A discussion with Oracle on chatbots

With recent advancements in machine learning and the convergence of compute power and big data bringing artificial intelligence into the mainstream, intelligent bots will transform every facet of every industry and dramatically improve the customer experience. We recently had a discussion with Mr. Suhas Uliyar from Oracle to chat about how Oracle is using this transformative technology to help their customers succeed, including the aspects of its chatbot service.

ADM: What chatbots interfaces are starting to pick up steam and what has Oracle done to help integrate these chatbot services into products?

Uliyar: There are several channels with different integration interfaces and developers have to spend a lot of time to understand these interfaces and also what is supported and not supported. FB Messenger, WeChat, Skype, Line, Telegram etc, VPAs like Alexa, GoogleHome, HomePod etc, Voice interfaces like Siri, Cortana etc. We are also seeing Chat interface extensions to existing mobile apps and web pages for users to engage with customer service via BOT. What we have done to help developers is abstract this integration and provide this out of the box for developers to integrate their Bot into these channels. This not only reduces the speed at which they can deploy but also their total cost of ownership as this world is changing very fast.

ADM: What kind of technology does Oracle provide to help developers jump the hurdle of Natural Language Processing?

Uliyar: We have made it easy for developers to focus on building their BOT with a UI to set up their intents, dialog, entities without having to worry about the algorithms to process and understand the natural language and classify these inputs. We use a series of technologies based on neural networks and spectral embedding that uses language incident and linguistic modeling to increase the accuracy of processing natural language from the end user. This leaves the Bot developer to focus on setting up the Bot instead of fine tuning these algorithms

ADM: How do you see the evolution of Oracle in the chatbot industry going forward?

Uliyar: We are continuously adding additional algorithms to simplify Bot development. We are adding algorithms to understand user sentiment, image analysis, language translation, self learning, behavioral analysis etc. Bot developers can decide which additional algorithms they want to add to their Bot by just including these algorithms in their pipeline without having to worry about the research, development and fine tuning of all these algorithms.

ADM: Does Oracle have any enterprise integrations?

Uliyar: The value of the Bot is to surface up data intelligently to the end users. Oracle is the leader in integration to enterprise and cloud data sources and this is also provided out of the box reducing the effort, time and cost to expose data to the Bot.

ADM: What are some of the advantages of having a chatbot service ran in the cloud?

Uliyar: Delivered as a complete solution in Oracle cloud, it reduces the cost, effort and complexity of managing patching, upgrades and provides a reliable and secure cloud that handles backups, high availability, and fail over.

ADM: How does an Oracle bot handle dialog and how hard is it to develop?

Uliyar: We provide an easy way for developer to construct a dialog - a conversation with the end user. End users by human nature could potentially branch into different states  / context in course of a conversations for e.g. Lets say I want to transfer funds from Account A to someone. Lets say I start by asking the Bot - Pay Tom for dinner. The bot responds with - from which account. Lets say the user picks checking but realizes he is not sure of how much he has in the account, he switches context to ask for balance and further ask what were the transactions etc - in other words change the state from transferring money to someone to checking balance and transactions. At some point he decided to Pay Tom which he comes back to. The Oracle Bot platform makes it very easy to model this with built in state management so the developer does not have code and maintain the solution.