+ **3.3** - bug fix: make sure DocumentSerializer.load() works when multiple documents are serialized into the same file.
+ **3.2** - Added a discourse parser to `FastNLPProcessor`. This performs marginally worse than the one in `CoreNLPProcessor`, but it is much faster for end-to-end processing, due to the shift-reduce syntactic parser.
+ **3.1** - Minimal functionality added to the learning package. Changed to CoreNLP 3.3.1.
+ **3.0** - Added a RST discourse parser to `CoreNLPProcessor`. Added the `edu.arizona.sista.learning` package. Utils classes are now under `edu.arizona.sista.utils` rather than `edu.arizona.sista.processors.utils`.
+ **2.2** - Various bug fixes. Added support for basic dependencies to `CoreNLPProcessor`.
+ **2.1** - Bug fix in FastNLPProcessor: better root detection algorithm, robust to malt inconsistencies.
+ **2.0** - We now support two processors: `CoreNLPProcessor` and `FastNLPProcessor`. Changed the package name from `e.a.s.processor` to `e.a.s.processors`. Added Java usage example to README. Updated to CoreNLP 3.3.0. Added better unit tests to check for thread safetiness.
+ **1.5** - Bug fixing. Made the string interning process (see `Processor.in`) local to each thread to avoid concurrency issues in multi-threaded programs. Added new unit tests. Added minor functionality to Lexicon.
+ **1.4** - Code cleanup. Added some minor new functionality such as finding base NPs in the Trees class.
+ **1.3** - Reverted back to the `1.x` version numbers, since we will add other software here not just CoreNLP. Added correct mvn dependencies for the CoreNLP jars. Removed the `install*.sh` scripts, which are no longer needed.
+ **3.2.0** - Updated to Scala 2.10.1 and CoreNLP 3.2.0. Changed versioning system to be identical to CoreNLP's, so it's clear which CoreNLP version is used.