Multi-Channel Textual Analytics for Customer Service Applications

The key aim of the Multi-Channel Textual Analytics for Customer Service Applications
project will be to demonstrate the use of intelligent automated response approaches for textbased
interactions with call centre-based customer services. Text based messaging e.g. Email,
Twitter, etc. in customer service interactions is becoming more prevalent, particularly by the
younger generation. The introduction of textual analytics is one way to significantly improve
customer services.
The textual analytics will be realised using Natural Language Processing (NLP) techniques
applied to the different text-based communications channels. The corresponding set of
interpretations and automated text-driven conversations will be seamlessly integrated with any
person-based interaction via the contact centre to provide coherent and seamless
communications with customers. The innovation of this project is through the development of
the:
a) Automated Interaction Engine that hosts the NLP algorithms to realise the automated textbased
interactions with the user. This will include mixed channel communication and will
ensure that the interaction appears to be ‘human-to-human’
b) Content Analysis Engine used to analyse the text-based content to obtain the relevant
sentiment and trend analytics. Sentiment analysis will be used to provide market acceptance
analytics and to enable real-time market response.
The benefits that will accrue to the users of new system are:
a) Service agents will be able to handle a significantly greater number of customers without
increasing the number of staff and while providing a better quality of service
b) Customers will encounter a more responsive and interactive customer service through the
provision of more accurate and up-to-date information on the handling of their issues
c) New business models will be possible based on the extent of automation, the range of
market sectors being supported and the richness of the automated interactions supplied.