KRDB PhD Students Workshop
For more information, please contact:
lubyte at inf.unibz.it
(+39) 0471 016 226
keet at inf.unibz.it
(+39) 0471 016 128
After the workshop we will have a dinner (more information will follow soon).
||A Proof Theory for DL-Lite
||Complexity of Reasoning in Entity Relationship Models
||Extracting Ontologies from Relational Databases
||Prospects for and issues with mapping the Object-Role Modeling language into DLRifd
||Expressing DL-Lite Ontologies with Controlled English
||From Conjunctive Queries to Text Plans
||An Extension of DIG 2.0 for Handling Bulk Data
||Semantic Personalization of Web Portal Contents
A Proof Theory for DL-Lite.
In this work we propose an alternative approach to inference in DL-Lite,
based on a reduction to reasoning in an extension of function-free Horn
Logic (EHL). We develop a calculus for EHL and prove its soundness and completeness.
We will also show how to achieve decidability by means of a specific
strategy, and how alternative strategies can lead to improved results in specific
cases. On the one hand, we will mimic the query-answering technique based on
first computing a rewriting and then evaluating it. On the other hand, we discuss a
strategy that allows one to anticipate the grounding of atoms, and that might lead
to better performance in the case where the size of the TBox is not dominated by
the size of the data.
Complexity of Reasoning in Entity Relationship Models.
In this work we investigate the complexity of reasoning over various
fragments of the Extended Entity Relationship (EER) language, which include
different combinations of the constructs for isa between concepts and relationships,
disjointness, covering, cardinality constraints, including their refinement.
Specifically, we show that reasoning over ER diagrams with I S A between relationships
is EXPTIME-hard, even when we drop relationship covering. Surprisingly, when we drop also
ISA between relations, reasoning becomes NP-complete. If we further remove boolean constructs,
reasoning becomes polynomial. Our lower-bound results are established through direct reductions, while
the upper-bounds follow from correspondences with expressive variants of the
Extracting Ontologies from Relational Databases.
The use of a conceptual model (or an ontology) to describe relational data sources has been proved
to be extremely useful to overcome many important data access problems. However, the task of wrapping relational data sources by means of an ontology is mainly done manually. In this paper we introduce
an automatic procedure for extracting a conceptual view from a relational database. The semantic
mapping between the database schema and its conceptualisation is captured by associating views
over the data source to elements of the extracted conceptual model. To represent the conceptual
model we use an ontology language, rather that a graphical notation, in order to provide a precise
formal semantics. In particular we adopt a variant of the DLR-Lite description logic because of its
nice computational properties, and ability to express the mostly used modelling constraints. In order
to uncover the connections between relational schema and the conceptual model, the heuristics
underlying the ontology extraction process are based on ideas of standard relational schema design
and normalisation. In fact, we assume that the relational source is in third normal form. Under this
assumption we can formally prove that the conversion preserves the semantics of the constraints in
the relational database. Therefore, there is no data loss, and the extracted model constitutes a
faithful wrapper of the relational database.
Prospects for and issues with mapping the Object-Role Modeling language into DLRifd.
Object-Role modellers miss the advantages of automated reasoning over their ORM conceptual models,
which could be addressed by DL reasoners. DLs are not considered user-friendly and could benefit
from the easy to use ORM diagrammatic and verbalization interfaces and modelling methodologies.
Relating the two would greatly expand the scope for automated reasoning with additional scenarios
to improve quality of software systems. Given that none of the extant DL languages are as expressive
as ORM or its successor ORM2, the ‘best-fit’ DLRifd was chosen to map the formal conceptual modelling
language ORM2. For the non-mappable constraints, pointers to other DL languages are provided,
which could serve as impetus for research into DL language extensions or interoperability between
existing DL languages.
Expressing DL-Lite Ontologies with Controlled English.
In this work we deal with the problem of providing natural
language front-ends to databases upon which an ontology layer has been
added. Specifically, we are interested in expressing ontologies formalized
in Description Logics in a controlled language, i.e., a fragment of natural
language tailored to compositionally translate into a knowledge representation (KR) language.
As KR language we have chosen DL-Lite R , a
representative of the well-known DL-Lite family, and we aim at understanding the kind of
English constructs the controlled language can and
cannot have to correspond to DL-Lite R . Hence, we compare the expressive power of
DL-Lite R to that of various fragments of English studied
by I. Pratt and A. Third, which compositionally translate into fragments
of first order logic. Our analysis shows that DL-Lite R , though polynomial, is incomparable
in expressive power with respect to intractable
fragments of English. Interestingly, it allows one to represent a restricted
form of relative clauses, which lead to intractability when used without
From Conjunctive Queries to Text Plans.
In this talk I will present our efforts in terms of building a bridge between an Intelligent
Query Interface and Natural Language Generation technologies. The current version of our query
interface enables users to access data sources by means of an ontology representing the knowledge
of a domain in a well defined formal semantics. The main challenge we are facing now is that the
underlying conjunctive query is to be presented to the user in natural language. I will explain
the steps needed to translate a conjunctive query into a text plan, focusing primarily on the
problem of maximizing the local referential coherence of the natural language query that will
be generated given the text plan. This is done by seeing this problem as a topological sort of
a directed acyclic tree (the query), more precisely finding the linear ordering of the predicates
that will maximize the local coherence and therefore the readability of the generated text,
according to the coherence measures offered by Centering Theory.
An Extension of DIG 2.0 for Handling Bulk Data.
The research community has noted the need to retrieve the
instance level of an ontology from bulk data stored in external data
sources (e.g., a relational database), and to delegate to the external
source all aspects of the actual management of the data. To achieve this,
several methodologies have been recently developed to represent and reason about what
we call here Ontologies with Linking Axioms. However,
existing DL reasoners cannot properly deal with such ontologies. Indeed,
including the instance level in the communication with a DL reasoner
can be a heavy burden on the communication line, and goes against the
requirement of delegating data management to the external source. To
overcome these problems, we present here an extension to the DIG 2.0
Interface that allows for the specification and management of Ontologies
with Linking Axioms. The extension is a general framework which can
accommodate any type of data source and linking axiom through specific
implementations. We present one such specific implementation aiming at
representing axioms linking RDBMS data sources to ontologies handled
Semantic Personalization of Web Portal Contents.
Enriching Web applications with personalized data is of major
interest for facilitating the user access to the published contents,
and therefore, for guaranteeing successful user navigation. We
propose a conceptual model for extracting personalized
recommendations based on user profiling, ontological domain
models, and semantic reasoning. The approach offers a high-level
representation of the designed application based on a domain-
specific metamodel for Web applications called WebML.