Papers by Abdullatif Köksal
Do We Know What LLMs Don’t Know? A Study of Consistency in Knowledge Probing (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Existing methods for probing knowledge gaps in large language models are inconsistent and inconsistent. |
| Approach: | They propose a process based on input variations and quantitative metrics to evaluate probing methods that are inconsistent on knowledge gaps. |
| Outcome: | The proposed process exposes two dimensions of inconsistency in knowledge gap probing. |
Language-Agnostic Bias Detection in Language Models with Bias Probing (2023.findings-emnlp)
Copied to clipboard
| Challenge: | Pretrained language models (PLMs) contain strong social biases, which are difficult to quantify because current methods focusing on fill-the-mask objectives are sensitive to slight changes in input. |
| Approach: | They propose a bias probing technique called LABDet to evaluate social bias in pretrained language models with a language-agnostic method. |
| Outcome: | The proposed method “surfaces” nationality bias by training a classifier on top of a frozen PLM on non-nationality sentiment detection. |
LongForm: Effective Instruction Tuning with Reverse Instructions (2024.findings-emnlp)
Copied to clipboard
| Challenge: | Prior work on instruction tuning relies on expensive human annotation and crowd-sourced datasets with alignment issues. |
| Approach: | They propose a method to generate instructions via LLMs from human-written corpus examples using reverse instructions. |
| Outcome: | The proposed method outperforms larger language models without instruction tuning on tasks such as story/recipe generation and long-form question answering. |
Balancing Methods for Multi-label Text Classification with Long-Tailed Class Distribution (2021.emnlp-main)
Copied to clipboard
| Challenge: | Multi-label text classification is a challenging task because it requires capturing label dependencies. |
| Approach: | They propose to use distribution-balanced loss functions to solve label dependency problems in multi-label text classification by capturing label dependencies from a fixed-set of labels. |
| Outcome: | The proposed loss function addresses both the class imbalance and label linkage problems and outperforms other loss functions. |
SynthEval: Hybrid Behavioral Testing of NLP Models with Synthetic Evaluation (2024.findings-emnlp)
Copied to clipboard
| Challenge: | Existing frameworks for benchmarking in NLP often overestimate performance . however, manually creating a variety of test types requires significant human labor . |
| Approach: | They propose a framework that leverages large language models to generate a wide range of test types . they first generate sentences via LLMs and then identifies challenging examples . |
| Outcome: | The proposed framework overestimates performance on two classification tasks. |
TurkishMMLU: Measuring Massive Multitask Language Understanding in Turkish (2024.findings-emnlp)
Copied to clipboard
| Challenge: | Existing multiple choice question answering benchmarks employ automatic translation for multilingual evaluation, but this approach is error-prone and potentially introduces culturally biased questions. |
| Approach: | They introduce the first multitask, multiple-choice Turkish QA benchmark, TurkishMMLU . they evaluate over 20 LLMs including open-source, closed-source and Turkish-adapted models . |
| Outcome: | The proposed benchmarks evaluate the reasoning, comprehension, and mathematical abilities of large language models. |
Consistent Document-level Relation Extraction via Counterfactuals (2024.findings-emnlp)
Copied to clipboard
| Challenge: | Document-level relation extraction models trained on factual data exhibit inconsistent behavior, relying on spurious signals such as specific entities and external knowledge to extract triples. |
| Approach: | They propose a counterfactual data generation approach for document-level relation extraction datasets using entity replacement to generate triples from factual data. |
| Outcome: | The proposed approach extracts triples from factual data but fails on counterfactual modification. |
The better your Syntax, the better your Semantics? Probing Pretrained Language Models for the English Comparative Correlative (2022.emnlp-main)
Copied to clipboard
| Challenge: | Construction Grammar posits constructions as the central building blocks of language . human-like performance of pretrained language models on many NLP tasks has been alleged . |
| Approach: | They propose to use construction grammar to posit constructions as the central building blocks of language . they conduct experiments with three pretrained language models to examine their ability to classify and understand English comparative correlative . |
| Outcome: | The proposed models are able to recognise the English comparative correlative (CC) but fail to use its meaning. |
The RELX Dataset and Matching the Multilingual Blanks for Cross-Lingual Relation Classification (2020.findings-emnlp)
Copied to clipboard
| Challenge: | Current approaches for relation classification are focused on the English language and require lots of training data with human annotations. |
| Approach: | They propose a baseline model based on Multilingual BERT and a new multilingual pretraining setup . they propose 'relationship classification' models that use distant supervision . |
| Outcome: | The proposed model significantly improves the baseline model with distant supervision. |
MEAL: Stable and Active Learning for Few-Shot Prompting (2023.findings-emnlp)
Copied to clipboard
| Challenge: | Existing methods for few-shot classification have high variance across different sets of few shots and finetuning runs. |
| Approach: | They propose novel ensembling methods that significantly reduce run variability and introduce a new active learning criterion for *data selection*. |
| Outcome: | The proposed method significantly reduces run variability and improves performance on five tasks. |
Evaluating Morphological Compositional Generalization in Large Language Models (2025.naacl-long)
Copied to clipboard
Mete Ismayilzada, Defne Circi, Jonne Sälevä, Hale Sirin, Abdullatif Köksal, Bhuwan Dhingra, Antoine Bosselut, Duygu Ataman, Lonneke Van Der Plas
| Challenge: | Large language models (LLMs) have demonstrated significant progress in various natural language generation and understanding tasks. |
| Approach: | They define morphemes as compositional primitives and design a suite of generative and discriminative tasks to assess morphological productivity and systematicity. |
| Outcome: | The proposed models can identify individual morphological combinations better than chance, but their performance lacks systematicity, leading to significant accuracy gaps compared to humans. |
TUMLU: A Unified and Native Language Understanding Benchmark for Turkic Languages (2025.acl-long)
Copied to clipboard
Jafar Isbarov, Arofat Akhundjanova, Mammad Hajili, Kavsar Huseynova, Dmitry Gaynullin, Anar Rzayev, Osman Tursun, Aizirek Turdubaeva, Ilshat Saetov, Rinat Kharisov, Saule Belginova, Ariana Kenbayeva, Amina Alisheva, Abdullatif Köksal, Samir Rustamov, Duygu Ataman
| Challenge: | preparing native language MMLU benchmarks is costly and limits representativeness of evaluation datasets. |
| Approach: | They propose to use a Turkic language MMLU benchmark to assess massive multitask language understanding capabilities. |
| Outcome: | The proposed benchmarks are based on a Turkic language morphosyntactic and cultural benchmark . the benchmarks evaluate a diverse range of open and proprietary multilingual large language models . |