Rating and Ranking
- the rating problem is to assign an artwork with a given score indicating its extrinsic value (like success of the artwork on the market in terms of bids and sales) as well as its intrinsic value (like the estimation given by an art expert or art curator)
- given a rating score for each artwork in a gallery, we have a ranking of the gallery artworks
- moreover, one can rate and rank artists and collectors by considering them as bags of artworks and extending the score from a single object to a set of objects in some suitable manner
- the rating problem is typically solved using centrality measures on networks that can be extracted from the dataset (like the sell or bid networks)
Price prediction
- the price prediction problem is to predict the price of an artwork at a given time in the future given a set of features of the artwork that are known at the present time
- the features for the prediction can be related to the bid and sale histories of the artwork (extrinsic features) or associated with the artwork itself (intrinsic features) such as the digital image or the more complex digital object (gif, video, sound, 3D object) representing the art
- in the simplest case, a multiple linear regression model is solved, with price as response (dependent) variable and the features as explanatory (independent) variables
- the price prediction can be used by an artist to set a reserve price for a new artwork or by a collector to have an estimate of how much to offer during an auction for an artwork
- it can be also used to estimate how the value of a collection will evolve in the near future
Art discovery
- the art discovery problem it to recommend art items to users that they might not have found otherwise
- users can be artists, interested in discovering similar artists for collaboration, or collectors (including art investors), willing to find new art to purchase that is somewhat overlapping with that already present in their collection
- recommender systems usually make use of either or both collaborative filtering and content-based filtering
- collaborative filtering approaches build a model from a user’s past behavior as well as similar decisions made by other users. For instance, if A and B are similar collectors in some meaningful sense and A purchased an art piece X that B does not have in their collection, then the system might recommend art X (or a similar one) to B as well
- content-based filtering approaches utilize a series of pre-tagged characteristics of an item in order to recommend additional items with similar properties. For instance, if a collector bought several artworks tagged as glitch art, then the system might suggest other glitch artworks not already in the in collector’s gallery
- a particular case of art discovery is the art radio feature: starting from a seed artwork, the system generates a compilation of related artworks that the user can sequentially visualize and stop when something of interest is found (possibly starting a new radio from this piece)
Internet of Things meets blockchain art
Internet of Things (IoT) e arte sono ottimi amici. L’interazione tra IoT e crypto art (arte su blockchain) è invece un argomento ancora in gran parte inesplorato. Questo lavoro si propone di esplorare questa connessione. In particolare, si vuole far interagire Arduino e Processing per creare un digital artwork che verrà tokenizzato come Non-Fungible Token in una galleria digitale quale SuperRare oppure async. In particolare, async permette di realizzare arte programmabile che cambia in base a determinati eventi.