IXAGroupEHUDiac: A Multiple Approach System towards the Diachronic Evaluation of Texts

Proceedings of the 9th International Workshop on Semantic Evaluation (SemEval 2015), pages 840–845, Denver, Colorado, June 4-5, 2015. c⃝2015 Association for Computational Linguistics IXAGroupEHUDiac: A Multiple Approach System towards the Diachronic Evaluation of Texts Haritz Salaberri†, Iker Salaberri‡, Olatz Arregi†, Beñat Zapirain† †IXA Group - Faculty of Computer Sciences, ‡Faculty of Arts University of the Basque Country, Spain firstname.lastname@ehu.eus Abstract This paper presents our contribution to the SemEval-2015 Task 7. The task was subdi- vided into three subtasks that consisted of au- tomatically identifying the time period when a piece of news was written (1,2) as well as automatically determining whether a specific phrase in a sentence is relevant or not for a given period of time (3). Our system tackles the resolution of all three subtasks. With this purpose in mind multiple approaches are un- dertaken that use resources such as Wikipedia or Google NGrams. Final results are obtained by combining the output from all approaches. The texts used for the task are written in En- glish and range from the years 1700 to 2000. 1 Introduction According to Mihalcea and Nastase (2012) cur- rent applications within human language technol- ogy work with languages as if they were constant. However, changes in language are taking place con- stantly, for example: new meanings for old words are coined; metaphoric and metonymic uses become so ingrained that they are considered literal from one specific point in time on; new words are constantly being created. These changes in language are what in part has motivated the task addressed by our system. In fact, subtasks (1) and (2) tackle the problem of computa- tionally identifying the time period in which a piece of news was written. This is undertaken based on, among other things, the changes that take place in language over time. The difference between sub- tasks (1) and (2) is that the texts in subtask (1) con- tain clear references to time anchors. This means that e.g. historical events, relevant people, commer- cial products etc. are mentioned in the text that are specific to the period of time in which the texts were written. Subtask (3), on the other hand, consists of determining whether a phrase within a clause is spe- cific or not to the period of time in which the text was written. The training corpus for this subtask is made up of the texts from other subtasks. As a con- sequence our system will be able to use information on both time anchors and language changes in order to generate the results for subtask (3). This paper is organized as follows: section 2 presents the resources available to the diachronic evaluation of texts; section 3, on the other hand, shortly reviews the relevant literature on this mat- ter. Section 4 makes a description of the developed system; results are then described in section 5 and, finally, our conclusions are given in section 6. 2 Resources To the extent of our knowledge, there exist two main resources as of …

Content truncated to 500 words