Applied Natural Language Processing Group

Middle East Technical University

Department of Computer Engineering

Applied Natural Language Processing Group

About Us

The Applied NLP Group at METU's Department of Computer Engineering is dedicated to advancing cutting-edge research in the field of natural language processing (NLP). Our research covers a wide spectrum of NLP topics, including but not limited to, large language models, generative AI, text mining, social media analysis, and voice assistants.

We collaborate with industry partners to apply our research to real-word use cases. By working on large-scale datasets and employing advanced machine learning models, we aim to improve the efficiency, accuracy, and scalability of language technologies.

At the Applied NLP Group, our mission is not only to contribute to the academic community through high-quality publications but also to create solutions that have a significant impact on everyday applications, such as virtual assistants, content moderation, and intelligent systems.

Research Topics

Generative AI (Large Language Models)
Social Media Analysis (Hate Speech Detection, Misinformation Detection, Topic Detection)
Text Mining (Text Classification, Similarity Search, Named Entity Recognition)
Voice Assistants (Intent Detection, Slot Filling)

Active Projects

Low-Resource Adaptation of Generative LLMs

Despite advancements in English-dominant Generative Large Language Models, further development is needed for low-resource languages to enhance global accessibility. This project aims to adapt generative large language models, primarily trained on English, to low-resource languages.

Turkish LLM Benchmark

METU LLM Benchmark is a comprehensive evaluation framework designed to assess the performance of Large Language Models on Turkish language tasks. This project aims to facilitate the development of more accurate and reliable Turkish language models.

Toxic Content (Disinformation and Hate Speech) Detection for Generative LLMs (partner Havelsan)

By leveraging advanced natural language processing and machine learning techniques, we can effectively identify and flag harmful content in real-time, reducing the spread of online toxicity and promoting a safer online environment. This project aims to develop accurate models for detecting toxic content, including hate speech and misinformation, on social media platforms.

News

EMNLP 2024 Conference Multilingual Representation Learning Workshop accepted our paper titled "Adapting Open-Source Generative Large Language Models for Low-Resource Languages: A Case Study for Turkish"
Research Article published in Natural Language Processing, Cambridge University Press. Click here to see the article.