Optimization of Queries Based on Foundational Ontologies
Author(s): Jana Ahmad, Petr Kremen, Martin Ledvinka
Full text: submitted version
Abstract: With increased global economic development, the volume of data has grown steadily over the past decade. Thus, managing, answering user’s queries and retrieving data from multiple data sources could be a significant challenge for any enterprise. Thus this paper discusses, how proper foundational-ontology-based index design can help in optimize query answering. For this purpose, we discuss our idea on the unified foundational ontology and a domain ontology based on it. We discuss several(Unified Foundational Ontology) UFO-specific indexing techniques and compare their suitabilities for different UFO-based queries.
Keywords: Foundational Ontologies; SPARQl; Indexing
Review 1 (by Valerio Basile)
(RELEVANCE TO ESWC) This is a paper about (foundational?) ontologies and triple stores, so I guess it falls in the areas of interest of ESWC. (NOVELTY OF THE PROPOSED SOLUTION) As the authors point out, there has been significant amount of work on how to efficiently store RDF triples and how to handle their retrieval. It is not clear what is supposed to be novel here, apart from applying existing techniques to one particular ontology. (CORRECTNESS AND COMPLETENESS OF THE PROPOSED SOLUTION) I think I understand the general idea underlying the paper, i.e., different low-level representations of RDF triples have an impact on the queries one may efficiently make on them. However, the "approach" is presented in such a confusing way that it is hard t draw any conclusion. (EVALUATION OF THE STATE-OF-THE-ART) No comparison is provided with any other solution. (DEMONSTRATION AND DISCUSSION OF THE PROPERTIES OF THE PROPOSED APPROACH) In addition to the other comments I made, the relevance of this work seems to be quite narrow. It proposes a way of optimizing the queries made only on ontologies that make use of UFO, while there is no mention of what would be the additional costs of maintaining three different stores for the same triple set. (REPRODUCIBILITY AND GENERALITY OF THE EXPERIMENTAL STUDY) There is next to no experimental study, only a few examples with no results. (OVERALL SCORE) Most of the sentences are ungrammatical. Those who are not, are still quite confusing. I wanted to give a chance to this paper so I read through it anyway. From what I understand, the main goal of this paper is to evaluate methods of querying ontologies that are built on top of UFO, in a way optimizing the queries by following the constraints of UFO and implement them at a low level in the triple store. Besides the presentation being seriously lacking, the evaluation consists of a few examples of queries with judgments like "the system should easily infer...", with no experimental evidence at all. No figures are provided, nor an acctual experiment for that matters. The only thing I learned from this paper is that there are at least three ways of representing UFO triples in a database, and you may choose one or another based on what property you want to query. This is hardly sufficient for a research paper. -- After reading the authors' rebuttal, I will keep my scores. The authors only briefly addressed my point about novelty, and I am still not convinced.
Review 2 (by anonymous reviewer)
(RELEVANCE TO ESWC) The paper proposes to use foundational ontologies as a tool to optimize SPAQL query answering over RDF data sets. The idea is that the knowledge stored in an ontology could be used make query answering more intelligent and efficient. Unfortunately, the proposal in the paper is very vague and informal. There is vast literature on ontology-based data access (OBDA), and this goes way beyond the cited paper . It seems to me the basic ideas of the submission have already been explored in the OBDA literature. It was very difficult to read this paper. The paper clearly needs a basic linguistic/grammar revision. It is clearly not ready for publication (e.g., see the last paragraph in page 5 or the sentence "Then, for example if we have the query: what objects participate in Action-i? (Q2)."). (NOVELTY OF THE PROPOSED SOLUTION) The paper lacks formal details on the proposal. Judging the novelty is thus very difficult. (CORRECTNESS AND COMPLETENESS OF THE PROPOSED SOLUTION) The paper lacks formal details on the proposal. It is impossible to judge the correctness based on the provided details. (EVALUATION OF THE STATE-OF-THE-ART) As mentioned, OBDA literature is extensive, but is mostly ignored by the authors. The authors should provide an extensive discussion on how their approach is related to OBDA. (DEMONSTRATION AND DISCUSSION OF THE PROPERTIES OF THE PROPOSED APPROACH) There little formal details in the paper, and there are no experiments. (REPRODUCIBILITY AND GENERALITY OF THE EXPERIMENTAL STUDY) From the few details given in the paper it is impossible to judge the possible practical implications. I don't see any experimental study on the performance gains based on the proposed approach. (OVERALL SCORE) The paper is clearly not ready for publication at ESWC. ======ADDED AFTER AUTHOR RESPONSE======== The response does not change my evaluation of the paper. In my review I mentioned Ontology-based Data Access, which is only briefly mentioned in the paper, but should receive much more attention. I encourage the authors to look it, e.g., to results on semantic indexing (Calvanese&Rodriguez-Muro). In general, the paper needs to be improved by adding formal details.
Review 3 (by Kuldeep Singh)
(RELEVANCE TO ESWC) The paper is in the direction of SPARQL query optimization, and a relevant fit for the conference. (NOVELTY OF THE PROPOSED SOLUTION) The paper proposes UFO-based indexing techniques for query optimization and aim towards improving the efficiency of queries based on foundational ontologies. (CORRECTNESS AND COMPLETENESS OF THE PROPOSED SOLUTION) The approach has been illustrated in the paper using competence questions describe in (1). All the questions has been addressed properly in the paper with an explaination. 1) Grninger, M. Mark S.: Methodology for the Design and Evaluation of Ontologies (1995). (EVALUATION OF THE STATE-OF-THE-ART) The approach has been evaluated using the created benchmark. Few things are not clear: 1. Besides the authors, what are the other work which have been evaluated using such competence questions? I could not find relevant references in the paper. 2. Why UFO based query optimization? Why not some other ontology-based query optimization, What I am missing since beginning of the paper is clear motivation of choosing UFO based indexing technique. Also, what are the parameters for optimisation-> is optimization also affected by execution time? If not, then what is the matric that affects query optimization. I am also missing the related work for query optimization in general and how the current work position itself with other approaches. (DEMONSTRATION AND DISCUSSION OF THE PROPERTIES OF THE PROPOSED APPROACH) The current approach has been well illustrated using competence questions. (REPRODUCIBILITY AND GENERALITY OF THE EXPERIMENTAL STUDY) The approach seems generic, and can be applied to other ontologies besides the discussed Aviation ontology as case study. (OVERALL SCORE) The paper proposes UFO-based indexing techniques for query optimization and aim towards improving the efficiency of queries based on foundational ontologies. It is well supported by evaluation using competence questions, and benchmark and aviation ontology is used as case study. +ves: The illustration of the approach using the set of questions and detailed description of how each competence question is addressed. -ves: Images can be much better in resolution. Image 1 is very unclear. For a non-expert user, it is hard to make sense of this image. Missing related work of query optimization, how the problem has been tackled in the research field. The motivation behind the work, why UFO based optimization is chosen, why not some other approach. Questions to Authors: 1. How will authors position their paper wrt. state-of-the-art techniques for query optimization? How their approach is different/novel from others. 2. Besides evaluating approach using competence question, is there any other evaluation of the approach possible?
Metareview by Hsofia Pinto
The paper discusses a unified foundational ontology and a domain ontology based on it and UFO-specific indexing techniques and compare their suitabilities for different UFO-based queries. While a relevant topic for the ESWC community it provides few details and no experiments. Even after rebuttal comments from the authors, the reviewers evaluated the contribution of this paper as falling short from the acceptance criteria for this conference. Authors are encouraged to further work on their approach based on review comments.