Onestop App

Περιγραφή

Weighting schemes are used in both Entity Resolution and Geospatial Interlinking, as a means of quickly and a-priori determining the pairs of records and geometries, respectively, that are more likely to satisfy a matching or topological relation. In a nutshell, all candidate pairs are sorted in a decreasing score such that the processing is focused on the K top-weighted, where K is determined by the available budget of the underlying application (e.g., the run-time that a cloud infrastructure is used or the number of times a function is called upon charge like the Lambda functions in AWS cloud). There is a plethora of weighting schemes for both integration tasks, with no clear winner among them. The goal of this work is to develop novel rank aggregation schemes that unify the sorting order of different weighting schemes into a single sorting list of pairs that achieves much higher recall and precision than the individual ones. Any rank aggregation library in Python can be used for this purpose, e.g., ranky and pyrankagg. The use of graphons should be exploited in these approaches.

References:
George Papadakis, Georgia Koutrika, Themis Palpanas, Wolfgang Nejdl: Meta-Blocking: Taking Entity Resolution to the Next Level. IEEE Trans. Knowl. Data Eng. 26(8): 1946-1960 (2014)
Luca Gagliardelli, George Papadakis, Giovanni Simonini, Sonia Bergamaschi, Themis Palpanas: Generalized Supervised Meta-blocking. Proc. VLDB Endow. 15(9): 1902-1910 (2022)
George Papadakis, Georgios M. Mandilaras, Nikos Mamoulis, Manolis Koubarakis: Progressive, Holistic Geospatial Interlinking. WWW 2021: 833-844
Georgios M. Mandilaras, Nikos Mamoulis, Manolis Koubarakis: Static and dynamic progressive geospatial interlinking. ACM TSAS (2022)

Στοιχεία Εργασίας