The media industry continues to evolve rapidly due to the growing number of delivery channels and consumption devices. Today, viewers consume content anytime, anywhere. Operators are shifting focus to experience-led, data-driven, open platform-based solutions to transform next-generation video services and provide immersive customer experiences. Services are becoming more integral to the overall user experience with the effort to make TV viewing more accessible, attractive and engaging. Innovations like Augmented Reality, Virtual Reality, Voice Controls, along with hyper-personalization, continue to evolve and enhance the end user experience.
ChatBots are one such value-added service that enable operators to engage with their customers in a more personal and contextual way. ChatBots are demonstrating how to create highly interactive and engaging conversations, offering a more holistic TV viewing experience for users.
So, what are ChatBots?
ChatBots are applications that help in simulating a personal interaction with the user either via written text messages or voice controls. Together with a well-defined user and voice interface, natural language processing (NLP), prediction-based context-mapped machine learning, and a backend fulfillment service, ChatBots can help map a unique customer journey with content preferences and provide the right personalized experience to the user - anytime, anywhere. Recently, service providers like Netflix, Hulu, AT&T and broadcasters like CNN and MTV have successfully used ChatBot technology to engage with their customers.
ChatBots provide an opportunity for a cross-product integration that can truly be a value-add for customers. The most obvious use cases for ChatBots are asking for information and performing a task. Here are some use cases that can be enhanced by using ChatBots:
- For advertising, pushing contextual advertisements to users, improving branding and ad impressions.
- For providing personalized and proactive conversation to TV watchers. ChatBots use contextual awareness to respond to queries quickly and efficiently.
- For messaging friends with links to a preview of currently watching content or an upcoming show.
- For set top interaction including the ability to record programs, schedule recordings, manage playlists, play content, display program details, etc.. An example would be the user texting on their chat application – “Record FRIENDS episode tonight for me”, and the action being performed on their connected set top device at home.
- For searching through content and getting instant results on the device of their choice. An example here would be the user texting on their chat application – “Are there any recent movies by Steven Spielberg” and the results being displayed on the chat window or bringing up the list of movies on the connected set-top-box at home.
- For pushing recommendations based on a feature like a genre, star rating, etc. or providing a recommendation based upon multiple criteria like genre, actor, etc.
Enabling a ChatBot Service:
ChatBots process the text or voice presented to them by the user, interpret what the user wants and make an inference to determine a series of appropriate responses based on this information. With the use of open platforms like Google Dialogflow, RASA, Microsoft LUIS, Wit.ai, etc., a ChatBot service can be executed. Below are some critical points to be considered for an operator rolled out ChatBot service:
- A user interface generally included as part of the operator’s OTT application or a standalone application linked to the subscriber's account. Voice-enabled controls can be built using virtual assistants such as Google Now, Apple Siri, and Microsoft Cortana. This is the human-machine interface (HMI) that takes user text or voice inputs, and works with the backend ChatBot related services before presenting the response back to the user along with performing some action.
- Bot Service is the primary backend service that acts as a gateway for receiving all utterances from the client application, leverages NLU to identify intents and entities, does intent matching and invokes the operator fulfillment service for further action. A Bot service can be implemented as a microservice and can be packaged as a Docker container.
- NLU (Natural Language Understanding) service performs natural language processing by parsing the given input data and establishing logical structure so as to understand the intent. NLU primarily helps in identifying intents and entities.
- Machine Learning (ML) core is the knowledge base which hosts and trains the ML model and helps to put context to the user’s requirement. Primary responsibility is managing contextual dialogues, and with each user interaction, the model will train itself and evolve for predictions.
- Finally, the operator’s fulfillment service, mainly for authentication, user data management, account management and post processing tasks or actions.
As operators continue to focus on improving user experience, incorporating automation into their end customer interactions and content watching behavior is becoming increasingly relevant. ChatBots are focused on a specific task and help interpret user intent and actions more accurately. They are in no means here to compete with Personal assistants like Amazon Echo, Google Assistant or Siri. Personal assistants have endless scope while the ChatBots serve a purpose. A ChatBot will indeed complement the video services landscape by creating new opportunities and forms of consumption and interaction. The key would be to find the right balance between being engaging and not intrusive. And regardless of the future of ChatBots, the emergence of a new engagement tool is indeed an exciting development in television viewing.