Video: Rules-Based Chatbot Design, Part 2: Visual Representation and Support
Learn more about chatbot design at the next SpeechTEK conference.
Read the complete transcript of this clip:Andreas Volmer: Something we also found to be extremely useful is a visual representation of the flows. So I mentioned earlier, in this Alexa Prize, the developers typically found the need for the conversations to have predefined snippets of conversations with turn-taking with the user and the system. That can typically be pretty easily implemented or displayed with a visual call flow, much - as you know - traditionally from IVR development. In this case it needs to be more flexible. IVR is typically very static, very confined. Here we're talking about much freer ways to be visualized. We try to find ways to visualize this in our platform.
And finally, in order to leverage the language capabilities of this English-language database, we needed to find ways to allow developers to take a similar approach to machine learning. A machine learning approach is where you start with simple phrases and train the system that way. This is what I'm showing here. I started a mockup of something that we're still currently implementing in our platform. In this workflow, there's no actual machine learning. But this particular step is very similar to machine learning where you start with a simple phrase like "What red wine goes with duck breast?" The system will automatically classify this according to certain generic intent categories, such as, “Is there somebody talking about a problem that they have? Is there somebody that needs information of the type ‘where,’ ‘what,’ ‘when,’ etc.? Is this a pleasantry that the system already knows about to accelerate the development?”
And then the user can say - In this phrase that I just mentioned, I think the following words are the most important ones to identify a category. So the customer might click on "duck breast" and then we do a search in the English-language database and find the different meanings of this term "duck breast.” "Duck breast" is pretty specific, but other words are much more loose and have more different types of class of meaning. The user can pick those meanings that really define the word in this context, the meaning of this word in this context, which has a big impact because this answer means every one of the different meanings would have different hypernyms and hyponyms. A bench, for example, can be something that you sit on, it can be something in court, "to bench" can mean to put some players back on the bench that have just been active on the field. Thinking about meaning is very important in the context of your bot.
We can also use hypernyms. "Duck breast,” to expand the hypernyms you see, is an extra-specialization of "meat," "poultry," "bird". So what is it that I'm really after? I can say any bird, really, or any poultry or any duck for that matter. And then you can test your rule finally, and then eventually say, "This rule is what I want to have in part of my application and use it for my conversational dialogue." This is using a phrase that we conceived of that can be used by non-technical users, even though what is being created internally is a rules-based approach, based on a highly trained, highly mature English-language model.
Another thing that is nice about this language model approach is, if you have a language model that is not only for the English language but for Spanish, German, French, etc., all the things that we do here in English--if this language model is properly represented and linked between the languages--we're allowed to work in other languages. Take the word "bench" again. "Bench" in court is a different meaning than "bench" just the wooden bench that you can sit on. And in a different language, this word "bench" might be represented by two different terms. If you use those proper categorizations here using such a tool, then what you end up with is a bot that very simply, very easily translates other languages, which is one of the great benefits of having a language model-based approach over a statistical-based approach where you have to retrain for every new language of your bot.
Omilia's Quinn Agen discusses the value of context and memory in delivering human-like interactions in self-service contact centers in this clip from SpeechTEK 2018.
SpeakEasy AI CEO Frank Schneider offers essential tips on using AI and active listening to enhance automatic speech recognition in this clip from SpeechTEK 2018.
Ulster University Professor Michael McTear discusses how contemporary natural language-based intelligent agents use intents and entities rather than traditional parse trees in this clip from SpeechTEK 2018.
Ulster University Professor Michael McTear identifies key resources for chatbot design best practices as well as tools and frameworks for building voice-user interfaces in this clip from SpeechTEK 2018.
Ulster University Professor Michael McTear discusses how interactions differ with different intelligent agents, from one-shot dialogue to system-directed dialogue and mixed-initiative in this clip from his presentation at SpeechTech 2018.
Ulster University Professor Michael McTear breaks down the intelligent agent field in terms of type, technology, and functionality in this clip from SpeechTEK 2018.
Omilia's Nikos Kolivas discusses the value of context and memory in delivering human-like interactions in self-service contact centers in this clip from SpeechTEK 2018.
Omilia's Quinn Agen discusses the way an an end-to-end conversational approach in machine-to-human interaction helps deliver satisfying contact center experiences in this clip from SpeechTEK 2018.
USAA's Brett Knight discusses how USAA has used machine learning and AI in virtual agent training at the enterpise level in this clip from SpeechTEK 2018.
Omilia's Quinn Agen discusses the advantages of taking a conversational approach in automated contact center customer care in this clip from SpeechTEK 2018.
USAA's Brett Knight describes all the players enterprises should assemble when planning conversational development for virtual agents in this clip from SpeechTEK 2018.
USAA's Brett Knight explains how to train virtual agents to anticipate customer needs more effectively in this clip from SpeechTEK 2018.
Amazon Connect General Manager Pasquale DeMaio demos Amazon Connect AI-driven Lex chatbot in this clip from his keynote at SpeechTEK 2018.
AWS Head of Product, Language Tech Vikram Anzabhagan outlines 4 essential strategies for leveraging technology to make contact center interactions more personal, conversational, agile, and engaging in this clip from his SpeechTEK 2018 keynote.
O'Reilly Media's Paco Nathan discusses the explosive growth in CPUs and GPUs that have opened new vistas for natural-language processing applications and rendered old approaches obsolete in this clip from his SpeechTEK 2018 keynote.
Paco Nathan of O'Reilly Media's R & D Group discusses the benefits of natural-language processing (NLP) when working with text converted from video in this clip from his SpeechTEK 2018 keynote.
Witlingo Product Manager Navya Nayaki Yelloji discusses the successful product development and QA strategy Witlingo created for its Motley Fool Alexa Skill in this presentation from SpeechTEK 2018.
You can hardly visit a company website without a chatbot popping up to offer you help these days. But these days, chatbots are helping with everything from mental health to identifying possible online predators. Is this trend here to stay? We asked Michael McTear, Allyson Boudousquie, Debra Cancro, and Crispin Reedy at SpeechTEK 2018.