitive system. If cognitive system finds correct customer service, the process will continue. If not, cognitive will give back a further question in order to get more details. The system received new voice and analyze it. The system transport both new key words and old key words to cognitive system again. Repeat this until the cognitive system match the correct service or repeat times are over 5. If cognitive system can’t pick up key word because dictionary and lexicon are insufficient. System will directly connect client to a common operator. And add new entity into the dictionary or lexicon after finishing service.

2) Cognitive module is responsible for tree construction and tree search.

Tree construction This tree is constructed based on bank’s various business layers. After self-learning, meaningful leaf nodes will be added. Normally leaf node indicates customer request in popular words. Before this system deploy on production, it need be combined with existing voice service system to complete its self-learning course. Each leaf node has a weight according to request frequency.

Tree search Cognitive module received input from voice recognition system and input is formatted as key words from customer request. 1. search key words in all the leaf nodes, if it is fully matched one of the leaf nodes, cognitive module send this node information to screen system and assign a most suitable operator to provide customer service. 2. search key words in all the leaf nodes, if i...