N-gram Based Croatian Language Network: Application in a Smart Environment

Abstract

In the field of natural language processing, language networks represent a method for observing linguistic units and their interactions in different linguistic contexts. This paper uses the previously presented Croatian language network for building a solution capable of generating spoken notifications in Croatian language. The novelty of this paper is that it proposes an approach for generating spoken notifications in smart environments by combining specialized services that enable interaction with the environment and human users. The process employed for generating spoken notifications is described in detail. Also, a novel contribution of this paper is the case-study evaluation of the proposed approach in a smart home environment.

Error Pattern Discovery in Spellchecking Using Multi-Class Confusion Matrix Analysis for the Croatian Language

“…This study highlights the efforts to develop language technologies specific to Croatian. Additionally, Šoić and Vuković [31] utilize a Croatian language network for building a solution capable of generating spoken notifications in Croatian, demonstrating the practical applications of language technologies in the Croatian context. Šantić et al [32] describe a system for automatic diacritic restoration in Croatian texts, which combines dictionary lookup and statistical language modeling, achieving high levels of accuracy.…”

A Comprehensive Dataset of Spelling Errors and Users’ Corrections in Croatian Language

“…The Natural Language Processing group at the University of Zagreb's Faculty of Humanities and Social Sciences created a high-quality web corpus, MaCoCu-Hr, for linguistic studies, training language models, and other language applications [6]. Additionally, researchers at the University of Zagreb Faculty of Electrical Engineering and Computing have developed an n-gram language model [7,8] and a contextual spellchecking model based on the n-gram system [9]. Finally, a group from the University of Rijeka examined sentiments in Twitter tweets during the COVID-19 pandemic and developed their own model by utilizing BERT tools and a linguistic corpus [10].…”