Re: RDB2RDF mapping

Hi, guys! Thank you for the response. I know, that questions are a bit
difficult to understand, and believe me, they are also difficult to
formulate in one sentence.
I will try to response some of your doubts:

1. Both ETL and on-demand data access must be implemented - the statement
is, that the best practice is to implement both of them, in order to
increase the possible ways to use the dataset.
2. In case of automated mapping, the direct mapping requirements must be
satisfied; in case of manual mapping, it must be represented in R2RML - if
you are absolutely agree, please choose the option in the form =)
3. To improve the quality, explicitly related data from the database should
be implicitly incorporated into the mapping. Often, not only
primaryKey-foreignKey relations is hidden in the DB model. In our case, for
example, was such a relation:

Presentation1-has version-version1
Presentation1-has version-version2
Presentation1-has language-language1

Implicitly, this also means (in our case), that version1 and version2 has
language1. (Hope, my syntax is understandable)

Q4 - the queries in the application domain outside the r2rml context.
Q6 - remember please, that the proposed best practices will be used to
evaluate the quality of mapping. So, the question is, what do you
personally think: the reuse of existing ontologies increases, decreases or
does not influende the quality of mapping in general?


Mit freundlichen Grüßen,
Kind regards,

Darya Tarasowa

Phd student

Enterprise Information Systems Department, University of Bonn,

Fraunhofer IAIS <http://www.iais.fraunhofer.de/>  <http://aksw.org/>

--




2013/12/13 Juan Sequeda <juanfederico@gmail.com>

> Darya,
>
> I was about to write an email with these exact comments. Ashok, beat me to
> it. I have the same concerns.
>
> Some other comments:
>
> Q4: Which complicated queries? Used in an application outside the RDB2RDF
> context? Or a complicated query used in the mapping. If it's the latter,
> isn't the query already in the R2RML mapping file?
>
> Q6: (reusing mappings) This is a common topic in the SW community. The
> answer is "depends". In a commercial setting (at least in my experience),
> users create their own target ontology and don't reuse. Should they reuse?
> If they want to be 100% in control, then no.
>
> You may also found of interest our following paper:
>
> Relational Database to RDF Mapping Patterns
> http://ceur-ws.org/Vol-929/paper9.pdf
>
> Best,
>
> Juan Sequeda
> +1-575-SEQ-UEDA
> www.juansequeda.com
>
>
> On Fri, Dec 13, 2013 at 9:09 AM, Ashok Malhotra <ashok.malhotra@oracle.com
> > wrote:
>
>>  Hi Darya:
>> I have a few comments on your questionnaire.
>> The problem that I have with it is that the mapping depends
>> on what you want to do with it.  For example, your first question is
>> "Both ETL and on-demand data access must be implemented".  Well,
>> that depends on how you want to use the data.
>>
>> The second question is "In case of automated mapping, the direct mapping
>> requirements must be satisfied; in case of manual mapping, it must be
>> represented in R2RML."  Well, yes, of course.
>>
>> The third question is "To improve the quality, explicitly related data
>> from the database should be implicitly incorporated into the mapping."   I
>> don't understand what this means.  What kind of explicit
>> relationships, other than primaryKey-foreignKey are you thinking about.
>>
>> Are you trying to write a Best Practices in Mapping Relational Data to
>> RDF?
>> All the best, Ashok
>>
>>  On 12/13/2013 8:15 AM, Darya Tarasowa wrote:
>>
>>  Hi,
>>
>>  My name is Darya Tarasova and I'm a 3rd-year PhD student at University
>> of Bonn.
>>
>> Currently I'm working on proposing the best practices in creating the
>> RDB2RDF mappings and I need community feedback on my investigations.
>>
>> Could you please fill out the form, available on http://slidewiki.org/application/questionnaire.php
>> . The questionnaire consists of about 15 questions and should not take
>> more than 5-10 minutes of your time.
>>
>> Answering the questionnaire will help as a lot, as we do not have any
>> other way to evaluate our work. You can choose either to stay incognito or
>> to fill out your personal data to be personally thanked in the
>> Acknowledgement section of the future paper.
>>
>> Please, share the link with anybody, who could help us!
>>
>>  Mit freundlichen Grüßen,
>> Kind regards,
>>
>> Darya Tarasowa
>>
>> Phd student
>>
>> Enterprise Information Systems Department, University of Bonn,
>>
>> Fraunhofer IAIS <http://www.iais.fraunhofer.de/>
>>
>> --
>>
>>
>>
>>
>