Microsoft Does Their Share to Make Chatbots Smarter

Chatbots are like a new, popular toy that everyone must have, but once they are played with the glamor wears off and you realize they are not that great.   For lack of better terms, chatbots are dumb.  They have minimal comprehension and can only respond with canned phrases.  Chatbots are getting better because companies are investing in linguistic resources and sentimental analysis.  InfoQ tells us about Microsoft’s contributions to chatbots’ knowledge in, “Microsoft Releases Dialogue Dataset To Make Chatbots Smarter.”

The Microsoft company, Maluuba, released a new chatbot dialogue dataset about booking vacations with the hopes to make chatbots more intelligent.  Maluuba accomplished this task by having two humans communicate via a chatbox, no vocal dialogue was exchanged.  One human was trying to find the best price for a flight, while the other human who played chatbot used a database to find the information.  Travel-related chatbots are some of the dumber of the species, because travel planning requires a lot of details, content comprehension, and digesting multiple information sources.

What makes travel planning more difficult is that users often change the topic of their conversation. Simultaneously you might discuss your plan to go to Waterloo, Montreal, and Toronto. We humans have no trouble with keeping apart different plans people make while talking. Unfortunately, If users explore multiple options before booking, computers tend to run into problems. Most chatbots forget everything you talked about when you suddenly enter a new destination.

Maluuba’s dataset is like enrolling a chatbot in a travel agent course and it will benefit anyone interested in a travel planning chatbot.  Maluuba is one of many companies, not just Microsoft owned, that are sharing their own expertise and building specific chatbot datasets.  One such company is Bitext, but their expertise lies in a number of languages they can teach a chatbot.

Whitney Grace, May 23, 2017

Comments

comments