The rapid spread of multimodal fake news (FN) on Online Social Networks (OSNs) threatens digital information ecosystems, particularly in low-resource languages. Existing multimodal fake news detection (FND) methods are largely limited to high-resource settings, restricting their global applicability. We propose an M&M-RAG, a Multilingual & Multimodal Retrieval-Augmented Generation framework, that leverages Large Vision-Language Models (LVLMs) and Large Language Models (LLMs) to verify news claims across English, Chinese and Urdu. M&M-RAG integrates real-time multilingual evidence retrieval, language-aware prompting, and cross-modal reasoning for fact verification. We further propose Multi-Ax-to-Grind Urdu, the first large-scale, multi-domain multimodal benchmark for FND in Urdu. Experiments on typologically diverse monolingual multimodal datasets demonstrate that M&M-RAG achieves state-of-the-art (SOTA) performance, with 94.6% accuracy and 94.2% F1 score, surpassing models such as SpotFake, MPFN, MMCFND, and Semi-FND. The proposed framework remains robust in zero-shot and cross-lingual scenarios under frozen-model inference without task-specific fine-tuning. The results underscore the scalability and interpretability of LVLM-based approaches for combating multimodal misinformation, particularly in under-represented and typologically diverse languages.

Abstract Android malware has become an increasingly important threat to individuals, organizations, and society, posing significant risks to data security, privacy, and infrastructure. As malware evolves in sophistication and complexity, the detection and mitigation of these malicious software instances have become more challenging and time consuming since the required number of features to identify potential malware can be very high. To address this issue, we have developed an effective feature selection methodology for malware detection in Android. The critical concern in the field of malware detection is the complexity of algorithms and the use of features that are used to detect malware. The present paper delivers a methodology for pre-processing datasets to select the most optimal features that will allow detecting malware, while maintaining very high accuracy. The proposed methodology has been tested on two real world datasets and the results indicate that the number of features is significantly reduced from 489 to between 19 and 28 for the first dataset and from 9503 to between 9 and 27 for the second dataset, whilst the accuracy is maintained as if all features were used.

Natural Language Processing (NLP) is the sub-field of Artificial Intelligence that represents and analyses human language automatically. NLP has been employed in many applications, such as information retrieval, information processing and automated answer ranking. Semantic analysis focuses on understanding the meaning of text. Among other proposed approaches, Latent Semantic Analysis (LSA) is a widely used corpus-based approach that evaluates similarity of text based on the semantic relations among words. LSA has been applied successfully in diverse language systems for calculating the semantic similarity of texts. LSA ignores the structure of sentences, i.e., it suffers from a syntactic blindness problem. LSA fails to distinguish between sentences that contain semantically similar words but have opposite meanings. Disregarding sentence structure, LSA cannot differentiate between a sentence and a list of keywords. If the list and the sentence contain similar words, comparing them using LSA would lead to a high similarity score. In this paper, we propose xLSA, an extension of LSA that focuses on the syntactic structure of sentences to overcome the syntactic blindness problem of the original LSA approach. xLSA was tested on sentence pairs that contain similar words but have significantly different meaning. Our results showed that xLSA alleviates the syntactic blindness problem, providing more realistic semantic similarity scores.

A network is a composition of many communities, i.e., sets of nodes and edges with stronger relationships, with distinct and overlapping properties. Community detection is crucial for various reasons, such as serving as a functional unit of a network that captures local interactions among nodes. Communities come in various forms and types, ranging from biologically to technology-induced ones. As technology-induced communities, social media networks such as Twitter and Facebook connect a myriad of diverse users, leading to a highly connected and dynamic ecosystem. Although many algorithms have been proposed for detecting socially cohesive communities on Twitter, mining and related tasks remain challenging. This study presents a novel detection method based on a scalable framework to identify related communities in a network. We propose a multilevel clustering technique (MCT) that leverages structural and textual information to identify local communities termed microcosms. Experimental evaluation on benchmark models and datasets demonstrate the efficacy of the approach. This study contributes a new dimension for the detection of cohesive communities in social networks. The approach offers a better understanding and clarity toward describing how low-level communities evolve and behave on Twitter. From an application point of view, identifying such communities can better inform recommendation, among other benefits.

Learning from outliers and imbalanced data remains one of the major difficulties for machine learning classifiers. Among the numerous techniques dedicated to tackle this problem, data preprocessing solutions are known to be efficient and easy to implement. In this paper, we propose a selective data preprocessing approach that embeds knowledge of the outlier instances into artificially generated subset to achieve an even distribution. The Synthetic Minority Oversampling TEchnique (SMOTE) was used to balance the training data by introducing artificial minority instances. However, this was not before the outliers were identified and oversampled (irrespective of class). The aim is to balance the training dataset while controlling the effect of outliers. The experiments prove that such selective oversampling empowers SMOTE, ultimately leading to improved classification performance.

A forum or social media post can express multiple emotions, such as love, joy or anger. Emotion classification has been proven useful for measuring aspects such as user satisfaction. Despite its usefulness, research in emotion classification is limited, because the task is multi-label and publicly available data sets and lexica are very limited. A number of emotion classifiers for general-domain text have been proposed recently, but only a few for text in the domain of Open Source Software (OSS), such as EmoTxt. In this paper, we explore different lexica and two multi-label algorithms for classifying emotions in text related to OSS. We trained various multi-label classifiers using HOMER and RAkEL on a data set of Stack Overflow posts and a data set of JIRA Issue Tracker comments. The classifiers have been enriched with features derived from different state-of-the-art lexica. We achieved multi-label Micro F-scores up to 0.811 and Subset 0/1 Loss of 0.290. These results represent a statistically significant improvement over the state-of-the-art.

Online Social Media platforms, such as Facebook and Twitter, enable all users, independently of their characteristics, to freely generate and consume huge amounts of data. While this data is being exploited by individuals and organisations to gain competitive advantage, a substantial amount of data is being generated by spam or fake users. One in every 200 social media messages and one in every 21 tweets is estimated to be spam. The rapid growth in the volume of global spam is expected to compromise research works that use social media data, thereby questioning data credibility. Motivated by the need to identify and filter out spam contents in social media data, this study presents a novel approach for distinguishing spam vs. non-spam social media posts and offers more insight into the behaviour of spam users on Twitter. The approach proposes an optimised set of features independent of historical tweets, which are only available for a short time on Twitter. We take into account features related to the users of Twitter, their accounts and their pairwise engagement with each other. We experimentally demonstrate the efficacy and robustness of our approach and compare it to a typical feature set for spam detection in the literature, achieving a significant improvement on performance. In contrast to prior research findings, we observe that an average automated spam account posted at least 12 tweets per day at well defined periods. Our method is suitable for real-time deployment in a social media data collection pipeline as an initial preprocessing strategy to improve the validity of research data.

Descriptive document clustering aims at discovering clusters of semantically interrelated documents together with meaningful labels to summarize the content of each document cluster. In this work, we propose a novel descriptive clustering framework, referred to as CEDL. It relies on the formulation and generation of 2 types of heterogeneous objects, which correspond to documents and candidate phrases, using multilevel similarity information. CEDL is composed of 5 main processing stages. First, it simultaneously maps the documents and candidate phrases into a common co‐embedded space that preserves higher‐order, neighbor‐based proximities between the combined sets of documents and phrases. Then, it discovers an approximate cluster structure of documents in the common space. The third stage extracts promising topic phrases by constructing a discriminant model where documents along with their cluster memberships are used as training instances. Subsequently, the final cluster labels are selected from the topic phrases using a ranking scheme using multiple scores based on the extracted co‐embedding information and the discriminant output. The final stage polishes the initial clusters to reduce noise and accommodate the multitopic nature of documents. The effectiveness and competitiveness of CEDL is demonstrated qualitatively and quantitatively with experiments using document databases from different application fields.

OBJECTIVE: The abundance of text available in social media and health related forums along with the rich expression of public opinion have recently attracted the interest of the public health community to use these sources for pharmacovigilance. Based on the intuition that patients post about Adverse Drug Reactions (ADRs) expressing negative sentiments, we investigate the effect of sentiment analysis features in locating ADR mentions. METHODS: We enrich the feature space of a state-of-the-art ADR identification method with sentiment analysis features. Using a corpus of posts from the DailyStrength forum and tweets annotated for ADR and indication mentions, we evaluate the extent to which sentiment analysis features help in locating ADR mentions and distinguishing them from indication mentions. RESULTS: Evaluation results show that sentiment analysis features marginally improve ADR identification in tweets and health related forum posts. Adding sentiment analysis features achieved a statistically significant F-measure increase from 72.14% to 73.22% in the Twitter part of an existing corpus using its original train/test split. Using stratified 10×10-fold cross-validation, statistically significant F-measure increases were shown in the DailyStrength part of the corpus, from 79.57% to 80.14%, and in the Twitter part of the corpus, from 66.91% to 69.16%. Moreover, sentiment analysis features are shown to reduce the number of ADRs being recognized as indications. CONCLUSION: This study shows that adding sentiment analysis features can marginally improve the performance of even a state-of-the-art ADR identification method. This improvement can be of use to pharmacovigilance practice, due to the rapidly increasing popularity of social media and health forums.

Deciding whether an open source software (OSS) project meets the required standards for adoption in terms of quality, maturity, activity of development and user support is not a straightforward process as it involves exploring various sources of information. Such sources include OSS source code repositories, communication channels such as newsgroups, forums, and mailing lists, as well as issue tracking systems. OSSMETER is an extensible and scalable platform that can monitor and incrementally analyse a large number of OSS projects. The results of this analysis can be used to assess various aspects of OSS projects, and to directly compare different OSS projects with each other.

OBJECTIVE: Drug named entity recognition (NER) is a critical step for complex biomedical NLP tasks such as the extraction of pharmacogenomic, pharmacodynamic and pharmacokinetic parameters. Large quantities of high quality training data are almost always a prerequisite for employing supervised machine-learning techniques to achieve high classification performance. However, the human labour needed to produce and maintain such resources is a significant limitation. In this study, we improve the performance of drug NER without relying exclusively on manual annotations. METHODS: We perform drug NER using either a small gold-standard corpus (120 abstracts) or no corpus at all. In our approach, we develop a voting system to combine a number of heterogeneous models, based on dictionary knowledge, gold-standard corpora and silver annotations, to enhance performance. To improve recall, we employed genetic programming to evolve 11 regular-expression patterns that capture common drug suffixes and used them as an extra means for recognition. MATERIALS: Our approach uses a dictionary of drug names, i.e. DrugBank, a small manually annotated corpus, i.e. the pharmacokinetic corpus, and a part of the UKPMC database, as raw biomedical text. Gold-standard and silver annotated data are used to train maximum entropy and multinomial logistic regression classifiers. RESULTS: Aggregating drug NER methods, based on gold-standard annotations, dictionary knowledge and patterns, improved the performance on models trained on gold-standard annotations, only, achieving a maximum F-score of 95%. In addition, combining models trained on silver annotations, dictionary knowledge and patterns are shown to achieve comparable performance to models trained exclusively on gold-standard data. The main reason appears to be the morphological similarities shared among drug names. CONCLUSION: We conclude that gold-standard data are not a hard requirement for drug NER. Combining heterogeneous models build on dictionary knowledge can achieve similar or comparable classification performance with that of the best performing model trained on gold-standard annotations.

Georgios Kontonatsios, Ioannis Korkontzelos, Jun’ichi Tsujii, Sophia Ananiadou. Proceedings of the 14th Conference of the European Chapter of the Association for Computational Linguistics, volume 2: Short Papers. 2014.

Automatically compiling bilingual dictio-naries of technical terms from comparable corpora is a challenging problem, yet with many potential applications. In this paper, we exploit two independent observations about term translations: (a) terms are of-ten formed by corresponding sub-lexical units across languages and (b) a term and its translation tend to appear in similar lex-ical context. Based on the first observa-tion, we develop a new character n-gram compositional method, a logistic regres-sion classifier, for learning a string similar-ity measure of term translations. Accord-ing to the second observation, we use an existing context-based approach. For eval-uation, we investigate the performance of compositional and context-based methods on: (a) similar and unrelated languages, (b) corpora of different degree of compa-rability and (c) the translation of frequent and rare terms. Finally, we combine the two translation clues, namely string and contextual similarity, in a linear model and we show substantial improvements over the two translation signals. 1