###Authors:
- Omar Shammas
- Tunde Agboola
###Heroku:
###Description
Ensemble is a system that allows users to interact naturally and in real-time with the crowd and receive suggestions based on their own preferences. The current model for task creation in online labour markets is very limiting. Requesters provide instructions, submit tasks, and await the results. This unidirectional communication model is unsuitable for a variety of tasks, in particular tasks dependent on the subjective tastes of the requester. Our solution creates a dialogue between workers on Amazon’s Mechanical Turk and requesters, allowing Turkers to build a profile detailing the requester’s preferences and perform nontrivial tasks.
Ensemble is a crowd driven fashion recommender. Users will be able to request an item of interest using Ensemble by simply typing a message in natural language or taking a photograph. Ensemble will route the request to the crowd. Given the objective specified by the user, the crowd will determine the salient features, offer proposals based on the criteria, garner feedback from the user, and repeat this iterative process until the user is satisfied. The crowd will simultaneously build a profile based on preferences that are implied from the requestors feedback. This user profile will allow Turkers to contribute relevant suggestions across different sessions.
The main component of Ensemble is a web interface that enables real-time interaction between turkers. The web interface will allow Turkers to communicate with one another and vote on suggestions posed by each other. The interface will also contain a user profile that contains all the preferences that have been inferred about the user from current and previous sessions. Suggestions and feedback will be sent to the requestor’s mobile device via SMS message or a lightweight mobile client.