I'm developing a recommender system for movies using a matrix containing data of films found in DBpedia. However, my matrix contains a lot of missing information (e.g. some films does not contain the year attribute, others the writer etc). So my question is if is it possible to use an imputation algorithm taking into account the features of DBpedia, which is mainly built from a collaborative scenario and therefore prone to errors, and that my information is mainly textual (only the year attribute is numeric). My attributes are Name, Country, Directors, Producers, Starring, Writers, Year, and Subject.  The missing data is denoted with a question mark (?). I've also attached an example of my matrix with only 100 films (the full matrix contains around 101000 films). 

More Cristhian Figueroa's questions See All
Similar questions and discussions