Estnltk — Open source tools for Estonian natural language processing

Estnltk is a Python 2.7/Python 3.4 library for performing common language processing tasks in Estonian, funded by Eesti Keeletehnoloogia Riiklik Programm under the project EKT57. Estnltk is licensed under GNU GPL version 2.

To get started right now, see Installation.

University of Tartu has a course that covers many aspects of the library: .

Quick example

The best way to learn Estnltk is to start using it right away, so without further ado, let’s begin with an example on how to extract some morphological information from text and display it as a nicely formatted DataFrame object:

from estnltk import Text
text = Text('Mine vanast raudteeülesõidukohast edasi ja pööra paremale, siis leiad Krokodilli!')
                   word_texts               lemmas postags  postag_descriptions
0                    Mine               minema       V             tegusõna
1                  vanast                 vana       A  omadussõna algvõrre
2   raudteeülesõidukohast  raudteeülesõidukoht       S             nimisõna
3                   edasi                edasi       D             määrsõna
4                      ja                   ja       J             sidesõna
5                   pööra              pöörama       V             tegusõna
6                paremale             paremale       D             määrsõna
7                       ,                    ,       Z            lausemärk
8                    siis                 siis       J             sidesõna
9                   leiad               leidma       V             tegusõna
10             Krokodilli            Krokodill       H            pärisnimi
11                      !                    !       Z            lausemärk

Here is the same data as a dictionary:

{'lemmas': ['minema',
 'postag_descriptions': ['tegusõna',
                         'omadussõna algvõrre',
 'postags': ['V', 'A', 'S', 'D', 'J', 'V', 'D', 'Z', 'J', 'V', 'H', 'Z'],
 'word_texts': ['Mine',


In recent years, several major NLP components for Estonian have become available under free open source licenses, which is an important milestone in Estonian NLP domain.

The goal of Estnltk is to become the main platform for Estonian NLP and glue together existing free components to make them easily usable. Current situation requires the researchers to write their own interfaces to the tools, which can be very time-consuming Also, a simple platform is a great resource for students who are interested in NLP domain.

The most important component is vabamorf, which is a C++ library for morphological analysis, disambiguation and synthesis [KA97]. For named entity recognition, Estner library provides necessary code and also a valuable training dataset, which is required for training the default models that come with the software [TK13]. Estnltk also includes temporal time expression (TIMEX) library [OR12]. The de facto library for NLP in English, the NLTK toolkit is also a dependency [BI06].

In addition to providing an API that is simple to use for software developers, Estnltk also aims to be useful for language researches and linguists in general. The library is used to create tools for sentiment analysis, text classification and information extraction, which requires no programming knowledge once they are set up. Including useful tools is a major goal in the future of Estnltk. See for a list of tools available.

[KA97]Kaalep, Heiki-Jaan. “An Estonian morphological analyser and the impact of a corpus on its development.” Computers and the Humanities 31, no. 2 (1997): 115-133.
[TK13]Tkachenko, Alexander; Petmanson, Timo; Laur, Sven. “Named Entity Recognition in Estonian.” ACL 2013 (2013): 78.
[OR12]Orasmaa, Siim. “Automaatne ajaväljendite tuvastamine eestikeelsetes tekstides.” Eesti Rakenduslingvistika Ühingu aastaraamat 8 (2012): 153-169.
[BI06]Bird, Steven. “NLTK: the natural language toolkit.” In Proceedings of the COLING/ACL on Interactive presentation sessions, pp. 69-72. Association for Computational Linguistics, 2006.



Table Of Contents

Next topic


This Page