Details
| Originalsprache | Englisch |
|---|---|
| Titel des Sammelwerks | SIGMOD '19 |
| Untertitel | Proceedings of the 2019 International Conference on Management of Data |
| Erscheinungsort | New York |
| Herausgeber (Verlag) | Association for Computing Machinery (ACM) |
| Seiten | 865-882 |
| Seitenumfang | 18 |
| ISBN (elektronisch) | 9781450356435 |
| Publikationsstatus | Veröffentlicht - 25 Juni 2019 |
| Veranstaltung | SIGMOD/PODS '19: International Conference on Management of Data - Amsterdam, Niederlande Dauer: 30 Juni 2019 → 5 Juli 2019 |
Abstract
Detecting erroneous values is a key step in data cleaning. Error detection algorithms usually require a user to provide input configurations in the form of rules or statistical parameters. However, providing a complete, yet correct, set of configurations for each new dataset is not trivial, as the user has to know about both the dataset and the error detection algorithms upfront. In this paper, we present Raha, a new configuration-free error detection system. By generating a limited number of configurations for error detection algorithms that cover various types of data errors, we can generate an expressive feature vector for each tuple value. Leveraging these feature vectors, we propose a novel sampling and classification scheme that effectively chooses the most representative values for training. Furthermore, our system can exploit historical data to filter out irrelevant error detection algorithms and configurations. In our experiments, Raha outperforms the state-of-the-art error detection techniques with no more than 20 labeled tuples on each dataset.
ASJC Scopus Sachgebiete
- Informatik (insg.)
- Software
- Informatik (insg.)
- Information systems
Zitieren
- Standard
- Harvard
- Apa
- Vancouver
- BibTex
- RIS
SIGMOD '19: Proceedings of the 2019 International Conference on Management of Data. New York: Association for Computing Machinery (ACM), 2019. S. 865-882.
Publikation: Beitrag in Buch/Bericht/Sammelwerk/Konferenzband › Aufsatz in Konferenzband › Forschung › Peer-Review
}
TY - GEN
T1 - Raha
T2 - SIGMOD/PODS '19
AU - Abedjan, Ziawasch
AU - Mahdavi, Mohammad
AU - Fernandez, Raul Castro
AU - Madden, Samuel Ross
AU - Quzzani, Mourad
AU - Stonebraker, M.R.
AU - Tang, Nan
N1 - Funding information: This project has been supported by the German Research Foundation (DFG) under grant agreement 387872445.
PY - 2019/6/25
Y1 - 2019/6/25
N2 - Detecting erroneous values is a key step in data cleaning. Error detection algorithms usually require a user to provide input configurations in the form of rules or statistical parameters. However, providing a complete, yet correct, set of configurations for each new dataset is not trivial, as the user has to know about both the dataset and the error detection algorithms upfront. In this paper, we present Raha, a new configuration-free error detection system. By generating a limited number of configurations for error detection algorithms that cover various types of data errors, we can generate an expressive feature vector for each tuple value. Leveraging these feature vectors, we propose a novel sampling and classification scheme that effectively chooses the most representative values for training. Furthermore, our system can exploit historical data to filter out irrelevant error detection algorithms and configurations. In our experiments, Raha outperforms the state-of-the-art error detection techniques with no more than 20 labeled tuples on each dataset.
AB - Detecting erroneous values is a key step in data cleaning. Error detection algorithms usually require a user to provide input configurations in the form of rules or statistical parameters. However, providing a complete, yet correct, set of configurations for each new dataset is not trivial, as the user has to know about both the dataset and the error detection algorithms upfront. In this paper, we present Raha, a new configuration-free error detection system. By generating a limited number of configurations for error detection algorithms that cover various types of data errors, we can generate an expressive feature vector for each tuple value. Leveraging these feature vectors, we propose a novel sampling and classification scheme that effectively chooses the most representative values for training. Furthermore, our system can exploit historical data to filter out irrelevant error detection algorithms and configurations. In our experiments, Raha outperforms the state-of-the-art error detection techniques with no more than 20 labeled tuples on each dataset.
UR - http://www.scopus.com/inward/record.url?scp=85069437614&partnerID=8YFLogxK
U2 - 10.1145/3299869.3324956
DO - 10.1145/3299869.3324956
M3 - Conference contribution
SP - 865
EP - 882
BT - SIGMOD '19
PB - Association for Computing Machinery (ACM)
CY - New York
Y2 - 30 June 2019 through 5 July 2019
ER -