Abstract: A system, method, and computer program product are provided for managing and retrieving metadata from an enterprise-wide database. An implementation receives a search query from a user. The implementation then generates an updated search query by at least providing the search query into a natural language processing engine. The implementation then retrieves a collection of metadata from a database by at least querying a search engine with the updated search query. The implementation then generates a collection of ranked metadata by at least providing, associated with the collection of metadata, a usage history of a table or attribute access by the user over a period of time into a ranking engine. The implementation then selects one or more top-ranked metadata of the collection of ranked metadata to generate a result of the updated search query.

Abstract: Disclosed herein are system, method, and computer program product embodiments for generating metadata element recommendations. For example, the method includes acquiring, by at least one processor and via a user interface, metadata associated with a data store. The method also includes performing natural language processing on the metadata to generate processed metadata, generating a candidate table name and a table description associated with the candidate table name for a table included in the metadata, and generating a first candidate attribute name, an attribute description associated with the first candidate attribute name, and a corresponding data type for each attribute associated with the table. The method also includes generating a second candidate attribute name for each attribute by extracting one or more keywords from the first candidate attribute name, and modifying the user interface to include at least the candidate table name.

Abstract: Disclosed are various embodiments for data anomaly detection. A variable profile is generated for each variable in source data. Then, the variable profiles are provided to each of a plurality of machine learning models. Next, it is determined, with each of the plurality of machine learning models, whether each variable profile is anomalous. The determination, from each of the plurality of machine learning models, whether each variable profile is anomalous is provided to an ensemble model. The ensemble model then generates a final determination whether each variable profile is anomalous. The final determination is then reported to an analysis service.

Abstract: Disclosed are various embodiments for data anomaly detection. A variable profile is generated for each variable in source data. Then, the variable profiles are provided to each of a plurality of machine learning models. Next, it is determined, with each of the plurality of machine learning models, whether each variable profile is anomalous. The determination, from each of the plurality of machine learning models, whether each variable profile is anomalous is provided to an ensemble model. The ensemble model then generates a final determination whether each variable profile is anomalous. The final determination is then reported to an analysis service.