Ok, So that is serializing the model to an OutputStream. http://opennlp.apache.org/documentation/1.7.2/apidocs/opennlp-tools/opennlp/tools/util/model/BaseModel.html#method.summary
Have you tried serializing to a ByteArrayStream. Getting the byte[] with toByteArray(), and writing to database as a blob? Daniel > On Apr 14, 2017, at 1:13 PM, Benedict Holland <benedict.m.holl...@gmail.com> > wrote: > > Sure. It is actually throughout the document. All I had to do was search > for "database" in > > https://opennlp.apache.org/documentation/1.7.2/manual/opennlp.html > > I pulled text above as a copy and paste. > > I think the solution I was looking for would be to take a model's database > connection (input steam equivalent), create tables, and manipulate the > database to store the internal model format---however the object would > save. It would then control the object loading in the same way. I wouldn't > need to save the byte representation to a database, essentially saving a > file within a database, but I was wondering if objects would save internal > state to a database and then load and re-read it. If not, that might be an > interesting feature. > > Thanks, > ~Ben > > On Fri, Apr 14, 2017 at 12:46 PM, Daniel Russ <dr...@apache.org> wrote: > >> I wont say it is impossible. I have custom BaseModel (like POSModel or >> ParseModel but it does something different), if you use the commandline >> tool there is call to CmdLineUtil.writeModel(String,File,BaseModel). >> Notice that it requires a file. If you want to save it in a database, I >> can think of a few ways of doing this. >> >> 1. (Expert knowlege required) Edit the code to read/write to db >> 2. Write to file, then read as a large byte[], write to a blob. read blob >> from db, write file, read into opennlp. >> 3. Use something like MapDB? Read in models. put them in a MapDB database >> (looks like a hashmap, but is actually a key-value database. >> >> I think 1 and 2 would be difficult. If 3 works for you, let me know. Even >> if 3 does work, I don't think it will work with the command line tool. The >> best option may be to stick with the file-based models. Is there some >> reason this wont work? You might wont to consider adding a feature >> request, opennlp.apache.org > issue tracker (though you may have to create >> a login). >> >> Can you tell me where in the documentation it talks about saving models to >> a database? It probably should be fixed, since we cant do it now. >> Daniel >>
