Mirjana Mazuran: Grasping and using the knowledge extracted from data: application scenarios

Mirjana Mazuran will give a talk on Wednesday 10th October, at 11am. The talk
will take place in the Turing Building, Inria Saclay, in the room Flowers (https://team.inria.fr/cedar/contact/).

Nowadays, huge amounts of data are generated both daily and from different applications and sources. Being able to understand their semantics is of great importance to take advantage of them. Database constraints contribute to the definition of such semantics but, during the life of a database, they might be violated by data updates, integration with other datasets, etc. In the presence of constraint violation, the problem of re-establish consistency is usually faced by changing the data, however, systematic and frequent violations of a given constraint may suggest that the represented reality is changing and thus the constraint should evolve with it. During the talk I will discuss a method that focuses on Functional Dependencies in particular and whose aim is to (i) find the functional dependencies that are violated by the current data, and (ii) support their evolution when it is necessary to update them. Being aware of the current semantics of data and of its evolution is crucial also in the countless applications that rely on the use of derived information as building blocks for complex scenarios. I will present Mercurio, a system that supports the decision-making process of financial investors through the automatic extraction and analysis of financial data coming from the Web. Mercurio formalizes the knowledge and reasoning of an expert in financial journalism and enriches it by performing automatic analysis of financial data to identify relevant events related to the stock market. Then, sequential pattern mining is used to predict exceptional events on the basis of the knowledge of their past occurrences and relationships with other events, in order to to warn investors about them. In the last part of the talk I will present the IDEAA project, the topic of my MSCA Individual Fellowship, whose aim is to support the exploration of huge and heterogeneous RDF graphs in a user-friendly fashion and by making use of succinct and meaningful knowledge. I will conclude by discussing the advancements made so far in the project and the future research directions.

Mirjana Mazuran has obtained her Ms.C in Computer Engineering (2008) and the PhD in Information technology (2012) at Politecnico di Milano working on the extraction of tree-based association rules from XML documents and on their use for intensional query-answering purposes. From September 2010 to May 2011 she has been a visiting scholar at UCLA; from 2012 to 2017 she has been a post-doc at Politecnico di Milano; from January 2018 to May 2018 she has worked at Human Technopole, a research institute focused on data analysis. She is currently a post-doc at Inria Saclay Ile-de-France holding an MSCA Individual Fellowship. Her main research interests include data mining applications, RDF analytics and semi-automatic constraint updates.

Comments are closed.