I was wondering are there plans to get KNIME to look at some of the bioinformatic sites as primary datasources eg. EMBL, MEDLINE, PUBMED etc. The problem is that KNIME would need a number of text mining nodes to analyse this type of data i.e. pattern search functions, extraction of data, text formating etc. Have you any plans to explore this area in the future? or development text mining nodes?
From looking at the package now for a couple of weeks, you have a great peice of software, but it still needs some more development in the variery of nodes available. Can you give any indication of the roadmap of KNIME and where you expect to be taking KNIME over the next 12-18 months.
Well - it all depends on what you are trying to do. I am sure someone will always find something missing in KNIME.
However, we are working on some text mining functionality internally here as well. If there is sufficient interest we may release it in the future but it's not that high on our list of things for the "stable" version. Moving functionality from "experimental" to "releasable" causes a lot of additional work.
I just like to say, KNIME is proving quite robust in my hands, and is a good piece of software. I totally agree with you, you cannot possibly please everyone, however, it would be useful if you could outline where you expect to take KNIME - which areas you are specifically interested in for example. You mention that you have some text mining, can you elaborate on this, or are these nodes not fully developed?. Would it be possible to publish these or other nodes on the site as Beta - use at own risk?. What other nodes/functions are in development?
I liked your comments stanage and second the view that Knime is a great piece of software! Thanks again to Michael and everyone else at Knime.
The roadmap and beta releases would be especially useful. The roadmap would help institutions to determine if the software is right to use based on it's (intended) future direction. The release of beta (experimental) nodes would help to get the community (at least some of us) more involved in Knime's development. It would also hopefully give you a few more testers.
Hmm, you guys insist, eh? ;-)
I am _very_ reluctant to talk about a roadmap - at least medium to long term - because our development right now is pretty much driven by research projects and collaborations. So the types of projects we have often shift priorities one way or the other. We *are* working on a new workflow manager that will allow to much easier create own metanodes (by collapsing parts of a workflow and expanding it later again). This will also add the ability to create simple loops and pass variables into the flow for use by subsequent nodes.
We are also constantly adding new nodes and will release package with a few additional nodes soon.
As for having an open platform to play with and continue develop experimental nodes - that is a great idea and we are looking into it. Especially for new types it would be important to converge to an accepted "standard" early in the development phase.
By now KNIME provides a lot of this functionality via the Textprocessing plugin which is available via the Labs update site. For further detail go to the Textprocessing website.