Hi Marco,

I was thinking of using the evaluation benchmarks of the Ontology  
Alignment Evaluation Initiative [1]. One of my professors is actively  
involved in this community and has confirmed that the benchmarks do  
generalize well, i.e. a good matcher would also perform well for the  
freebase and dbpedia schemas. People also publish their matching  
algorithms.

A straightforward way of approaching the GSoC task would be to pick  
one up and experiment how well it actually performs in our setting.  
Then to incorporate own ideas such as the one you proposed [2] and  
identify the critical parts to be improved, which is what I meant by  
research oriented. What do you think?

Best regards,

Robert

[1] http://oaei.ontologymatching.org/
[2] http://wiki.knoesis.org/index.php/Property_Alignment

Zitat von Marco Fossati <hell.j....@gmail.com>:

> Hi Robert,
>
> On 3/8/15 8:20 PM, rlits...@mail.uni-mannheim.de wrote:
>> Hi Thiago, Hi DBPedia-Team,
>>
>> thanks for your reply. I'd like to clarify a fundamental question:
>>
>> - In the previous GSoC the participant seem to have built his own
>> goldstandard of mappings. Are standard benchmarks for quality
>> measurement insufficient
> Which standards are you thinking about? Could you reference them?
> , i.e. does the schema matching quality vary
>> and depend that much on the source schemas used?
>>
>> - In my opinion one could tackle this task either practically oriented
>> by implementing a promising approach, or research oriented by working
>> on an improvement of existing solutions. Which approach is likely the
>> way to go?
> For the scope of GSoC, I would advocate the former.
>>
>> I have done quite some research and gone through a few papers and I
>> think I have fair understanding. Are there any particular warm-up task
>> related to this task that you could suggest?
> Have a look at the unsolved issues on the specific repo of GSoC 2014 project:
> https://github.com/dbpedia/wikidata-mapper/issues
> Those tasks can be applied to Freebase schema too.
>>
>> Best regards
>>
>> Robert
>>
>> Zitat von Thiago Galery <tgal...@gmail.com>:
>>
>>> Hi Robert, I would advise taking a look at Marco's response to another
>>> prospective student. He points to these links for a summary of a similar
>>> project in 2014
>>>
>>>
>>> -idea: http://wiki.dbpedia.org/gsoc2014/ideas#h359-11
>>> -proposal:
>>> https://docs.google.com/document/d/16lAqKLAsAGQW0cp9SA0Egb1vlb6mPCcHYezVN-zB870/edit?pli=1
>>> -stuff
>>> <https://docs.google.com/document/d/16lAqKLAsAGQW0cp9SA0Egb1vlb6mPCcHYezVN-zB870/edit?pli=1-stuff>
>>> done:
>>> https://github.com/dbpedia/extraction-framework/wiki/GSoC-2014-Progress-Sergey-Skovorodkin
>>>
>>>
>>> On Fri, Mar 6, 2015 at 12:04 PM, <rlits...@mail.uni-mannheim.de> wrote:
>>>
>>>> Hello everybody,
>>>>
>>>> first off I'd like to introduce myself . I'm Robert, a current Masters
>>>> student at the Mannheim University. I'm studying Business Informatics
>>>> and pursuing
>>>> the Data and Web Science Specialization Track. One of my major
>>>> interests lies in
>>>> Data Mining and I constantly complement my studies with Data Mining
>>>> related online
>>>> courses (MOOCs) during my free time. Alongside my studies I'm also
>>>> employed as a
>>>> student researcher at the Data and Web Science research group [1]  
>>>> under the
>>>> supervision of Prof. Bizer. You will find many professors mentioned in
>>>> many of the
>>>> papers you suggest as a starting point. A major part of the research
>>>> is particularly
>>>> dedicated at Open Linked Data, hence the education is close-knit with
>>>> examples
>>>> and from research projects.
>>>>
>>>> Furthermore, during one of my previous internships I have been involed
>>>> in building
>>>> an Active Learning system for Named Entity Recognition which has also
>>>> enhanced my
>>>> experience within this field. The first time I got in touch with NLP
>>>> and Machine Learning
>>>> was during my Bachelor Thesis that concerned with the classification
>>>> of Scientific Papers.
>>>>
>>>> Now coming to the GSoC project:
>>>>
>>>> My first priority would be to work on "5.7. Reverse Engineering and
>>>> Aligning Freebase
>>>> with DBpedia." I have a working knowledge of Sparql and the Freebase
>>>> MQL query language
>>>> if needed. During my prior semester I have used DBPedia and Freebase
>>>> to perform web
>>>> data integration in a closed domain. So I'm aware of schema
>>>> integration and schema matching
>>>> procedures, which I think qualifies me along with my programming
>>>> experience fairly well.
>>>> After digging into the proposal of the project there are some
>>>> uncertainties that aroused.
>>>> In the descriptin you mention the introduction of new properties and
>>>> classes if needed.
>>>> Your first reference [2] concerns mainly with the reduction/fusion of
>>>> closely related
>>>> or equivalent properties.
>>>>
>>>> - Can you give me an intuition of a situation where a need for a new class
>>>> or
>>>> property would arise?
>>>>
>>>> - Can you also please give an example of tools that are based on
>>>> freebase and that
>>>> should be easily migrated to DBpedia?
>>>>
>>>> - Speaking of the current approaches of mapping classes and
>>>> properties, is there any
>>>> work currently going on that deal with hierarchies of subjects  
>>>> and objects?
>>>>
>>>> - Related to [2], do S1 and O1 represent actual subjects and objects
>>>> or rdf:type classes
>>>> of S1 and O1? I think one problem could (at least partially) solve the
>>>> other, namely
>>>> using a trustful class mapping could assist in working out equivalent
>>>> property mappings
>>>> and vice versa.
>>>>
>>>> I would be available full-time during the time period of GSoC and it comes
>>>> naturally for me that I get myself into the latest research prior  
>>>> the start
>>>> of the GSoC period.
>>>>
>>>> - Can you please advise me what would be the next step?
>>>>
>>>> - The project mentioned above is only one of my interests given your
>>>> proposals. Do I
>>>> have to elaborate my interest to my second and third priority in a
>>>> similar way?
>>>>
>>>> Best regads
>>>>
>>>> Robert
>>>>
>>>> [1] http://dws.informatik.uni-mannheim.de/en/home/
>>>> [2] http://wiki.knoesis.org/index.php/Property_Alignment
>>>>
>>>>
>>>>
>>>> ------------------------------------------------------------------------------
>>>> Dive into the World of Parallel Programming The Go Parallel Website,
>>>> sponsored
>>>> by Intel and developed in partnership with Slashdot Media, is your hub for
>>>> all
>>>> things parallel software development, from weekly thought leadership blogs
>>>> to
>>>> news, videos, case studies, tutorials and more. Take a look and join the
>>>> conversation now. http://goparallel.sourceforge.net/
>>>> _______________________________________________
>>>> Dbpedia-gsoc mailing list
>>>> Dbpedia-gsoc@lists.sourceforge.net
>>>> https://lists.sourceforge.net/lists/listinfo/dbpedia-gsoc
>>>>
>>
>>
>>
>>
>> ------------------------------------------------------------------------------
>> Dive into the World of Parallel Programming The Go Parallel  
>> Website, sponsored
>> by Intel and developed in partnership with Slashdot Media, is your  
>> hub for all
>> things parallel software development, from weekly thought  
>> leadership blogs to
>> news, videos, case studies, tutorials and more. Take a look and join the
>> conversation now. http://goparallel.sourceforge.net/
>> _______________________________________________
>> Dbpedia-gsoc mailing list
>> Dbpedia-gsoc@lists.sourceforge.net
>> https://lists.sourceforge.net/lists/listinfo/dbpedia-gsoc
>>
>
> -- 
> Marco Fossati
> http://about.me/marco.fossati
> Twitter: @hjfocs
> Skype: hell_j




------------------------------------------------------------------------------
Dive into the World of Parallel Programming The Go Parallel Website, sponsored
by Intel and developed in partnership with Slashdot Media, is your hub for all
things parallel software development, from weekly thought leadership blogs to
news, videos, case studies, tutorials and more. Take a look and join the 
conversation now. http://goparallel.sourceforge.net/
_______________________________________________
Dbpedia-gsoc mailing list
Dbpedia-gsoc@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/dbpedia-gsoc

Reply via email to