Plagiarism detection system for Armenian language

Abstract

In the academic context, it is very important to evaluate the uniqueness of reports, scientific papers and other documents that are everyday disseminated on the web. There are already several tools with this purpose but not for Armenian texts. In this paper, a system to analyze the similarity of Armenian documents is presented. The idea is to collect a set of documents of the same domain in order to identify keywords. Then, based on that information, the system receives two documents and compares them calculating the probability of plagiarism. For that, an approach based on several levels of analysis is implemented and some of those steps allow the user interaction choosing options or adding more information.

Large Language Models and Low-Resource Languages: An Examination of Armenian NLP

“…Initially, Armenian NLP research focused on developing basic tools and methods, such as keyword identification and stemming, to handle specific tasks like plagiarism detection (Margarov et al, 2017). These early studies aimed to overcome the language's unique challenges and lay the founda-tion for further advancements.…”

“…The recent advancements in natural language processing (NLP) have been significantly driven by the advent and development of large language models (LLMs), such as BERT, GPT-3, and T5. However, despite their evident success, the adoption of LLMs in Armenian NLP tasks has been somewhat limited, as reflected in the current literature (Avetisyan et al, 2023;Yeshilbashian et al, 2022;Ter-Hovhannisyan and Avetisyan, 2022;Margarov et al, 2017). Only a handful of studies in Armenian NLP have explored the use of Large Language Models (LLMs).…”

“…Despite the proven efficacy of Large Language Models (LLMs) in various language processing tasks for well-resourced languages, their utilization in Armenian NLP remains limited. The prevailing approaches in Armenian NLP lean towards traditional methods, such as clustering, bag-of-words, and rule-based techniques (Yeshilbashian et al, 2022;Margarov et al, 2017). While these methods have shown relative success, the underutilization of LLMs highlights a research gap and the potential to enhance the performance of Armenian NLP systems.…”

“…Once sufficient resources and fine-tuned models are in place, efforts should build practical applications for Armenian NLP. These applications could include Armenian sentiment analysis systems, named entity recognizers, chatbots, and machine translation systems (Margarov et al, 2017).…”