Temporal Text Classification for Romanian Novels set in the Past

29240

Authors

nan

Abstract

In this paper we look at a task in historical linguistics and the study of language de- velopment, namely that of identifying the time when a text was written. The nov- elty is that we evaluate our classifier and our selected features on literary texts hav- ing their action placed in the past and writ- ten so as to give off the impression of the respective epoch. We investigate several types of features and ultimately go with a very simple set of 10 features which very accurately classifies the texts based on the century they were actually written in. We use random forests to obtain high perfor- mance. 1

Keywords

nan

Full Content

Proceedings of Recent Advances in Natural Language Processing, pages 136–140, Hissar, Bulgaria, 7-13 September 2013. Temporal Text Classification for Romanian Novels Set in the Past Alina Maria Ciobanu Liviu P. Dinu Octavia-Maria S, ulea Faculty of Mathematics and Computer Science Center for Computational Linguistics University of Bucharest alinamaria.ciobanu@yahoo.com ldinu@fmi.unibuc.ro mary.octavia@gmail.com Anca Dinu Faculty of Foreign Languages University of Bucharest anca d dinu@yahoo.com Vlad Niculae University of Wolverhampton vlad@vene.ro Abstract In this paper we look at a task in historical linguistics and the study of language de- velopment, namely that of identifying the time when a text was written. The nov- elty is that we evaluate our classifier and our selected features on literary texts hav- ing their action placed in the past and writ- ten so as to give off the impression of the respective epoch. We investigate several types of features and ultimately go with a very simple set of 10 features which very accurately classifies the texts based on the century they were actually written in. We use random forests to obtain high perfor- mance. 1 Introduction and Motivation Determining the time when a document was writ- ten is a task not only with implications in cul- tural heritage but one which proves important to many other domains such as historical and literary criticism, diachronic linguistics, manuscript phy- logeny and stemmatics, and the elaboration of crit- ical theories about the author of the texts in ques- tion. A more practical, coarser grained approach is to classify according to the century in which a text was written, approach that we take in this paper. Within many instances of this task, disputes be- tween linguists and historians appear. For ex- ample, among the first texts written in Romania there are four religious texts, Codicele Voronet¸ean, Psaltirea Scheian˘a, Psaltirea Voronet¸ean˘a and Psaltirea Hurmuzachi, for which the dating is dis- puted between the 15th century (idea promoted by historians such as Nicolae Iorga) and the end of the 16th century (idea maintained by linguists such as Rosetti) (Tagliavini, 1972). Often times, the texts present characteristics of a translation, yet they are not original translations but modern copies of lost originals. For Romanian, the 16th century represents the beginning of Romanian writing. In (Dimitrescu, 1994, p. 13) the author states that the modern Romanian vocabulary cannot be completely un- derstood without a thorough study of the texts written in this period, which should be consid- ered the source of the literary language used to- day. In the 17th century, some of the most im- portant cultural events took place, such as the im- provement of the education system and the es- tablishing of several printing houses, and this led to a new development of the Romanian language (Dimitrescu, 1994, p.75). Then, in the 18th cen- tury, a diversification of the philological interests in Romania took place through writing the first Romanian-Latin bilingual lexicons, the draft of the first monolingual dictionary, the first Romanian grammar, and the earliest translations from French (Lupu, 1999, p. …

Content truncated to 500 words

Labeled Text

In this paper we look at a task in historical linguistics and the study of language development, namely that of identifying the time when a text was written. The novelty is that we evaluate our classifier and our selected features on literary texts having their action placed in the past and written so as to give off the impression of the respective epoch. We investigate several types of features and ultimately go with a very simple set of 10 features which very accurately classifies the texts based on the century they were actually written in. We use random forests to obtain high performance. 1

Publication Details

Paper ID:
29240
Year:
2013
Publication:
TIPSTER
Publication Type:
conf
DOI:
nan