selimfirat / turkishnlp.github.io Goto Github PK

This is a collection of tools and resources for processing Turkish. The list is still in its infancy. It most certainly is not the definitive or an exhaustive list. If you have suggestions or corrections please contact me.

• METU Corpus (there are a few more specific corpora linked from this page)
• Turkish National Corpus is a larger corpus built in a more recent effort. However, the corpus is only accessible thorough the query interface. The full data (to my knowledge) is not released.
• TS Turkish Corpus releases corpora in various domains including tweets, news columns, idioms and abstracts.

• Universal Dependencies (UD) project includes three freely available treebanks, a few more are likely to arrive soon.

  • The UD_Turkish-IMST treebank has its roots in IMST and METU-Sabancı treebanks listed below (although content seems to differ). There is also a version with a full correction/re-annotation effort. However, the changes are not incorporated to the version currently released by the UD project.

    Home pages of the earlier versions (with non-UD annotations) are: - IMST - METU-Sabancı treebank

  • The UD_Turkish-PUD treebank is created as part of the effort in CoNLL 2017 shared task.

  • The UD_Turkish-GB is a treebank constructed from grammar book examples (Göksel and Kerslake (2005)). The most up-to-date version of the treebank can be found here.

• Turkish WordNet

• Prof. Deniz Yüret's list of Turkish NLP resources