Abstract: A method includes generating recommendations and user structures by applying a recommender machine learning model to training user information and item information, and generating, from the user structures and by applying a demographic machine learning model, demographic predictions of users represented by the user structures. The method further includes generating a first accuracy measure of the demographic machine learning model based on a first comparison of the demographic predictions with demographics of the users. A recommender loss function is generated based on the first accuracy measure and a second comparison of the recommendations with selections of users, where the recommender loss function uses the first accuracy measure to suppress detectability by the demographic machine learning model. The method further includes updating the recommender machine learning model according to the recommender loss function.

Abstract: One or more embodiments provide for a method. The method includes receiving a document including text. The method also includes receiving different definitions of a position in the document of an identifier-value pair within the text. The identifier-value pair includes an identifier identifying an information type in the text and a value including an entry of the information in the text. The method also includes determining which of the different definitions correctly point to the position to form a number of correct definitions. Each of the different definitions correctly points to the position when the identifier-value pair is found on a first attempt using a definition of the different definitions to search for the identifier-value pair. The method also includes, responsive to the number of correct definitions at least matching a threshold value, issuing a command to perform data scraping on the document to extract the value of the identifier.