The purpose of this work is threefold: (i) to facilitate knowledge discovery in art historical photo archives, (ii) to support users’ decision-making process when evaluating contradictory artwork attributions, and (iii) to provide policies for information quality improvement in art historical photo archives. The approach is to leverage Semantic Web technologies in order to aggregate, assess, and recommend the most documented authorship attributions. In particular, findings of this work offer art historians an aid for retrieving relevant sources, assessing textual authoritativeness (i.e. internal grounds) of sources of attribution, and evaluating cognitive authoritativeness of cited scholars. At the same time, the retrieval process allows art historical data providers to define a low-cost data integration process to update and enrich their collection data. The contributions of this thesis are the following: (1) a methodology for representing questionable information by means of ontologies; (2) a conceptual framework of Information Quality measures addressing dimensions of textual and cognitive authoritativeness characterising art historical data, (3) a number of policies for metadata quality improvement in art historical photo archives as derived from the application of the framework, (4) a ranking model leveraging the conceptual framework, (5) a semantic crawler, called mAuth, that harvests authorship attributions in the Web of Data, and (6) an API and a Web Application to serve information to applications and final users for consuming data. Despite findings are limited to a restricted number of photo archives and datasets, the research impacts on a broader number of stakeholders, such as archives, museums, and libraries, which can reuse the conceptual framework for assessing questionable information, mutatis mutandi, to other near fields in the Humanities.