The present invention encompasses computerized methods and systems for finding and ranking domain-specific entities in a text-based inquiry. Existing text data sources are analyzed to create a training set of data comprising ranked entities with associated weighted features within each of a plurality of domains. The training set of data for each of the plurality of domains is stored in a ranked entities database. A text-based inquiry may be received from a user and analyzed, within a specific one of the domains, to locate candidate words or phrases based at least in part on the ranked entities in the specific domain. The candidate words or phrases can then be ranked based on the features for corresponding entities from the training set. The ranked candidate words or phrases can be stored in a ranked candidate database for further use by at least one of an analyst or a recommendation engine.