A Uniform Morphological Analyzer for the Kazakh and Turkish Languages

Abstract

The Kazakh and Turkish languages belong to the group of the Turkic languages and have much in common. The detailed comparison of the ontologies on the example of the Kazakh and Turkish nouns allowed entering the analysis of morphological rules of these languages and the uni ed system of designations to create the uniform morphological analyzer based on the general algorithm of the morphological analysis.

Keywords

• morphological analysis of the Kazakh and Turkish languages
• ontology
• analysis of morphological rules

1 Introduction

One of the methods to reduce the semantic barrier between the human and the computer is searching new methods of a natural language processing. Nowadays it is obvious that in order to implement the human-computer interaction in a natural language and to create a linguistic support of the information processes the study of the language itself is required. Besides the resources consumed could be decreased due to formalization of language rules providing the storage of information in procedural but not in declarative form. For the Kazakh and the Turkish languages which morphological regularities are quite well yielded to formalization, it would produce an excellent result.

All language levels are characterized by existence of basic elements. A language studying can take place from two positions { the analysis and synthesis because the revealed rules of synthesis can assist to carry out the analysis and vice versa. In this case the Kazakh and Turkish languages are studied from both positions the analysis and synthesis. This very integrated approach allows to study in details all regularities and to reveal such nuances which when using only of one of approaches would remain outside our attention. For researching and the maximum formalizing of each language subsystem it is necessary to create the program tools implementing the studying process by identifying and verifying the analysis and synthesis rules. There-with it will greatly automate the research process and a researcher does not need to accumulate and collect information. And the labor intensity is very low.

The morphology modeling is related to all applications such as natural language and tasks processing and includes information search, moods analysis, spelling correction, detection of the generated texts, parts of speech marking and entity extraction. The morphology is used in linguistics to refer to the study of structure and formation of words. The Agglutinative languages (agglutinare from Latin means to stick together) are languages which morphological system is characterized by agglutination ("pasting") of various formants. As a formant either pre xes or su xes act and each of them makes its own sense. As the Kazakh and Turkish languages belong to the group of Turkic languages and the languages of this group can be classi ed as agglutinative languages. These languages are full of word forms (in ections). In ections are formed by addition of su xes. The su xes are attached in the strict sequence and the resulting new words can belong to the other part of speech. The possessive form in Kazakh is similar to a possessive form in English [ 1, 2 ]. Plenty of researches covering formalization of morphological rules and morphological analyses of [ 3-6 ] the Turkic languages are avail-able. The rst morphological analyzer of Kazakh was developed in 2009 and based on the procedural method. The procedural method implies the preliminary systematization of morphological knowledge about a natural language and development of morphological information assignment algorithms to a separate word form [ 7, 8 ].The procedural morphological analyzer of Kazakh consisted of the following stages: marking the stem in the current word form, its identi cation, assigning to a word form the corresponding list of morphological information. The disadvantage of this method is high labor intensity while compiling the dictionaries of compatibility. This challenge is difcult to be settled and cannot be automated completely for languages which are characterized by a large number of counterexamples. The implementation of this method occupies considerably smaller memory size, but at the same time the morphological analysis period due to splitting a word form into components and applying the procedures of compatibility increases [ 8 ]. The second version of the morphological analyzer was developed in 2012 and based on the formal morphological rules [ 9 ]. Later versions were based on using the ontological models and the hyper graphs [ 10-13 ]. The other research groups developed their own morphological analyzers [ 15-16 ].

The works on creation of the morphological analysis for the Turkish language are carried out for a long period of time and presented in papers [ 17-21 ]. In this paper the results received in [ 17 ] were used. The peculiarity of this morphological analyzer is the methodology for carrying out the analysis. The Turkish words with a xation were used without any lexicon. This morphological analyzer is completely based on the rules and implies using only the dictionary of counterexamples. The analyzer is based on the nal automatic model. 2

The generalized ontologic models of parts of speech of the Kazakh and Turkish languages Ontology is a powerful and widely used tool for modelling relationships between objects which belong to the di erent subject area. Ontology should be classi ed based on the degree of dependence on the task or application area, ontological model for knowledge representation and expression as well as other criteria [ 22 ]. We used the ontology editor Protg (http://protege.stanford.edu) to build the ontology. It is a free open source ontology editor and a framework for building knowledge bases. It was developed at Stanford University in collaboration with the University of Manchester.

The morphological features of initial forms of nouns (N) are as follows. A noun can be either animate (anim) or inanimate (inanim); this feature determines the trajectory of the in ection of a noun. Nouns in the Kazakh language can be conjugated (pers end) and vary for case (cases) and number (number), as well as have a possessive form (poss end) [ 10 ]. Figure 1 shows the ontological model of the Kazakh noun with its morphological features. Concepts and relationships used in this ontological model are explained

Anim

Sign animacy Inanim Sign inanimacy Cases of

Animate of v16

Inanimate

The ontology model of the Kazakh parts of speech allows us to completely de-scribe the morphological rules and their relationships. On the basis of this ontological model we developed generalized ontological models of the Kazakh and Turkish language parts of speech. The developed ontological model of nouns of the Kazakh language in Protege environment is displayed in the Figure 2, and the ontological model of nouns of the Turkish language is shown below in the Figure 3. In this way the comparative ontological models of noun for machine translation system include all the categories of morphological features, for instance, noun is divided as stem and complex according to the structure of noun in the Kazakh language whereas in the Turkish language there is not such division, furthermore, a noun can be common, proper, concrete, abstract, animated, inanimate according to meaning in the Kazakh language, while in the Turkish language a noun can be common, proper, animated, inanimated. In both languages the divisions of a xation are similar, e.g., the forms of cases, number, possessives and conjugations. There are seven cases in Kazakh whereas in Turkish there are ve.

The uniform morphological analyzer for the Kazakh and Turkish languages The comparison of the ontological models allowed creating the general symbol system of morphological markers which are used in morphological analyzer. For example, the line 4 of the above-mentioned table does not have any meanings in the Kazakh and Turkish columns, there is no analogue in English, but preserved name means that for the other Turkic languages such morphological marker for noun exists. In the lines 12 and 16 the blank value in the Turkish language means that this morphological marker exists only for the Kazakh language. Metalanguage is one of key concepts of system of the description of an object of science and is de ned as arti cial language of "the second order" in relation to which natural human language acts as "language object", that is as a subject of a linguistic research. In our case a natural language are the Kazakh and Turkish

NON. POSS

languages enter-ing into the Turkic group of languages.

The uni ed symbol system (UNIFIED TAG) was developed based on the idea of creating uni ed metalanguage for Turkic Languages.

Firstly, the idea of creating metalanguage was proposed at the 1st so-called International Conference on Computer processing of the Turkic Languages (TurkLang2013) which was held in Astana on 3-4 October, 2013. A group of famous professors of technical sciences A.A. Sharipbay (Astana, Kazakhstan), D.SH. Suleimenov (Kazan, Tatarstan, Russia), Esref Adal (Istanbul, Turkey) is working on the creation of metalanguage.

At the UniTurk scienti c-practical seminar of the conference the discussion of problems related to the uni cation of grammatical categories for the corpuses of the Turkic languages raised a great interest and held successfully. For computerizing the Kazakh language it is very important step to research the computational linguistics of the other Turkic-speaking countries. From this point studying the structures of agglutinative languages that are similar to Kazakh and make comparisons between them leads to a successful computer transforming of all languages belonging to the Turkic languages group. We are very con dent that it will bring great success in development of the Kazakh language computerizing. Our goal is to use correctly these similarities and di erences in the language automating direction. While entering to a computer the similarities between languages help to solve the unsolved problems in one language by supplementing the achievements of another language, moreover, studying the di erences of languages according to its features in cooperation gives us an opportunity to implement a method in one language which didn't give any results in another language. The analysis made revealed that the Kazakh and Turkish languages have much in common. The Table 3 shows the comparison of the rules for a noun window in the Kazakh and Turkish languages. The record of morphological rules in the uni ed form allowed to create the uniform rule-based algorithm of morphological analysis for the Kazakh and Turkish languages in the papers [ 9, 10, 17 ]. 4

Conclusion

In the present scienti c paper the morphological features of the Kazakh and Turkish languages are analyzed. The ontologies comparison is made, the uniform symbol system of morphological features is developed and the morphological rules of the Kazakh and Turkish languages are written over via new symbol system. The uni ed morphological analyzer is developed based on the general morphological analysis algorithm.

In the future it is supposed to create the uni ed metalanguage of the Turkic languages that will allow reaching the new level the Turkic languages processing.

References

1. Batayeva , Z. ( 2012 ). Colloquial Kazakh, Routledge

2. Kazakh grammar. ( 2002 ). Phonetics, word formation, morphology, syntax (in Kazakh) . Astana

3. Sharipbayev A. A. , Bekmanova G. T. : The synthesis of word forms of Turkic language using semantic neural networks . Modern problems of applied mathematics and information technologies: abstracts Al Khorezmy , pp. 145 ( 2009 )

4. Sharipbayev , A.A. , Bekmanova , G. T.: The building of logical semantics of the Kazakh words. The materials of the all-Russian conference with International participations Knowledge-Ontology-Theory (ZONT-09 ), Novosibirsk, pp. 246 - 249 ( 2009 )

5. Tantug , A. C. , Adal , E., O azer, K.: Computer Analysis of the Turkmen Language Morphology . In: Salakoski T., Ginter F. , Pyysalo S. , Pahikkala T . (eds) Advances in Natural Language Processing. Lecture Notes in Computer Science , vol 4139 , pp. 186 193 , Springer, Berlin, Heidelberg ( 2006 )

6. Orhun , M. , Tantug , A. C. , Adal , E.: Rule Based Analysis of the Uyghur Nouns . Proceedings of the International Conference on Asian Language Processing (IALP) . Chiang Mai , Thailand, 19 ( 1 ): pp. 33 - 43 ( 2008 )

7. Bekmanova , G. T. : Some approaches to the problems of automatic in ection and morpholog-ical analysis in the Kazakh language . The newsletter of D. Serikbayev East Kazakhstan state technical university, Ust-Kamenogorsk, pp. 192 - 197 ( 2009 )

8. Dobrushina , E. P. , Savina , G. B. , Gelbukh , A. G. : The system of an accurate morphological analysis and synthesis . The software of new information technology , Kalinin ( 1989 )

9. Sharipbayev , A. , Bekmanova , G. , Mukanova, ., Buribayeva , A. , Yergesh , B. , Kaliyev , A. : Semantic neural network model of morphological rules of the agglutinative languages . The 6th International Conference on Soft Computing and Intelligent Systems The 13th International Symposium on Advanced Intelligent Systems , Kobe, Japan, 20 -24 November 2012 , pp. 1094 - 1099

10. Yergesh , B. , Mukanova , A. , Bekmanova , G. , Sharipbay , A. , Razakhova , B. : Semantic hy-per-graph based representation of nouns in the Kazakh language . Computacion y Sistemas; Volume 18, Issue 3 , 1 July 2014 , pp. 627 - 635

11. Mukanova , A. , Yergesh , B. , Bekmanova G. , Razakhova , B. , Sharipbay , A. : Formal models of nouns in the Kazakh language . Leonardo Electronic Journal of Practices and Technologies; Is-sue 25 ( July-December ) , 2014 ( 13 ), pp. 264 - 273

12. Zetkenbay , L. , Sharipbay , A. , Bekmanova , G. , Kamanur , U. : Ontological modeling of morphological rules for the adjectives in Kazakh and Turkish languages . Journal of Theoretical and Applied Information Technology , Vol. 91 . No. 2 , pp. 257 - 263 ( 2016 )

13. Kamanur U. , Sharipbay A. , Altenbek G. , Bekmanova G. , Zhetkenbay L.: Investigation and Use of Methods for De ning the Extends of Similarity of Kazakh Language Sentences . In: Sun M. , Huang X. , Lin H. , Liu Z., Liu Y . (eds) Chinese Computational Linguistics and Natural Language Processing Based on Naturally Annotated Big Data. CCL 2016 , NLP-NABD 2016 . LNCS , vol 10035 . Springer, Cham ( 2016 )

14. Tukeyev , U. , Zhumanov , Zh., Rakhimova , D. , Kartbayev , A. : Combinational Circuits Model of Kazakh and Russian Languages Morphology . Abstracts of International Conference Computational and Informational Technologies in Science, Engineering and Education , pp. 241 - 242 . Al-Farabi KazNU Press, Almaty ( 2015 )

15. Kessikbayeva , G. , Cicekli , I. : Rule-Based Morphological Analyzer of Kazakh Language . Linguistics and Literature Studies 4 ( 1 ): pp. 96 - 104 ( 2016 )

16. Makhambetov , O. , Makazhanov , A. , Sabyrgaliyev , I. , Yessenbayev , Z. Data-Driven Morphological Analysis and Disambiguation for Kazakh . In: Gelbukh A. ( eds) Computational Linguistics and Intelligent Text Processing . CICLing 2015 .LNCS, vol 9041 , pp. 151 - 163 . Springer, Cham ( 2015 )

17. Erygit , G. , Adal , E. : An a x stripping morphological analyzer for turkish . In Proceedings of the IASTED International Conference on Arti cial Intelligence and Applications , Innsbruck, Austria, pp. 299304 ( 2004 )

18. Erygit , G. , Adal , E.: Synthetic Turkish Word Root Generation . Proceedings of the Turkish Arti cial Intelligence and Neural Networks , TAINN , Canak-kale, Turkey ( 2003 )

19. Ak n , A.A. , Ak n , M.D.: Zemberek, an open source nlp framework for Turkic languages . Available at http://zemberek.googlecode.com

20. Hakkani-Tu r, D. Z., O azer, K. , Tur, G.: Statistical morphological disambiguation for agglutinative languages . In Proceedings of COLING. ICCL ( 2000 )

21. Hankamer , J. Finite State Morphology and Left to Right Phonology . Proceedings of the West Coast Conference on Formal Linguistics 5 . Stanford University ( 1986 )

22. Gruber , T.R. : Toward Principles for the Design of Ontologies Used for Knowledge Sharing . International Journal Human-Computer Studies Vol. 43 , Issues 5- 6 , 907928 ( 1995 )