Everybody is excited about Google Wave and curious about what new forms of interactions with other people become possible. One special thing about Google Wave are the Wavelets. Wavelets are a form of collaboration workspace where people chat and share information. During my christmas holidays, finally, I decided to get my hands dirty and get my hands on the Google Wave technology to create a cool application.
Since I’m very into movie recommender systems, why not creating an app supporting the participants of a Wavelet what movie to watch together? That was my question and I came up with the following simple idea: Gather the movie preferences of each participants of a Wavelet and combine these movie preferences in a smart way such that the perfect movie recommendation can be provided to these combination of participants.
But how to capture the participants’ preferences. How to provide them a movie recommendation? In other words, how can a user interface of a recommender system look like in a Wavelet? Well, a Wavelet is a conversation of people. Why not make a recommender system part of the conversation? Why not creating a chat bot that captures the preferences from the participants’ conversation, learn their preferences and come up with a movie recommendation? Well, that’s exactly why I did as you can see in the following figure.
Chat bot participating in the conversation using natural language processing
I did not yet publish the chat bot. But I will soon. I named Colifri. the name Colifiri is a mixture of Colibri and Collaborative Filtering. There is no reason except for fitting the story why it must be a colibri. In the following, the architecture of the chat bot is presented.
Architecture of the chat bot which provides group recommendations
For the mean time, check out this presentation with the demo and architectural details of the chat bot and creating group recommendations:
