Artificial intelligence is steadily improving work efficiency across a wide range of industries, largely by decreasing human capital costs. AI-empowered technologies such as natural language processing (NLP) are increasingly active in the labour-intensive world of call centres — concentrated offices used for sending or receiving a large volume of requests by telephone.
Inbound call centres tend to be operated by a company to administer its customer product support or information enquiries. Outbound call centres meanwhile are widely used for telemarketing, debt collection, soliciting charitable or political donations, market research, etc.
The technology used in call centres roughly falls into four categories: real-time speech recognition; intent analysis, which frames conversations based on context to predict customer intent in real time; conversation management, which ensures simultaneous processing of multi-pass conversations; and conversational analysis, which broadly analyzes users’ dialogue.
Inbound call centre systems
The inbound call centre system has been supported by NLP technology for many years. In today’s marketplace NLP technology is applied both in general intelligent service platforms and in intelligent service platforms for specific professional fields.
NLP technology and artificial intelligence are increasingly active in the financial vertical field, banking and wealth management services. Through multiple rounds of question and answer, a customer service digital assistant can gradually screen and clarify users’ specific needs, and give targeted and appropriate responses.
Customer service on the e-commerce platform Youzan is an example of this. Countless customer queries are dealt with by the Youzan intelligent assistant, and the system can understand and meet needs for various types of e-commerce services.
Another inbound telephone system scenario is information consulting services. This is mostly a task-driven dialogue scenario where the goal of the system is to help guide the user to quickly and smoothly achieve a specified task using the lowest number of dialogue rounds.
Rsvp.ai provides a 114 information query service system for China’s Guizhou province. This is an example of a general information consultation call centre: the user makes a telephone call to query local information such as traffic, accommodation, food, weather and so on; and the system returns relevant information to the user in a specific form.
Telephone outbound systems
Telephone outbound systems are the other main type of call centre. These systems are more challenging than the inbound class of services, as there is almost no opportunity for a human to intervene or guide the dialogue interaction process. Current outbound call systems can generally be divided into two main types: debt collection services and message/information promotion services.
In the debt collection scenario the system need not place great consideration on the customer’s psychological experience as long as a clear intention and purpose are expressed in the conversation content. In a typical debt collection call the dialogue process between the system and user can be simplified to a three-round dialogue, which is relatively easy to implement.
The high demand for debt collection services is attracting many NLP startups. Yizhi Technology is telephone outbound startup whose NLP system is being used by many financial service institutions in China.
In addition to debt collection, outbound systems are also being deployed in other scenarios such as information promotion services. This includes product recommendations, advertisements, surveys, message notifications and so on. Although this is not very different in form from debt collection scenarios, there is a greater need to build a positive user experience into the design for information promotion services. People tend to be suspicious or annoyed by automated calls, and this creates challenges for artificial intelligence and NLP integration. Although NLP tech companies continue to explore ways to provide users with a better interactive experience, there are currently no outstanding products in this market.
Data analysis of call records
Call record data analysis is a relatively new but growing application for NLP technology. In order to provide better and more personalized services, machines can be tasked with assessing users’ emotions, intentions, and potential thoughts by analyzing associated telephone content information data. Demand for this type of service is now high in the e-commerce sector. E-commerce platforms usually have massive amounts of user telephone consultations and text message data. Through NLP technology, a machine can automatically organize and analyze this data.
One of the biggest Chinese e-commerce agencies, JD.com uses text data analysis technologies to provide users with more personalized product recommendation services and to help reduce user complaint rates.
As more and more merchants are now recording and retaining their customer service data, the calling record data analysis field can be expected to grow. NLP technology is able to parse text data from a calling system to discover and understand the content in a user’s calling data to benefit both merchants and customers.
As a cutting edge artificial intelligent technology, NLP and its intelligent voice assistants are now a part of our everyday lives, whether on e-commerce platforms, reservation centres or government agency information lines. As NLP technology and AI usher the call centre into an era of intelligence, labour costs will continue to fall and humans will be liberated from redundant work as machines make call centre services more personalized, automated, efficient and convenient.
Author: Ying Shan | Editor: Michael Sarazen