Bilingual Word Level Language Identification for Omotic Languages

Alexander Gelbukh; Atnafu Lambebo Tonja; Girma Yohannis Bade; Grigori Sidorov; Mesay Gemeda Yigezu; Olga Kolesnikova

REVIEW 3 major objections 5 minor 44 references

Bilingual Word Level Language Identification for Omotic Languages

T0 review · 3 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read BERT embeddings plus an LSTM tag Wolayta and Gofa words at 0.72 F1

desk verdict A useful first dataset and evaluation for Wolayta-Gofa LID, but the reported 0.72 F1 is undermined by an unexplained preprocessing step that removes shared words and by the fact that the dataset is not released. read the letter →

arxiv 2509.07998 v1 pith:5AJLNJSY submitted 2025-09-05 cs.CL cs.AI

Mesay Gemeda Yigezu , Girma Yohannis Bade , Atnafu Lambebo Tonja , Olga Kolesnikova , Grigori Sidorov , Alexander Gelbukh This is my paper · ORCID

classification cs.CLcs.AI

keywords bilinguallanguageidentificationword-levelLIDOmoticlanguagesWolaytaGofaBERTLSTMlow-resourceNLP

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper tries to establish that word-level bilingual language identification is possible for Wolayta and Gofa, two closely related Omotic languages of southern Ethiopia whose shared vocabulary makes them hard to tell apart. The authors built and manually annotated a 144,000-word dataset with three labels: Wolayta, Gofa, and common words used by both. Among ten model configurations, a pipeline that feeds BERT-base-uncased embeddings into an LSTM classifier performed best, with a macro F1 score of 0.72 on the test set. If the result holds, it offers a practical tool for detecting which language a social media post or comment is written in and a first benchmark for NLP on these languages.

What carries the argument

The central mechanism is a hybrid pipeline: BERT-base-uncased produces contextualized word embeddings, and those embeddings are passed to an LSTM layer with 128 units, followed by batch normalization, two dense layers of 768 units with ReLU, a dropout layer of 0.1, and a softmax output over the three classes wal, gof, and wal-gof. The BERT embeddings do the load-bearing work of encoding subtle lexical and morphological differences between the two languages, while the LSTM models the sequential structure of the input and the classifier assigns the language label.

What would settle it

Take a random sample of the original unprocessed data, have fresh native speakers annotate every word without removing common words first, and compute inter-annotator agreement; then retrain the BERT+LSTM model on the unmodified labels and compare F1. If agreement is near chance or the F1 drops substantially, the reported 0.72 depends on the filtering and majority-vote procedure.

Watch

Extended reading notes

Core claim

On the authors' annotated three-class test set, the BERT-base-uncased plus LSTM model achieves 0.76 macro precision, 0.69 recall, and 0.72 F1, beating logistic regression (0.47), LSTM with attention (0.55), BiLSTM with attention (0.62), CNN (0.64), CNN+LSTM (0.66), CNN+BiLSTM (0.69), and RoBERTa+LSTM (0.69). The authors claim that pretrained contextual embeddings combined with an LSTM classifier are an effective approach for word-level bilingual LID of closely related, low-resource Omotic languages, and that the work provides the first dataset and benchmark for this language pair.

Load-bearing premise

The ground-truth labels are correct: each word was labeled by majority vote among three native speakers with no reported agreement measure, and shared words were removed with Excel Kutools before annotation, so the definition of the common-word class is not demonstrated on naturally occurring data.

Editorial extensions

If this is right

The proposed model identifies the language of short, word-level texts, which is the setting typical of tweets and comments.
The annotated 144,000-word dataset gives a first resource for further NLP work on Wolayta and Gofa.
Word-level LID can separate languages even when they share a large common vocabulary, as long as contextualized representations are used.
The approach is presented as transferable to other low-resource Ethiopian languages and to applications such as hate-speech detection and machine-translation prefiltering.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

If the labels are reliable, the same BERT+LSTM recipe should generalize to other closely related Ometo languages, but its margin over simpler models may shrink as the shared vocabulary grows.
Removing common words before annotation likely made the wal-gof class sparse or artificial; on raw, unfiltered text the reported 0.72 F1 could drop, so a test on unmodified data would be needed for deployment.
Reporting per-class precision and recall, plus an inter-annotator agreement score, would show whether the wal-gof class is a genuine linguistic category or mainly an annotation artifact.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Colleague,

Quick take: this is a modest but genuinely new piece of work for a language pair nobody has benchmarked before. The authors built a 144K-word annotated corpus for Wolayta and Gofa, ran a standard battery of ML, DL, and transformer+LSTM classifiers, and got a plausible 0.72 macro F1 with BERT-base+LSTM. The architecture is explicitly borrowed from Tonja et al. on Kannada-English, and the authors don't pretend otherwise. That's fine—the contribution is the dataset and the first reported evaluation for these Omotic languages.

The paper does several things well. The annotation procedure is carefully described for a low-resource setting: three native speakers per language, majority vote, author review. The results table is internally coherent; the precision/recall numbers match the F1 values. The related work situates the task within existing bilingual LID efforts. If the dataset were shipped, this would be a useful baseline for future work on these languages.

Now the soft spots, and they're not minor. Section 4.2 says: "We used Excel Kutools to remove the matching common words from both Wolayta and Gofa to annotate the data." The stress-test note is right: if identical words are removed before annotation, the wal-gof class (shared words) is either emptied or reduced to whatever survives a heuristic spreadsheet filter. The test set then no longer samples the actual bilingual LID task—the hardest cases are gone. The abstract's claim about identifying between the two languages is measured on a truncated problem. This is load-bearing, and the paper never explains what the wal-gof label means after that filtering.

Other issues: the dataset is described as "open source" but no link is given. There is no train/test split description, no inter-annotator agreement, no error bars. The BiLSTM attention F1 appears as 0.62 in Table 3 and 0.68 in Section 6—probably a typo, but it needs fixing. Some references look like padding, but that's secondary.

Who should read this? Anyone working on low-resource Ethiopic NLP. But the value depends on the data being released and the preprocessing clarified.

Verdict on peer review: yes, send it out. The data collection effort and the specific language pair deserve referee time. But the revision must address the common-word removal, release the dataset, and fill in the experimental details. As it stands, the result is not independently verifiable.

Best, [You]

Referee Report

3 major / 5 minor

Summary. The paper presents a word-level bilingual language identification (LID) system for Wolayta and Gofa, two closely related Omotic languages spoken in southern Ethiopia. The authors collected 144K words from religious, educational, and social media sources, annotated them as wal, gof, or wal-gof by majority vote of native speakers, and evaluated several models: logistic regression, LSTM and BiLSTM with attention, CNN, CNN+LSTM, CNN+BiLSTM, RoBERTa+LSTM, and BERT-base-uncased+LSTM. They report the best result for BERT-base-uncased+LSTM with macro precision 0.76, recall 0.69, and F1 0.72 on the test set. The paper claims this is the first LID work for the Omotic language family and motivates applications in social-media content analysis and downstream NLP tasks.

Significance. If the dataset and preprocessing are sound, this is a useful first benchmark for an under-resourced language pair, comparing classical, deep, and transformer-hybrid approaches on a new word-level annotation. The strongest contribution is the creation of a new annotated dataset for Wolayta and Gofa and the systematic comparison of multiple architectures. However, the significance is currently limited by unclear dataset release, ambiguous preprocessing, and inconsistent result reporting. The central performance claim is plausible and the top model's F1 is internally consistent with its precision and recall, but the empirical comparison must be verified after the reported issues are resolved.

major comments (3)

[Sections 4.2, 4.3 and Table 2] The preprocessing description contradicts the definition of the wal-gof class. Section 4.2 states: "We used Excel Kutools to remove the matching common words from both Wolayta and Gofa to annotate the data." Section 4.3 then defines wal-gof as a label for words shared by both languages, and Table 2 lists examples such as Kaallidi, Hara, Eridi, and Doonan. If matching common words were literally removed before annotation, the wal-gof class could not contain those examples; if they were not removed, the sentence is misleading. This is load-bearing because the reported 0.72 F1 is measured on the resulting test set, and removing the most confusable shared words would change the task from the claimed bilingual LID problem. The authors should clarify exactly what was removed, how the wal-gof class was constructed after filtering, and report the per-class distribution in the final training and test sets.
[Table 3 and Section 6] The result table is internally inconsistent. For Logistic Regression, Table 3 reports precision 0.55 and recall 0.76, whose harmonic mean is 0.64, but the table lists F1 as 0.47. For BiLSTM with attention, precision 0.70 and recall 0.66 give F1 0.68, but the table lists 0.62; Section 6 text says BiLSTM with attention achieved F1 0.68. These discrepancies make the reported comparison unreliable and must be corrected. The central ranking (BERT+LSTM best at 0.72) may survive correction, but the magnitude of differences between models depends on this fix.
[Sections 4.1, 4.4 and 4.5] The paper omits essential experimental details for reproducibility. It does not report the train/test split ratio, the test set size, the random seed, or whether the split was stratified by source domain. It claims an "open source dataset" in Section 1 but provides no repository link or availability statement. Section 4.4 reports only majority voting among three annotators and gives no inter-annotator agreement measure, so the reliability of the ground-truth labels cannot be assessed. These details are necessary to evaluate whether the 0.72 F1 result is stable and whether the dataset can be used by other researchers.

minor comments (5)

[Table 3] There is a typo: "BiLSTM wiith attention" should be "BiLSTM with attention."
[Keywords line] The keywords line reads "Keywords:Keywords: LID..."; the duplicated prefix should be removed.
[Section 1 contributions] Some listed contributions are future plans rather than contributions of this paper, e.g., "The proposed model will be applied in the other low-resource Ethiopian languages." Please rephrase to describe what is actually delivered.
[Figure 3 and Section 4.5] The dataset distribution is shown only as percentages in Figure 3; please provide exact counts or a table with the number of words per class and per domain, as this is important for interpreting the class imbalance.
[References] Several reference entries are incomplete or inconsistently formatted (e.g., [36] and [37] lack full publication details, and some entries appear multiple times with different numbering). Please check the reference list against the venue's style guidelines.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the 0.72 F1 is an empirical measurement on an independently annotated test set, not derived from the model's fitted inputs.

full rationale

The paper's central claim is an empirical comparison of classifiers on a newly annotated corpus. The BERT+LSTM F1 score of 0.72 is computed on a test split of word-level annotations produced by three native speakers via majority vote, and no equation in the paper reconstructs this score from model parameters, fitted constants, or label definitions. No parameter is fitted to a subset of data and then renamed as a prediction. The wal-gof class is defined as words shared by Wolayta and Gofa, with examples in Table 2; even if Section 4.2's removal of matching common words truncated that class, that is a data-quality or task-definition concern, not a circular derivation, because the labels and the test labels do not depend on the model being evaluated. The self-citations in Sections 2 and 5, such as Yigezu et al. [15] and Tonja et al. [23], describe prior work on other language pairs and motivate the architecture, but they do not supply the reported numbers or forbid alternative explanations. No uniqueness theorem is invoked, and no ansatz is smuggled in via citation as a load-bearing premise. The main credibility risks are the unstated inter-annotator agreement, the absence of train/test split details, and the Excel Kutools filtering, all of which affect soundness and external validity rather than circularity. Accordingly, no circular step can be exhibited, and the score is 0.

Assumptions & free parameters 6 free parameters · 4 assumptions · 0 invented entities

The central claim rests on a manually annotated dataset whose labels are treated as ground truth, unstated dataset split and leakage assumptions, transfer assumptions for English BERT tokenization, and hyperparameters chosen by trial and error. No new theoretical entities are introduced.

free parameters (6)

learning rate = 0.001
Set by hand in Section 5.2; no sensitivity analysis reported.
batch size = 128
Set in model configuration in Section 5.2; no sensitivity analysis reported.
LSTM hidden units = 128
Section 6 states the LSTM layer has 128 units; chosen by hand.
dense layer size = 768
Section 6 describes a dense layer with 768 units before the softmax layer; chosen by hand.
dropout rate = 0.1
Dropout layer rate 0.1 is specified in Section 6; no tuning reported.
number of epochs = not stated (range 10 to 30)
Section 5.2 gives a range of 10 to 30 epochs but the actual value used in the reported experiments is not specified.

assumptions (4)

domain assumption Majority vote of three native-speaker annotators produces correct language labels.
Section 4.4 defines labels by majority vote but reports no inter-annotator agreement metric, so label reliability is assumed and is load-bearing for all reported F1 scores.
domain assumption The collected train/test data are representative and free of leakage.
Section 4 says the compiled dataset was shuffled, but no split proportions, source separation, or speaker overlap controls are described.
domain assumption English BERT-base-uncased subword tokenization transfers adequately to Wolayta and Gofa text.
Section 5.2 uses bert-base-uncased as the embedding layer; the tokenizer was trained on English and its applicability to these Omotic languages is not discussed.
ad hoc to paper Kutools-based removal of matching common words does not distort the label distribution.
Section 4.2 states matching common words were removed to annotate the data, but the effect on class balance and the wal-gof class is not described.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Bilingual Word Level Language Identification for Omotic Languages." pith.science (2026). https://pith.science/paper/5AJLNJSY

@misc{pith2026250907998,
  author       = {Pith},
  title        = {Pith review of: Bilingual Word Level Language Identification for Omotic Languages},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/5AJLNJSY}},
  note         = {Machine review of arXiv:2509.07998}
}

read the original abstract

Language identification is the task of determining the languages for a given text. In many real world scenarios, text may contain more than one language, particularly in multilingual communities. Bilingual Language Identification (BLID) is the task of identifying and distinguishing between two languages in a given text. This paper presents BLID for languages spoken in the southern part of Ethiopia, namely Wolaita and Gofa. The presence of words similarities and differences between the two languages makes the language identification task challenging. To overcome this challenge, we employed various experiments on various approaches. Then, the combination of the BERT based pretrained language model and LSTM approach performed better, with an F1 score of 0.72 on the test set. As a result, the work will be effective in tackling unwanted social media issues and providing a foundation for further research in this area.

Figures

Figures reproduced from arXiv: 2509.07998 by the authors.

**Figure 2.** Southern region of Ethiopia map [31] Omo, the southern part of Ethiopia. LID is still a difficult subject to solve, especially when trying to distinguish between languages that are quite similar [33]. A number of factors influenced the performance of BLID, including the similarity of the languages being identified and the length of the text. For example, the proposed languages that are more similar to each other a… view at source ↗

**Figure 3.** Distribution of tags among whole data sets(in %) [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗

Figures from the paper (1 more)

**Figure 4.** Figure 4: Architecture of proposed model 5.2 Experiment and parameter setting Experiment In this study, we present an experimental evaluation of several approaches to BLID. We developed the proposed model using the following approaches: Machine learning is a branch of artificia…

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

44 extracted references · 40 canonical work pages

[1]

H., and Baldwin, T

Lui, M., Lau, J. H., and Baldwin, T. (2014). Automatic detection and LID of multi- lingual documents. Transactions of the Association for Computational Linguistics, 2, 27-40

work page 2014
[2]

Alshami, and Jugal Kalita

Mersha, Melkamu, Khang Lam, Joseph Wood, Ali K. Alshami, and Jugal Kalita. ”Explainable artificial intelligence: A survey of needs, techniques, applications, and future direction.” Neurocomputing 599 (2024): 128111

work page 2024
[3]

Lillian Lee, Mark Johnson, Kristina Toutanova, and Brian Roark. 2018. Transac- tions of the Association for Computational Linguistics, Volume 6. Transactions of the Association for Computational Linguistics, 6

work page 2018
[4]

”Semantic-driven topic modeling using transformer-based embeddings and clustering algorithms.” Procedia Computer Sci- ence 244 (2024): 121-132

Mersha, Melkamu Abay, and Jugal Kalita. ”Semantic-driven topic modeling using transformer-based embeddings and clustering algorithms.” Procedia Computer Sci- ence 244 (2024): 121-132

work page 2024
[5]

(2006, May)

Hughes, B., Baldwin, T., Bird, S., Nicholson, J., and MacKinlay, A. (2006, May). Reconsidering LID for Written Language Resources. In LREC (pp. 485-488)

work page 2006
[6]

”Evaluating the effectiveness of XAI techniques for encoder-based language models.” Knowledge- Based Systems 310 (2025): 113042

Mersha, Melkamu Abay, Mesay Gemeda Yigezu, and Jugal Kalita. ”Evaluating the effectiveness of XAI techniques for encoder-based language models.” Knowledge- Based Systems 310 (2025): 113042

work page 2025
[7]

Indhuja, K., Indu, M., Sreejith, C., Sreekrishnapuram, P., and Raj, P. R. (2014). Text based LID system for indian languages following devanagiri script. Interna- tional Journal of Engineering, 3(4)

work page 2014
[9]

F., Dussias, P

Kroll, J. F., Dussias, P. E., Bogulski, C. A., and Kroff, J. R. V. (2012). Juggling two languages in one mind: What bilinguals tell us about language processing and its consequences for cognition. In Psychology of Learning and Motivation (Vol. 56, pp. 229-262). Academic press

work page 2012

Show all 44 references

[10]

”Smart-vision: survey of modern action recog- nition techniques in vision.” Multimedia Tools and Applications (2024): 1-72

AlShami, Ali K., Ryan Rabinowitz, Khang Lam, Yousra Shleibik, Melkamu Mer- sha, Terrance Boult, and Jugal Kalita. ”Smart-vision: survey of modern action recog- nition techniques in vision.” Multimedia Tools and Applications (2024): 1-72

2024
[11]

Bade, G. Y. (2021). Natural Language Processing and Its Challenges on Omotic Language Group of Ethiopia. Journal of Computer Science Research, 3(4), 26-30

2021
[12]

”Ethio-fake: Cutting-edge approaches to combat fake news in under-resourced languages using explainable ai.” Procedia Computer Science 244 (2024): 133-142

Mersha, Melkamu Abay, Girma Yohannis Bade, Jugal Kalita, Olga Kolesnikova, and Alexander Gelbukh. ”Ethio-fake: Cutting-edge approaches to combat fake news in under-resourced languages using explainable ai.” Procedia Computer Science 244 (2024): 133-142

2024
[13]

A review of natural language processing techniques for opinion mining systems

Shiliang Sun, Chen Luo, and Junyu Chen. A review of natural language processing techniques for opinion mining systems. Information fusion, 36:10–25, 2017

2017
[14]

”Explain- ability in neural networks for natural language processing tasks.” arXiv preprint arXiv:2412.18036 (2024)

Mersha, Melkamu, Mingiziem Bitewa, Tsion Abay, and Jugal Kalita. ”Explain- ability in neural networks for natural language processing tasks.” arXiv preprint arXiv:2412.18036 (2024)

2024 arXiv
[15]

G., Tonja, A

Yigezu, M. G., Tonja, A. L., Kolesnikova, O., Tash, M. S., Sidorov, G., and Gel- bukh, A. (2022, December). Word Level LID in Code-mixed Kannada-English Texts 14 M. G. Yigezu et al. using Deep Learning Approach. In Proceedings of the 19th International Conference on Natural La...

2022
[16]

”First Attempt at Building Parallel Corpora for Machine Translation of Northeast India’s Very Low-Resource Languages.” arXiv preprint arXiv:2312.04764 (2023)

Tonja, Atnafu Lambebo, Melkamu Mersha, Ananya Kalita, Olga Kolesnikova, and Jugal Kalita. ”First Attempt at Building Parallel Corpora for Machine Translation of Northeast India’s Very Low-Resource Languages.” arXiv preprint arXiv:2312.04764 (2023)

2023 arXiv
[17]

Jauhiainen, T., Ranasinghe, T., and Zampieri, M. (2021). Comparing approaches to Dravidian LID. arXiv preprint arXiv:2103.05552

2021 arXiv
[18]

AlShami, Sanghyun Byun, and Jugal Kalita

Mersha, Melkamu Abay, Mesay Gemeda Yigezu, Hassan Shakil, Ali K. AlShami, Sanghyun Byun, and Jugal Kalita. ”A unified framework with novel metrics for eval- uating the effectiveness of xai techniques in llms.” arXiv preprint arXiv:2503.05050 (2025)

2025 arXiv
[19]

”Code mix- ing: A challenge for LID in the language of social media.” In Proceedings of the first workshop on computational approaches to code switching, pp

Barman, Utsab, Amitava Das, Joachim Wagner, and Jennifer Foster. ”Code mix- ing: A challenge for LID in the language of social media.” In Proceedings of the first workshop on computational approaches to code switching, pp. 13-23. 2014

2014
[20]

”Explainable AI: XAI- guided context-aware data augmentation.” Expert Systems with Applications 289 (2025): 128364

Mersha, Melkamu Abay, Mesay Gemeda Yigezu, Atnafu Lambebo Tonja, Hassan Shakil, Samer Iskander, Olga Kolesnikova, and Jugal Kalita. ”Explainable AI: XAI- guided context-aware data augmentation.” Expert Systems with Applications 289 (2025): 128364

2025
[21]

Automatic LID: An alternative unsupervised approach using a new hybrid algorithm

Abdelmalek Amine, Zakaria Elberrichi, and Michel Simonet. Automatic LID: An alternative unsupervised approach using a new hybrid algorithm. Int. J. Comput. Sci. Appl., 7 (1):94–107, 2010

2010
[22]

”Text summarization using rhetorical structure trees.” In Recent Advances in Natural Language Processing, p

Trenk, Edward, Melkamu Mersha, and Jugal Kalita. ”Text summarization using rhetorical structure trees.” In Recent Advances in Natural Language Processing, p
[23]

Transformer-based model for word level LID in code-mixed Kannada-English texts

Atnafu Lambebo Tonja, Mesay Gemeda Yigezu, Olga Kolesnikova, Moein Shahiki Tash, Grigori Sidorov, and Alexander Gelbuk. Transformer-based model for word level LID in code-mixed Kannada-English texts. arXiv preprint arXiv:2211.14459, 2022

2022 arXiv
[24]

Mager, M., C ¸ etino˘ glu,¨O., and Kann, K. (2019). Subword-level LID for intra-word code-switching. arXiv preprint arXiv:1904.01989

2019 arXiv
[25]

waldemariam, H. (2007). The challenges of mother-tongue education in Ethiopia: The case of North Omo area. Language Matters, 38(2), 210-235

2007
[26]

Wakasa, M. (2014). A sketch grammar of walaytta. Nilo-Ethiopian Studies, 2014(19), 31-44

2014
[27]

Azeb, A. (1996). Aspects of the verb in Ometo: Ethiopia. J. Lang. Lit., 6, 43-64

1996
[28]

Amha, A. (2017). The Omotic language family. Cambridge University Press

2017
[29]

T., and waldeyohannis, M

Bedecho, A. T., and waldeyohannis, M. M. (2022, November). walaytta-English Cross-lingual Information Retrieval using Neural Machine Translation. In 2022 In- ternational Conference on Information and Communication Technology for Devel- opment for Africa (ICT4DA) (pp. 96-101). IEEE

2022
[30]

Michael, L., de Carvalho, F., Chacon, T., Rybka, K., Sabogal, A., Chousou- Polydouri, N., and Kaiping, G. (2022). Deriving calibrations for Arawakan using archaeological evidence. Interface Focus, 13(1), 20220049

2022
[31]

Melesse, A., and Negesse, T. (2011). Phenotypic and morphological characteri- zation of indigenous chicken populations in southern region of Ethiopia. Animal Genetic Resources/Resources g´ en´ etiques animales/Recursos gen´ eticos animales, 49, 19-31

2011
[32]

Hirboro, S. C. (2015). Documentation and grammatical description of Gofa. Addis Ababa Univeristy. Bilingual Word-Level LID for Omotic Languages 15

2015
[33]

Bjerva, J. (2016). Byte-based LID with deep convolutional networks. arXiv preprint arXiv:1609.09004

2016 arXiv
[34]

L., waldeyohannis, M

Tonja, A. L., waldeyohannis, M. M., and Yigezu, M. G. (2021, November). Parallel corpora for bi-directional neural machine translation for low-resourced Ethiopian languages. In 2021 International Conference on Information and Communication Technology for Development for Africa...

2021
[35]

Kumari, S. (2022). Text mining and pre-processing methods for social media data extraction and processing. In Handbook of Research on Opinion Mining and Text Analytics on Literary Works and Social Media (pp. 22-53). IGI Global

2022
[36]

G., Kolesnikova, O., Sidorov, G., & Gelbukh, A

Yigezu, M. G., Kolesnikova, O., Sidorov, G., & Gelbukh, A. (2023). Transformer- Based Hate Speech Detection for Multi-Class and Multi-Label Classification

2023
[37]

G., Bade, G

Yigezu, M. G., Bade, G. Y., Kolesnikova, O., Sidorov, G., & Gelbukh, A. (2023). Multilingual Hope Speech Detection using Machine Learning

2023
[38]

Mahesh, B. (2020). Machine learning algorithms review. International Journal of Science and Research (IJSR).[Internet], 9, 381-386

2020
[39]

Fourie, C. M. (2003). Deep learning? What deep learning?: research in higher education. South African Journal of Higher Education, 17(1), 123-131

2003
[40]

G., Kebede, T., Kolesnikova, O., Sidorov, G., & Gelbukh, A

Yigezu, M. G., Kebede, T., Kolesnikova, O., Sidorov, G., & Gelbukh, A. (2023, September). Habesha@ DravidianLangTech: Utilizing Deep and Transfer Learn- ing Approaches for Sentiment Analysis. In Proceedings of the Third Workshop on Speech and Language Technologies for Dravidia...

2023
[41]

G., Kanta, S., Kolesnikova, O., Sidorov, G., & Gelbukh, A

Yigezu, M. G., Kanta, S., Kolesnikova, O., Sidorov, G., & Gelbukh, A. (2023, September). Habesha@ DravidianLangTech: Abusive Comment Detection using Deep Learning Approach. In Proceedings of the Third Workshop on Speech and Language Technologies for Dravidian Languages (pp. 244-249)

2023
[42]

Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., ... and Polosukhin, I. (2017). Attention is all you need. Advances in neural information processing systems, 30

2017
[43]

G., Mehamed, M

Yigezu, M. G., Mehamed, M. A., Kolesnikova, O., Guge, T. K., Gelbukh, A., & Sidorov, G. (2023, October). Evaluating the Effectiveness of Hybrid Features in Fake News Detection on Social Media. In 2023 International Conference on Information and Communication Technology for Dev...

2023
[44]

S., Ahani, Z., Tonja, A., Gemeda, M., Hussain, N., and Kolesnikova, O

Tash, M. S., Ahani, Z., Tonja, A., Gemeda, M., Hussain, N., and Kolesnikova, O. (2022, December). Word Level LID in Code-mixed Kannada-English Texts using traditional machine learning algorithms. In Proceedings of the 19th International Conference on Natural Language Processin...

2022
[45]

G., waldeyohannis, M

Yigezu, M. G., waldeyohannis, M. M., and Tonja, A. L. (2021, November). Mul- tilingual neural machine translation for low resourced languages: Ometo-english. In 2021 International Conference on Information and Communication Technology for Development for Africa (ICT4DA) (pp. 8...

2021

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.

[1] [1]

H., and Baldwin, T

Lui, M., Lau, J. H., and Baldwin, T. (2014). Automatic detection and LID of multi- lingual documents. Transactions of the Association for Computational Linguistics, 2, 27-40

work page 2014

[2] [2]

Alshami, and Jugal Kalita

Mersha, Melkamu, Khang Lam, Joseph Wood, Ali K. Alshami, and Jugal Kalita. ”Explainable artificial intelligence: A survey of needs, techniques, applications, and future direction.” Neurocomputing 599 (2024): 128111

work page 2024

[3] [3]

Lillian Lee, Mark Johnson, Kristina Toutanova, and Brian Roark. 2018. Transac- tions of the Association for Computational Linguistics, Volume 6. Transactions of the Association for Computational Linguistics, 6

work page 2018

[4] [4]

”Semantic-driven topic modeling using transformer-based embeddings and clustering algorithms.” Procedia Computer Sci- ence 244 (2024): 121-132

Mersha, Melkamu Abay, and Jugal Kalita. ”Semantic-driven topic modeling using transformer-based embeddings and clustering algorithms.” Procedia Computer Sci- ence 244 (2024): 121-132

work page 2024

[5] [5]

(2006, May)

Hughes, B., Baldwin, T., Bird, S., Nicholson, J., and MacKinlay, A. (2006, May). Reconsidering LID for Written Language Resources. In LREC (pp. 485-488)

work page 2006

[6] [6]

”Evaluating the effectiveness of XAI techniques for encoder-based language models.” Knowledge- Based Systems 310 (2025): 113042

Mersha, Melkamu Abay, Mesay Gemeda Yigezu, and Jugal Kalita. ”Evaluating the effectiveness of XAI techniques for encoder-based language models.” Knowledge- Based Systems 310 (2025): 113042

work page 2025

[7] [7]

Indhuja, K., Indu, M., Sreejith, C., Sreekrishnapuram, P., and Raj, P. R. (2014). Text based LID system for indian languages following devanagiri script. Interna- tional Journal of Engineering, 3(4)

work page 2014

[8] [9]

F., Dussias, P

Kroll, J. F., Dussias, P. E., Bogulski, C. A., and Kroff, J. R. V. (2012). Juggling two languages in one mind: What bilinguals tell us about language processing and its consequences for cognition. In Psychology of Learning and Motivation (Vol. 56, pp. 229-262). Academic press

work page 2012