It looks like you're new here. Sign in or register to get started.
Ingo Mierswa wrote:Hi Motaz,there are no new and updated text mining sample processes yet, sorry. We are currently working on (re-)writing the complete documentation and probably there will be some new processes available in a few weeks.Cheers,IngoP.S.: To all others: the value series plugin / extension was not yet released but we try to release it as early as possible.
with RapidMiner 5.0 we will have a much more powerful extension mechanism, but it didn't make it into the beta, because the development took longer than expected.
Although the new TextProcessing and the new ValueSeriesProcessing extensions, which by the way will replace the old plugins, are ready to be deployed, we will publish them together with the final version of RapidMiner 5. Then hopefully available over a comforting auto update system.
Is there a planned release date for 5.0?
I'm interested in using the new text mining plugin to help research different term weighting methods. I'm expecting I'll need to implement some of the newer/less common term weighting methods myself. So, I wanted to wait and use the new plugin architecture from 5.0 before I began working.
the final release of RapidMiner 5 together with a few plugins is planned for mid to late December this year.
Text Processing is not included in the beta release. will it be included in the full release or it will be available as separate plug in?
it will be available as a separate extension as it used to be with version 4.6. However, probably you do not have to download it independently but you will be able to install it using a very comfortable updating mechanism directly within RapidMiner.
finally we can use text mining processing in RM5. by the way, we should RM as administrator/root to make update process successful.
I am asking if there is text mining processing sample processes ? because this would help in discovering new text mining operators in RM5 RC
there are no new and updated text mining sample processes yet, sorry. We are currently working on (re-)writing the complete documentation and probably there will be some new processes available in a few weeks.
P.S.: To all others: the value series plugin / extension was not yet released but we try to release it as early as possible.
I've just downloaded RM from sf.net (revision9), built it and played around with it. A question came up concerning the plugin system: Is the update server part of RM and thus part of the distribution? Could I set up an own update-server internally within my company to deploy some inhouse plugins? For sure I could inform my colleagues by mail about such updates, but using the update-system seems to be more elegant. Is there any means to propagate extensions of the global rapid-i update-server into such a local update-server?
Just for clarification: For sure, I'd like/plan to publish plug-ins as open-source where possible, however some special purpose solutions are unlikely to be of general interest.
Is there already a scheme which allows to submit own extensions to rapid-i to make it available to a wider audience via the rapid-i update server?
text mining became more popular task now. Is there any particular reason to provide it in RM RC as a separate expansion?
the same applies for value series plugin / extension
I will be waiting for TM sample processes
we are currently working on an enterprise server for RapidMiner, which then could propagate new versions of the core and extensions via intranet and as well would server for distributing your own inhouse extensions.
Until this is finished, there's no such possibility, but we are willing to host third party extensions which are not too specific and might be of value for anybody else.
And now to your question, Motaz:
We decided to somehow distinguish between general data mining algorithms usable in every task and task specific operators like the Text operators or Value Series operators. The main reason is, that RapidMiner already takes some time to start and it is not necessary to load all operators you will never touch, because you're not working in this field. Another reason is, that the integration of the new extensions is much smoother, as well as for the user as for the developer than in 4.x with the old plugins, so it's of no disadvantage to separate it.
Even if you load a process that contains operators of a extension you don't even have installed, it costs only two clicks to install it!
The last reason is, that we can make bugfix releases for each extension separately.