Question Classification for Albanian Language: An Annotated Corpus and Classification Models

Kote, Nelda; Trandafili, Evis; Plepi, Gjergj

doi:10.1007/978-3-031-19945-5_1

Cited by 2 publications

(6 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The Albanian question corpus was used to evaluate the performance of three question classification models, which utilized the Support Vector Machine (SVM), Logic Regression www.ijacsa.thesai.org (LR) and Random Forest (RF) algorithms. The SVM model achieved the highest accuracy of 75.7% using FastText, while the use of RoBERTa resulted in a slightly lower accuracy of 0.6% [5]. The question-answering system proposed for the Albanian language in [6] does not implement the question classification component.…”

Section: Related Workmentioning

confidence: 99%

“…In our Question Classification task, we used the implementation of Support Vector Machine (SVM), Random Forest and Logistic Regression from Scikit-learn to build Question Classification models [5]. The main object in our supervised learning is the Scikit-learn estimator that implements a fit method with two arguments: an array with data and an array with labels.…”

Section: B Scikit-learn (Sklearn)mentioning

confidence: 99%

“…Due to its generalization capabilities and the ability to compute word embeddings for out-of-vocabulary words, FastText has shown good accuracy for small datasets and for low resource languages. In [5] we used tf-idf, FastText and RoBERTa to represent word embedding fed to Machine Learning models and the results showed that FastText outperformed in terms of accuracy. a) Classical machine learning algorithms: Support Vector Machines, Random Forest and Logistic Regression with tf-idf, FastText and RoBERTa as vector representations.…”

Section: Dataset Preparationmentioning

confidence: 99%

“…a) Classical machine learning algorithms: Support Vector Machines, Random Forest and Logistic Regression with tf-idf, FastText and RoBERTa as vector representations. We used the results obtained in [5].…”

Section: Dataset Preparationmentioning

confidence: 99%

“…The implementation of FastText for vector representation is shown to increase the classification performance in all the classical machine learning models used for classification [5]. In Table V Experiment 2: Deep learning classification models using the preprocessed Albanian dataset.…”

Section: Experimental Evaluationmentioning

confidence: 99%

See 4 more Smart Citations

Question Classification in Albanian Through Deep Learning Approaches

Trandafili¹,

Kote²,

Plepi³

2023

IJACSA

View full text Add to dashboard Cite

In recent years, there is growing interest in intelligent conversation systems. In this context, Question Classification is an essential subtask in Question Answering systems that determines the question type, therefore, also the type of the answer. However, while there is abundant research for English, little research work has been carried out for other languages. In this paper we deal with classification of questions in the Albanian language which is considered a complex Indo-European language. We employ both machine learning and deep learning approaches on a large corpus in Albanian based on the six-class TREC dataset with approximately 5000 questions. Experiments with and without stop-words show that the impact of stop-words is significant in the accuracy of the classifier. Extensive comparison of algorithms for the task of question classification in Albanian show that deep learning algorithms outperform conventional machine learning approaches. To the best of our knowledge this is the first approach in literature for classifying questions in Albanian and the results are highly comparable to English.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: B Scikit-learn (Sklearn)mentioning

confidence: 99%

Section: Dataset Preparationmentioning

confidence: 99%

Section: Dataset Preparationmentioning

confidence: 99%

Section: Experimental Evaluationmentioning

confidence: 99%

See 3 more Smart Citations

Question Classification in Albanian Through Deep Learning Approaches

Trandafili¹,

Kote²,

Plepi³

2023

IJACSA

View full text Add to dashboard Cite

show abstract

Morphosyntactic probing of multilingual BERT models

et al. 2023

View full text Add to dashboard Cite

We introduce an extensive dataset for multilingual probing of morphological information in language models (247 tasks across 42 languages from 10 families), each consisting of a sentence with a target word and a morphological tag as the desired label, derived from the Universal Dependencies treebanks. We find that pre-trained Transformer models (mBERT and XLM-RoBERTa) learn features that attain strong performance across these tasks. We then apply two methods to locate, for each probing task, where the disambiguating information resides in the input. The first is a new perturbation method that “masks” various parts of context; the second is the classical method of Shapley values. The most intriguing finding that emerges is a strong tendency for the preceding context to hold more information relevant to the prediction than the following context.

show abstract

Question Classification for Albanian Language: An Annotated Corpus and Classification Models

Cited by 2 publications

References 29 publications

Question Classification in Albanian Through Deep Learning Approaches

Question Classification in Albanian Through Deep Learning Approaches

Morphosyntactic probing of multilingual BERT models

Contact Info

Product

Resources

About