Currently, most of the projects relying on ML are based on Python.
That’s because Python is simple, easy to learn, implement and test.
However, as being a script language, Python 3 can be 10 fold slower than compiled languages such as Java [1].
Further, the constant updates in the language make it hard to contribute and reuse NSPM project and models.
Many of the NSPM project issues are related to the Python version.
Machine Learning usually relies on n-cubic runtime algorithms and long training times to deliver models and predictions.
In an attempt to alleviate the aforementioned problems, in this work, we are interested in creating a Java implementation of NSPM.
@emarx the other two topics you posted have a good relation to DBpedia. This topic here seems a bit off-topic. Could you clarify the relation to DBpedia or LOD? If this connection is not clear, then any potential applications for GSOC might be rejected on that criteria.
Thnks for your feedback. I was thinking in evaluate the implementantion using the DBpedia Neural QA model, but I guess you are right. It is indeed out of topic because there is no direct gain to DBpedia.
Software testing is one of the main use cases of DBpedia like all SW databases and tools and many AI approaches are evaluated with DBpedia, but this then using DBpedia not improving DBpedia.