Ιδρυματικό Αποθετήριο
Πολυτεχνείο Κρήτης
Πολυτεχνείο Κρήτης
EN
|
EL
| URI: | http://purl.tuc.gr/dl/dias/9AC9C1B2-435D-4E49-971E-4D808FF2D7C6 | ||
| Έτος | 2010 | ||
| Τύπος | Πλήρης Δημοσίευση σε Συνέδριο | ||
| Άδεια Χρήσης |
|
||
| Βιβλιογραφική Αναφορά | D. Z. Wang, M. J. Franklin, M. Garofalakis and J. M. Hellerstein, "Querying probabilistic information extraction", in 36th International Conference on Very Large Data Bases, 2010. https://doi.org/10.14778/1920841.1920974 | ||
| Εμφανίζεται στις Συλλογές |
Recently, there has been increasing interest in extending relationalquery processing to include data obtained from unstructured sources.A common approach is to use stand-alone Information Extraction(IE) techniques to identify and label entities within blocks of text;the resulting entities are then imported into a standard database andprocessed using relational queries. This two-part approach, however,suffers from two main drawbacks. First, IE is inherently probabilistic,but traditional query processing does not properly handleprobabilistic data, resulting in reduced answer quality. Second,performance inefficiencies arise due to the separation of IE fromquery processing. In this paper, we address these two problems bybuilding on an in-database implementation of a leading IE model—Conditional Random Fields using the Viterbi inference algorithm.We develop two different query approaches on top of this implementation.The first uses deterministic queries over maximumlikelihoodextractions, with optimizations to push the relational operatorsinto the Viterbi algorithm. The second extends the Viterbialgorithm to produce a set of possible extraction “worlds”, fromwhich we compute top-k probabilistic query answers. We describethese approaches and explore the trade-offs of efficiency and effectivenessbetween them using two datasets
Copyright © DIAS 2013
