Can someone guide me on using jasen through java code in an Email Client(developed in Java)
Sorry for the delayed response.. too much work!
Unfortunately jASEN is not really well suited to a client application implementation at present. This is really because of the memory requirements in the default implementation, and the size of the data file needed to store all the spam/ham training data.
Having said that, most of the memory usage is done because we are pre-loading most of the data so we can process large volumes of email quickly. In the case of client application this is not so important, and you could potentially create an alternate JasenMapStore which loads the training set from some form of database for each email scan. This would probably still be quite slow though, even for a single client app.
We have toyed with the idea of using Lucene as the repository for training data. This would allow us to maintain the data set on disk without having a serious impact on performance, and also allow us to more easily provide "live updates" of training information. This is something that would also be ideally suited to a client app environment (ie, if the user hits "flag as spam", the training data is updated) But.. nothing actually done on this to date.
The reality is that jASEN was created to solve a problem with server-based spam filtering, so we haven't really been cautious about memory usage/thread counts etc. Clearly the engine has uses outside this situation, but we haven't yet taken the time to retro-fit a lighter version.
I'll add it to the list.. but I wouldn't hold out any hopes of a significant release in this area in the near future.
Log in to post a comment.