Ontology learning in TBC-ME

[Martina Husáková]

Good afternoon,

I would like to ask whether it is possible to practise ontology learning in TBC-ME?

Thank you a lot for your answer(s).

Best regards,

Martina

[Irene Polikoff]

Not sure what you mean by this. You can create ontologies in TBC, if this is your question.

--
You received this message because you are subscribed to the Google Groups "TopBraid Suite Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to topbraid-user...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/topbraid-users/7699218e-7cc0-47b0-8dca-53386fa8a18dn%40googlegroups.com.

[Martina Husáková]

Good afternoon,

my question was whether the ontology learning can be realized in the TBC-ME. It means whether we can use non-structured texts for automatical or semi-automatical development of the ontological structures.

In other words, if we can apply some specific algorithms of NLP, text mining, etc. in automatical/semi-automatical building of the ontological structures, because their manual development is often very time consuming.

Best regards,

Martina Husáková

To view this discussion on the web visit https://groups.google.com/d/msgid/topbraid-users/BC106487-7E58-4754-86E6-C736F692173A%40topquadrant.com.

[Irene Polikoff]

We define an ontology as a data model - classes, their properties and associated rules. There is no support for getting such models from non-structured text. There is support for getting this from a structured text such as spreadsheet (the columns will be turned into properties).

TopBraid EDG (launchable from TBC) offers support for creating taxonomies from unstructured text. It will extract terms from a document, sort them by frequency and let you pick which ones you want to have as skos:Concepts. Organizing of terms into hierarchies would be done through post processing. You can write business logic specific to your text to automated some of it. For example, if you have terms that contain other terms (e.g., Dog and Small Dog), you can run a query using text functions to identify these situations and make one term a child concept of another.

To view this discussion on the web visit https://groups.google.com/d/msgid/topbraid-users/CAAn62ofrdvUBW%2B0%2B7%3DMO_nax-%3Dm8%3DtABA_aNsytbRJ0tRBELVA%40mail.gmail.com.

[Martina Husáková]

Good afternoon,

I have already tried to create the taxonomy from the unstructured text in TB EDG. It works perfectly :-)

Thank you for your feedback.

Best  regards,

Martina Husáková

To view this discussion on the web visit https://groups.google.com/d/msgid/topbraid-users/0AFE197D-EDD3-45DD-BB3F-765002855941%40topquadrant.com.