Skip to content

Latest commit

 

History

History
62 lines (62 loc) · 5.92 KB

CHANGES.md

File metadata and controls

62 lines (62 loc) · 5.92 KB

#Changes

  • 1.5.1 - Update to bioresources 1.1.28.
  • 1.5.1 - Updated to processors 7.5.3.
  • 1.5.1 - Event polarity is now detected using a hybrid approach that uses rules for simpler texts, and deep learning for more complex statements.
  • 1.5.1 - Added verbose text to the index card format.
  • 1.5.1 - Implements variables for grammar rules.
  • 1.5.0 - We now use Universal Dependency syntax for all grammar rules rather than Stanford dependencies.
  • 1.4.1 - Switched default processor to BioNLPProcessor rather than the server. The client/server config remains available.
  • 1.4.1 - Update to use released Processors 6.3.0 as new Reach base. Update Scala to 2.11.11.
  • 1.4.1 - Update Wiki pages on configuring and running Reach. Update README and CHANGES files.
  • 1.4.1 - Add argument to restrict NXML input file size.
  • 1.4.0 - Update Reach to use Processors 6.1.5 (refactored Client/Server). Allow use of new CLUProcessor.
  • 1.4.0 - Refactor Processors Client to main Processors project. Update top level scripts to start/stop Client & Server.
  • 1.3.5 - Update to use Akka 2.5.4. Update to use Processors 6.1.4. Allow args in POST body of built-in server.
  • 1.3.5 - Update to use Processors 6.1.2 and Bioresources 1.1.24, which includes the Harvard BioEntities updates of 8/22/2017.
  • 1.3.4 - Refactor internal Akka servers, resources, and file locations. Add configurable mainClass for FAT JAR.
  • 1.3.4 - Added Akka-based processors client and updated processor-client classes. Updated to use Processors 6.0.7.
  • 1.3.3 - Updated to use Bioresource 1.1.23 and Processors 6.0.6.
  • 1.3.3 - Syntax fix in activation tests. Remove prep_in from rule.
  • 1.3.3 - Added option to use FastBioNLPProcessor, which makes Reach ~40% faster.
  • 1.3.3 - Add serial JSON output format to API. Switch to using Jackson JSON library.
  • 1.3.3 - Fix lib clash: allow ai.lum.common to dicate typesafe.config version.
  • 1.3.3 - Use consistent SBT version across projects. Update ai.lum.nxmlreader to special release 0.0.9.
  • 1.3.3 - Find shortest path between heads in actions. Add/allow outputter for JSON serialization format.
  • 1.3.3 - Fix trigger affecting countSemanticNegatives. Add restart capability and more logging stats to ReachCLI.
  • 1.3.3 - Added a new tabular format for the DyCE CMU model. Update Phase3 use cases. Add/use root path in config.
  • 1.3.3 - The "arizona" tabular format was extended with source/destination for Translocation events.
  • 1.3.3 - Use Processors 6.0.1 and Bioresources 1.1.19, which contains Harvard Bioentities project KBs. Update many tests for BE KBs.
  • 1.3.3 - Mentions are created from previously unrecognized aliases.
  • 1.3.3 - Sub-project split into main, assembly, export.
  • 1.3.2 - Optimizations to json serialization/deserialization of CorefMention.
  • 1.3.2 - Uses bioresources 1.1.15 and processors 5.9.6. Introduces json serialization/deserialization of CorefMention (including grounding, modifications, etc.).
  • 1.3.2 - Better handling of nested events.
  • 1.3.2 - Update to use Bioresources 1.1.15 and Processors 5.9.5.
  • 1.3.1 - Update of all edu.arizona.sista packages to org.clulab
  • 1.3.1 - Updated rules for parsing changes. Add rules for "VBN with" regulations. Correct amino acid rule. Add syntax -> surface rule. Several changes to activation/regulations rules.
  • 1.3.1 - Refactor protein domain table. Add protein kinase lookup table.
  • 1.3.1 - Fix bugs in grounding. Add ID overrides from collaborators. Add tests.
  • 1.3.1 - Replace use of manual KB files with NER/grounding override file. Add tests.
  • 1.3.1 - Refactor grounding tables to a standardized 2-5 column format. Update KB tests.
  • 1.3.0 - Use Bioresources 1.1.9, Processors 5.8.5. Allow regulations of regulations. Identify X inhibitors as chemicals. Use MITRE model for NER/grounding. Add NER stop list. Add Translocation mention support to Assembly. Various rule fixes and enhancements. Allow PaperReader to read .csv files.
  • 1.3.0 - Update to use Bioresources version 1.1.8, which includes NER stop list.
  • 1.3.0 - Update code and tests to replace homemade KBs, in conjunction with Bioresources version 1.1.7.
  • 1.3.0 - Sieve-based assembly system that includes a feature-based classifier and rule-based sieves for detecting causal precedence.
  • 1.3.0 - Bug fix and optimizations to the AssemblyManager.
  • 1.3.0 - Updated to Bioresources 1.1.6, Processors 5.8.4. Add/use Uniprot Tissue type KB.
  • 1.3.0 - Add new context frames and pointers to FRIES output event stream.
  • 1.3.0 - Replace ChEBI and HMDB KBs with PubChem KB.
  • 1.3.0 - Add isDirect flag to BioEventMentions to track if a regulation is direct or not.
  • 1.3.0 - Add "reverse" PTM events (e.g., dephosphorylation, deubiquitination, etc).
  • 1.3.0 - User PFAM protein family KB for NER and grounding.
  • 1.3.0 - Context is assigned to all type of mentions, not just event mentions.
  • 1.3.0 - Improvements to event extraction grammars.
  • 1.2.2 - Added cellular locations to context. Context engine now can be configured by the user. Added a new deterministic context policy that extends context before and after a mention. Added support for Pandas output format. Bug fixes: context is now enabled by default; coref engine now matches the Hobbs antecedent search heuristic.
  • 1.2.1 - Bug fix in the unboxing of controller events. New Year's Eve release!
  • 1.2.0 - First release of context extraction! Context includes: species, organs, cell lines and types. Improved coreference resolution with constraints on determiner type.
  • 1.1.3 - Bug fixes in coreference resolution plus output formats
  • 1.1.2 - Bug fixes in the index card output
  • 1.1.1 - Improved event coreference resolution
  • 1.1.0 - Added entity and event coreference resolution
  • 1.0.0 - Initial release