An online database and front-end interface

The goal is to design an online database and front-end interface that presents website visitors with images, one at a time, and solicits feedback about whether or not those images contain features that are randomly drawn from the database. For example, participants may be shown a picture of cat, and asked if they see “a tail” in the picture. The pictures that are displayed will also be randomly chosen from a large set of potential images. We want participants to be able to navigate through this process by having a button that allows visitors to submit feedback about the presence or an absence of a specific feature. Upon submission of this information, visitors will then be presented with a new feature from the database for the same image. We also want a second button that will allow visitors to generate a new image accompanied by a new feature. This button will also need to submit their feedback regarding the feature that was present when they requested a new image. Initially, when asked about the features, we want participants to respond on a rating scale (e.g. to what extent is there a tail in this image? 0%-100%). However, as we accrue feedback, we would like to be able to change the nature of the response option from a rating scale to a dichotomous (Yes / No) response when everyone agrees about the clear presence or absence of that feature across images (e.g. always 0% or 100%, few responses in between). This database also needs to be flexible enough to allow us to add new images and new features at later stages of the project. We would also like a system of weighting for the randomization, so that images / features that have been used more frequently are less likely to be presented than images / features that have been used less frequently. We would also like to have a way to track the responses of specific users (e.g. individual accounts requiring login), by keeping a running summary of how many features they have responded to, and a “credibility” score. Credibility scores will be based on responses to specific diagnostic features that are consistently answered in same way by nearly all visitors.