Interactive Mapping Specification with Exemplar Tuples

Bonifati, Angela; Comignani, Ugo; Coquery, Emmanuel; Thion, Romuald

doi:10.1145/3035918.3064028

Cited by 21 publications

(5 citation statements)

References 33 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We omit here the precise definition, but we point out that the GPS normal form has the following main features: it satisfies certain desirable minimality criteria, is unique up to renaming variables, and is used to define semantics of queries in data exchange, so that the semantics does not depend on the syntactic representation of the underlying schema mapping. We also note that schema mappings produced in the aforementioned Interactive Mapping Specification framework [16] are in GPS normal form. From the proof of Theorem 13 (and, more specifically, from Claim 15 and the use of critically sound constraints in the learning algorithm), it follows that the GAV mapping returned by GAVLearn is guaranteed to be in GPS normal form (we do not give the details for lack of space).…”

Section: From Claim 15 We Obtainmentioning

confidence: 99%

“…Our experimental evaluation of GAVLearn is carried out using iBench, which is a metadata generator capable of creating schema mappings of varying complexity [10]. We note that iBench was also used to evaluate the Interactive Mapping Specification (IMS) tool developed in [16] and mentioned in the Introduction of the present paper. We use iBench to create both goal schema mappings and universal data examples.…”

Section: Experimental Evaluationmentioning

confidence: 99%

“…More recently, data examples have been used as the basis of alternative approaches to schema-mapping design [1,[16][17][18]24]. Here, a data example is a pair (I, J ) consisting of a source instance Session: Learning and Streaming PODS'18, June 10-15, 2018, Houston, TX, USA I and a target instance J .…”

Section: Introductionmentioning

confidence: 99%

“…The algorithms in the fitting framework and the learning framework involve interaction with the user or the learner. More recently, a different framework, called Interactive Mapping Specification (IMS), was developed in [16]. Starting with an initial set E of data examples, IMS derives a GLAV mapping M through a sequence of interactions with a non-expert user.…”

Section: Introductionmentioning

confidence: 99%

“…Concerning tools, there are systems for schema-mapping discovery of GLAV mappings in the fitting and in the IMS frameworks [2,16]; there is also a system for producing approximately optimal schema mappings in the repair framework, but this system only covers the very restricted class of single-head LAV mappings [18]. So far, there are no tools for schema-mapping discovery in the learning framework.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Active Learning of GAV Schema Mappings

Cate

Kolaitis

Qian

et al. 2018

Proceedings of the 37th ACM SIGMOD-SIGACT-SIGAI Symposium on Principles of Database Systems

View full text Add to dashboard Cite

Schema mappings are syntactic specifications of the relationship between two database schemas, typically called the source schema and the target schema. They have been used extensively in formalizing and analyzing data inter-operability tasks, especially data exchange and data integration. There is a growing body of research on deriving schema mappings from data examples, that is, pairs of source and target instances that depict the behavior of the unknown schema mapping. One of the approaches used in this endeavor casts the derivation of a schema mapping from data examples as a learning problem. Earlier work has shown that GAV mappings (global-as-view schema mappings) are learnable in Angluin's model of exact learning with membership queries and equivalence queries. Here, we validate the practical applicability of this theoretical result by designing and implementing an active learning algorithm, called GAV-Learn that derives a syntactic specification of a GAV mapping from a given set of data examples and from a "black-box" implementation. We analyze the properties of GAV-Learn and, among other results, we show that it produces a GAV mapping that has minimal size and is a good approximation of the unknown GAV mapping. Furthermore, we carry out a detailed experimental evaluation that demonstrates the effectiveness of GAV-Learn along different metrics. In particular, we compare GAV-Learn with two earlier approaches for deriving GAV mappings from data examples, and establish that it performs significantly better than the two baselines.

show abstract

Section: From Claim 15 We Obtainmentioning

confidence: 99%

Section: Experimental Evaluationmentioning

confidence: 99%