Social media offers the potential to provide detection of outbreaks or public health incidents faster than traditional reporting mechanisms. In this paper, we developed and tested a pipeline to produce alerts of influenza-like illness (ILI) using Twitter data. Data was collected from the Twitter API, querying keywords referring to ILI symptoms and geolocated to Wales. Tweets that described first-hand descriptions of symptoms (as opposed to non-personal descriptions) were classified using transformer-based language models specialised on social media (BERTweet and TimeLMs), which were trained on a manually labelled dataset matching the above criteria. After gathering this data, weekly tweet counts were applied to the regression-based Noufaily algorithm to identify exceedances throughout 2022. The algorithm was also applied to counts of ILI-related GP consultations for comparison. Exceedance detection applied to the classified tweet counts produced alerts starting four weeks earlier than by using GP consultation data. These results demonstrate the potential to facilitate advanced preparedness for unexpected increases in healthcare burdens.

This study explores the differences between textual and multimodal sentiment annotations on videos and their impact on transcript-based sentiment modelling. Using the UniC and CH-SIMS datasets which are annotated at both the unimodal and multimodal level, we conducted a statistical analysis and sentiment modelling experiments. Results reveal significant differences between the two annotation types, with textual annotations yielding better performance in sentiment modelling and demonstrating superior generalization ability. These findings highlight the challenges of cross-modality generalization and provide insights for advancing sentiment analysis.

A formidable challenge regarding scraped corpora of social media is the omission of whitespaces, causing pairs of words to be conflated together as one. In order for the text to be properly parsed and analyzed, these missing spaces must be detected and restored. However, it is particularly hard to restore whitespace in languages such as Hebrew which are written without vowels, because a conflated form can often be split into multiple different pairs of valid words. Thus, a simple dictionary lookup is not feasible. In this paper, we present and evaluate a series of neural approaches to restore missing spaces in scraped Hebrew social media. Our best all-around method involved pretraining a new character-based BERT model for Hebrew, and then fine-tuning a space restoration model on top of this new BERT model. This method is blazing fast, high-performing, and open for unrestricted use, providing a practical solution to process huge Hebrew social media corpora with a consumer-grade GPU. We release the new BERT model and the fine-tuned space-restoration model to the NLP community.

This paper addresses the problem of identifying and analyzing ‘noisy’ spelling errors in texts written by second language (L2) learners’ texts in a written corpus. Using Python, spelling errors were identified in 5774 texts greater than or equal to 66 words (total=1,814,209 words), selected from a corpus of 4.2 million words (Authors-1). The statistical analysis used hurdle() models in R, which are appropriate for non-normal, count data, with many zeros.

This study explores the automatic normalization of noisy and highly technical anomaly reports by an LLM. Different prompts are tested to instruct the LLM to clean the text without changing the structure, vocabulary or specialized lexicon. The evaluation of this task is made in two steps. First, the Character Error Rate (CER) is calculated to assess the changes made compared to a gold standard on a small sample. Second, an automatic sequence labeling task is performed on the original and on the corrected datasets with a transformer-based classifier. If some configurations of LLM and prompts can reach satisfying CER scores, the sequence labeling task shows that the normalization has a small negative impact on performance.

We present a suite of experiments that allow us to understand the underlying challenges of language model adaptation to nonstandard text. We do so by designing interventions that approximate core features of user-generated text and their interactions with existing biases of language models. Applying our interventions during language model adaptation to nonstandard text variations, we gain important insights into when such adaptation is successful, as well as the aspects of text variation and noise that are particularly difficult for language models to handle. For instance, on text with character-level variation, out-of-the-box performance improves even with a few additional training examples but approaches a plateau, suggesting that more data is not the solution. In contrast, on text with variation involving new words or meanings, far more data is needed, but it leads to a massive breakthrough in performance. Our findings reveal that existing models lack the necessary infrastructure to handle diverse forms of nonstandard text, guiding the development of more resilient language modeling techniques. We make the code for our interventions, which can be applied to any English text data, publicly available.

This paper investigates whether Large Language Models (LLMs), fine-tuned on synthetic but domain-representative data, can perform the twofold task of (i) slot and intent detection and (ii) natural language response generation for a smart home assistant, while running solely on resource-limited, CPU-only edge hardware. We fine-tune LLMs to produce both JSON action calls and text responses. Our experiments show that 16-bit and 8-bit quantized variants preserve high accuracy on slot and intent detection and maintain strong semantic coherence in generated text, while the 4-bit model, while retaining generative fluency, suffers a noticeable drop in device-service classification accuracy. Further evaluations on noisy human (non-synthetic) prompts and out-of-domain intents confirm the models’ generalization ability, obtaining around 80–86% accuracy. While the average inference time is 5–6 seconds per query—acceptable for one-shot commands but suboptimal for multi-turn dialogue—our results affirm that an on-device LLM can effectively unify command interpretation and flexible response generation for home automation without relying on specialized hardware.

Detecting cynical comments in online communication poses a significant challenge in human-computer interaction, especially given the massive proliferation of discussions on platforms like YouTube. These comments often include offensive or disruptive patterns, such as sarcasm, negative feelings, specific reasons, and an attitude of being right. To address this problem, we present a web platform for the Spanish language that has been developed and leverages natural language processing and machine learning techniques. The platform detects comments and provides valuable information to users by focusing on analyzing comments. The core models are based on pre-trained architectures, including BETO, SpanBERTa, Multilingual BERT, RoBERTuito, and BERT, enabling robust detection of cynical comments. Our platform was trained and tested with Spanish comments from car analysis channels on YouTube. The results show that models achieve performance above 0.8 F1 for all types of cynical comments in the text classification task but achieve lower performance (around 0.6-0.7 F1) for the more arduous token classification task.

Controlled text generation, originally a task to generate coherent, contextually relevant text with specified attributes such as sentiment, topic, or style, has seen a lot of development with methods that use PPLM, FUDGE, and diffusion-based models. However, most state-of-the-art methods balance control precision with fluency. Classifier-guided approaches, like PPLM, are well-known for unstable updates of gradients, yielding incoherent outputs, while autoregressive models, like FUDGE, depend on rigid templates that limit creativity. While recent diffusion models show promises in iterative refinement and diversity, they often lack mechanisms to explicitly incorporate task-specific knowledge and hence require various complicated auxiliary classifiers for training and inference.We now propose a prompt-guided diffusion framework that integrates structured prompts seamlessly into the process of diffusion for precise and flexible control of generated texts.Each prompt combines a target condition (e.g., sentiment label), an in-class example (e.g., a positive movie review), and a placeholder for the generated sentence. Explicit, human-readable guidance is thereby given, spanning high-level intent to low-level text generation.Our approach encodes prompts using large pre-trained language models, e.g., BART, fusing these in a cross-attention manner with the diffusion dynamics, achieves new state-of-the-art results for all benchmarks, including IMDB for sentiment, AG News for topic, and E2E for structured data-to-text generation.

We introduce FaBERT, a Persian BERT-base model pre-trained on the HmBlogs corpus, encompassing both informal and formal Persian texts. FaBERT is designed to excel in traditional Natural Language Understanding (NLU) tasks, addressing the intricacies of diverse sentence structures and linguistic styles prevalent in the Persian language. In our comprehensive evaluation of FaBERT on 12 datasets in various downstream tasks, encompassing Sentiment Analysis (SA), Named Entity Recognition (NER), Natural Language Inference (NLI), Question Answering (QA), and Question Paraphrasing (QP), it consistently demonstrated improved performance, all achieved within a compact model size. The findings highlight the importance of utilizing diverse corpora, such as HmBlogs, to enhance the performance of language models like BERT in Persian Natural Language Processing (NLP) applications.

Evaluating machine translation (MT) of user-generated content (UGC) involves unique challenges such as checking whether the nuance of emotions from the source are preserved in the target text. Recent studies have proposed emotion-related datasets, frameworks and models to automatically evaluate MT quality of Chinese UGC, without relying on reference translations. However, whether these models are robust to the challenge of preserving emotional nuances has been left largely unexplored. To this end, we introduce a novel method inspired by information theory which generates challenging Chinese homophone words related to emotions, by leveraging the concept of *self-information*. Our approach generates homophones that were observed to cause translation errors in emotion preservation, and exposes vulnerabilities in MT models struggling to preserve relevant emotions. We evaluate the efficacy of our method using human evaluation and compare it with an existing one, showing that our method achieves higher correlation with human judgments. The generated Chinese homophones, along with their manual translations, are utilized to generate perturbations and to probe the robustness of existing quality evaluation models, including models trained using multi-task learning, fine-tuned variants of multilingual language models, as well as large language models (LLMs). Our results indicate that LLMs with larger size exhibit higher stability and robustness to such perturbations. We release our data and code for reproducibility and further research.

Multi-domain text analysis presents significant challenges, particularly in Persian name entity recognition (NER). Using a single model for multiple domains often fails to capture the specific features of different domains. That is why many scientists have focused on prompting chatbots for this issue. However, studies show that these models do not achieve remarkable results in NER tasks without proper fine-tuning while training and storing a chatbot is extremely costly. This paper presents a new approach using one core model with various sets of domain-specific parameters. By using techniques like LoRAs and pre-fix tuning, along with extra layers, we train each set of trainable parameters for a specific domain. This allows the model to perform as well as individual models for each domain. Tests on various formal and informal datasets show that by using these added parameters, the proposed model performs much better than existing practical models. The model needs only one instance for storage but achieves excellent results across all domains. This paper also examines each adaptation strategy, outlining its strengths, weaknesses, and the best settings and hyperparameters for Persian NER. Lastly, this study introduces a new document-based domain detection system for situations where text domains are unknown. This novel pipeline enhances the adaptability and practicality of the proposed approach for real-world applications.

Named Entity Recognition (NER), a fundamental task in Natural Language Processing (NLP), has shown significant advancements for high-resource languages. However, due to a lack of annotated datasets and limited representation in Pre-trained Language Models (PLMs), it remains understudied and challenging for low-resource languages. To address these challenges, in this paper, we propose a data augmentation technique that generates culturally plausible sentences and experiments on four low-resource Pakistani languages; Urdu, Shahmukhi, Sindhi, and Pashto. By fine-tuning multilingual masked Large Language Models (LLMs), our approach demonstrates significant improvements in NER performance for Shahmukhi and Pashto. We further explore the capability of generative LLMs for NER and data augmentation using few-shot learning.

Automated content moderation for collaborative knowledge hubs like Wikipedia or Wikidata is an important yet challenging task due to multiple factors. In this paper, we construct a database of discussions happening around articles marked for deletion in several Wikis and in three languages, which we then use to evaluate a range of LMs on different tasks (from predicting the outcome of the discussion to identifying the implicit policy an individual comment might be pointing to). Our results reveal, among others, that discussions leading to deletion are easier to predict, and that, surprisingly, self-produced tags (keep, delete or redirect) don’t always help guiding the classifiers, presumably because of users’ hesitation or deliberation within comments

We present ongoing research on automatic post-processing approaches to enhance the readability of noisy speech transcripts in low-resource languages, with a focus on conversational speech in Latvian. We compare transformer-based sequence-labeling models and large language models (LLMs) for the standard punctuation and capitalization restoration task, while also considering automatic correction of mispronounced words and disfluency, and partial inverse text normalization. Our results show that very small LLMs (approx. 2B parameters), fine-tuned on a modest text corpus, can achieve near state-of-the-art performance, rivaling orders of magnitude larger LLMs. Additionally, we demonstrate that a fine-tuned Whisper model, leveraging acoustic cues, outperforms text-only systems on challenging conversational data, even for a low-resource language. Error analysis reveals recurring pitfalls in sentence boundary determination and disfluency handling, emphasizing the importance of consistent annotation and domain adaptation for robust post-processing. Our findings highlight the feasibility of developing efficient post-processing solutions that significantly refine ASR output in low-resource settings, while opening new possibilities for editing and formatting speech transcripts beyond mere restoration of punctuation and capitalization.

We manually normalize noisy Japanese expressions on social networking services (SNS) to improve the performance of sentiment polarity classification.Despite advances in pre-trained language models, informal expressions found in social media still plague natural language processing.In this study, we analyzed 6,000 posts from a sentiment analysis corpus for Japanese SNS text, and constructed a text normalization taxonomy consisting of 33 types of editing operations.Text normalization according to our taxonomy significantly improved the performance of BERT-based sentiment analysis in Japanese.Detailed analysis reveals that most types of editing operations each contribute to improve the performance of sentiment analysis.