Hi Joshua, I think the main selling points are that Apache UIMA would allow you to modularize the tasks in your application, and that for each NLP task, there are numerous open-source UIMA annotators available that would allow you to build a starter pipeline you could improve upon by building and training your own custom annotators, or finding different, perhaps better (for your domain), open-source annotators that perform the same task. For me, the customizable, modular "plug-and-play" nature of UIMA pipelines has been essential to optimizing performance.
In short, anything is possible with sentiment analysis in UIMA! But UIMA itself doesn't DO the sentiment analysis. - Jessica On Fri, Nov 18, 2016 at 9:19 AM, Joshua Moody <[email protected]> wrote: > I want to know more about what is possible with sentiment analysis. > Specifically, in an application in which the content providers are > relatively static, always known, and often provide opinions on the same > topic. I would like to systematically discern the opinion of the content > providers to show a trend. Reviewer A is anti-citrus fruits and pro-apples > as seen by these 10 reviews. The end goal is to reveal more about the > content providers by consistently analyzing their sentiments. > > Please advise on how UIMA could assist in this project. I am a software > developer looking at multiple text engineering applications to build on. > > Thanks, > Josh >
