Map-On: A web-based editor for visual ontology mapping

Tracking #: 1266-2478

Alvaro Sicilia
German Nemirovski1
Andreas Nolle1

Responsible editor: 
Krzysztof Janowicz

Submission type: 
Tool/System Report
This paper presents Map-On, a web-based editor for visual ontology mapping developed at the Architecture, Representation and Computation research group of La Salle, Ramon Llull University. The Map-On editor provides a graphical environment for the ontology mapping creation using an interactive graph layout. A point-and-click interface simplifies the mapping creation process. The editor automatically generates a R2RML document based on user inputs, particularly producing IRI patterns and SQL queries. It has been used in real scenarios alleviating the effort of coding R2RML statements which is one of the main barriers for adopting R2RML in research and in the industry communities.
Full PDF Version: 


Solicited Reviews:
Click to Expand/Collapse
Review #1
By Valerio Santarelli submitted on 22/Feb/2016
Minor Revision
Review Comment:

The authors present Map-On, a web-based graphical mapping editor for creating R2RML mappings from ontologies to SQL databases. Map-On uses graphs to represent
the ontology, the database and the mappings, and allows the user to map elements of the ontology to the database tables, thus automatically creating R2RML mappings.
One of the critical aspects of OBDA is the difficulty of generating the mappings, as it requires extensive background knowledge, and is in general a time-consuming and error-prone task if performed manually.
The authors provide a description of the features and interface of the tool, of its technical architecture, and provide a step-by-step overview of the process of creating a mapping with Map-On.
The authors also present an introduction to R2RML mappings, a fairly extensive overview of other mapping editors or visualization tools, a user evaluation of Map-On, and a real-world project in which the tool was used successfully.
The tool has been made accessible through a custom guest read-only account.

The revised submission indeed addresses the aspects which were in my opinion most critical in the previous version, i.e., the lack of a user evaluation test, which is necessary to back up the claim that Map-On is a useful tool even for users not skilled in ontology and mapping design to generate ontology mappings, and the overall poor quality of clarity and readability of the paper.

Indeed, the authors present a user evaluation in which 5 users with no background knowledge of the Semantic Web or of OBDA are introduced to these subjects and to the tool, and are asked to perform three mapping tasks for a benchmark scenario. I have some experience with user evaluation tests on semantic-web related aspects, and I appreciate the difficulty of organizing such tests, so I commend the authors for doing so.
However, the validity of the test itself could be subject to some discussion, since the users were given a very brief introduction to the Semantic Web, ontologies, RDF, an R2RML mapping, the Map-On tool (each test took about 50 minutes, and the three tasks took more than 30 minutes total), so I'm not sure that the results of the test are very meaningful.
Specifically, the authors do not discuss the results of Task 2, which seems the most interesting, since average time was vastly superior to the other tasks. I would like to see at least some comment in this direction, possibly also addressing the user reaction when a proposed task took so long (was the user frustrated?).
More in general, I'm not sure there is enough evidence in the presented results to back the final claims made by the authors that the tool was successfull in achieving the desired goal during the tests. A more detailed discussion of the results would be appreciated.

The quality of the writing has been fairly improved throughout the paper (see minor comments below), with the notable exception of the new User testing section, which is quite poor. In fact, in some points it looks more like a rough draft, e.g., "Participant had a similar profile was; they were graduates and post-graduates" or "of languages such as OWL, RDF, SPARQL, and R2RML; either." or "The screen was recorded and the voice of the participants who were beforehand informed". Also, there is a discrepancy between one of the results reported in the text and in the tables: in Figure7, the completion time for Task 3 seems to be 3 or 4 minutes, while the paragraph above says 7 minutes.
I think a proof-read by a native english-speaking person of this section is necessary before the paper can be published.

Minor comments:
1)pg.1: The queries can be processed with reasoning purposes..... -> The queries can be processed over the ontology for reasoning purposes in order to allow using hidden relations....
2)pg.1: Experts require from knowledge on the data schemas -> Experts are required to have knowledge of the data schemas
3)pg.1: creating R2RML mappings requires from advanced skills -> creating R2RML mappings requires advanced skills
4)pg.2: Indeed, and given a fixed screen space -> Indeed, given a fixed screen space
5)pg.2: and the most common used -> and the most commonly used
6)pg.3: The mapping editor presented in this paper, supports the creation -> The mapping editor presented in this paper supports the creation
7)pg.6: the mapping space where mapping belongs to. -> the mapping space to which the mapping belongs.
8)pg.7: the link to the tool is wrong, it should point to Map-On, not Click-On
9)pg.8: One person has been administering the test -> The test was administered by a single person.
10)pg.8: is easy to be understood by the -> is easy to understand for the
11)pg.8: Both, the ontology -> Both the ontology
12)pg.9: The user satisfaction was measured with -> The user satisfaction was measured through
13)pg.9: were the participants rated some -> where the participants rated some
14)pg.9: to the interface raised during the test -> to the interface were raised during the test

Review #2
By Daniel Faria submitted on 21/Mar/2016
Review Comment:

The paper describes Map-On, a web-tool for editing and visualizing mappings between relational databases and ontologies.

In comparison with the previous version of the manuscript, the authors added:
- a few sentences discussing the choice between graph and tree layout for representing ontologies
- a comparison between existing tools summarized in a table
- a section detailing a usability test of their tool

These changes address my main concerns with the original version of the manuscript, and greatly contribute to improving it.
Ideally, the usability test should have been comparative, to demonstrate the superiority of Map-On over one of the tools adopting a tree layout. However, I concede that performing a comparative usability test involving a third-party tool is far from trivial, and thus that the usability test the authors carried out is quite reasonable.
On the whole, I am convinced about the quality and importance of Map-On.
As for the paper itself, it is on the whole quite clear and easy to read.
The discussion between graph and tree layout for representing ontologies could perhaps be more streamlined, but again, this is not a substantial issue.
What is somewhat substantial are the grammar errors that were not corrected from the original version of the manuscript or have since been introduced (namely a few erroneous uses of "from" in the Introduction). I think a grammar check of the document before it is published is paramount.

Review #3
By Amruta Nanavaty submitted on 19/Jun/2016
Review Comment:

This is a paper describing Map-On, a web-based graphical environment tool for mapping a database schema to a domain ontology. It supports different kinds of users to manually establish relations between the elements of a database and domain ontology in context of an Ontology-based Data Access (OBDA) scenario. The editor, based on the user’s input, automatically generates an R2RML document, produces Internationalized Resource Identifier (IRI) patterns and SQL queries. The authors describe the architecture and how the tool works very well. The automatic generation of R2RML mappings from the user actions and the graph based visualization layout seem to be unique features of Map-On. Please find below the major comments:

Most of the issues reported in the first review have been addressed in the latest paper submission. There is a lot of improvement in the paper compared to the last review. An entire new section on User testing has been added which includes the details of the setup, participant information, metrics used to evaluate the test and important observations (including limitations of the current Map-On tool). This additional information gives a boost to the paper as their "experience on a real scenario" section still does not cover much details about the tool's contribution. A summary table has also been added to highlight the features of Map-On tool with respect to the existing systems. However, explanation of the SQL query generation example mentioned in Section 4.3 "Automated generation of IRI patterns and SQL queries" is confusing. For example, it is difficult to understand how "Building concept - Id column" mapping is adjacent to "Age_class concept - Age_Class column" mapping with respect to the interactive graphical visualization shown in Figure 3.