Skip to content

Latest commit

 

History

History
85 lines (58 loc) · 3.26 KB

README.md

File metadata and controls

85 lines (58 loc) · 3.26 KB
Raw

Note: Original repo was at Google Code (https://code.google.com/p/jacana/). It was too big and after conversion, commit history was lost. Also, the data files had to be separated. Please download the data files and untar it to the same folder as the project files.

Jacana (/dʒəˈkɑːnə/) consists of three parts:

  • jacana-qa, a Question Answering engine for TREC-style questions, written in Java.
  • jacana-align, a monolingual word aligner for English, written in Java and Scala.
  • jacana-freebase, a Question Answering engine for web-style questions on Freebase, written in Java and Scala.

download:

[http://www.cs.jhu.edu/~xuchen/packages/jacana-align.bin.20141029.tar.bz2 jacana-align.bin.20141029.tar.bz2] (103M, just the binary version of the aligner)

[http://www.cs.jhu.edu/~xuchen/packages/jacana.20141029.tar.bz2 jacana.20141029.tar.bz2] (660MB, everything (qa+align+freebase, source code only, no compiled jars), mostly data files and lexical resources)

jacana-qa

contains:

  • a ranker for ranking whether a sentence contains an answer for a given question,
  • an answer extractor for extracting exact answer segments from sentences.

It provides software implementation and dataset for the following two papers:

[http://cs.jhu.edu/~xuchen/paper/yao-jacana-qa-naacl2013.pdf Answer Extraction as Sequence Tagging with Tree Edit Distance]. Xuchen Yao, Benjamin Van Durme, Peter Clark and Chris Callison-Burch. Proceedings of NAACL 2013.

[http://cs.jhu.edu/~xuchen/paper/yao-jacana-ir-acl2013.pdf Automatic Coupling of Answer Extraction and Information Retrieval]. Xuchen Yao, Benjamin Van Durme and Peter Clark. Proceedings of ACL 2013, short papers.

HOWTO: JacanaQA

jacana-align

is a token-based aligner described in the following paper:

[http://cs.jhu.edu/~xuchen/paper/yao-jacana-wordalign-acl2013.pdf A Lightweight and High Performance Monolingual Word Aligner]. Xuchen Yao, Benjamin Van Durme, Chris Callison-Burch and Peter Clark. Proceedings of ACL 2013, short papers.

[http://jacana.clsp.jhu.edu/ online demo]

HOWTO: JacanaAlign

jacana-freebase

is a question answering engine that extracts answers from Freebase:

Information Extraction over Structured Data: Question Answering with Freebase. Xuchen Yao and Benjamin Van Durme. Proceedings of ACL 2014.

HOWTO: JacanaFreebase

Change Log

2015-08-18:

  • imported from Google Code. Original repo exceeded GitHub limit (less 1G repo, single file less than 100MB). Had to separate data from main repo and lost commit history.

2014-10-29:

  • jacana-align: is now thread-safe, model is also updated/re-trained

2014-10-2

  • jacana-align: removed word2vec since it's too slow
  • jacana-align: added parallel support (by default), use the "--single" parameter to disable it

2014-3-19

  • released jacana-freebase

2013-11-3

  • added more lexical resources: word2vec, wiktionary, better PPDB support
  • feature tuning for token alignment
  • full UTF-8 support in AJAX demo

2013-8-5

  • retrained model on larger alignment corpus
  • now compiles with Scala-2.10.2
  • fixed Windows compatability issue of several lexical models
  • demo pre-loads model first

2013-5-28

  • Initial release