Arabic diacritic recovery i.e. diacritization is necessary for proper vocalization and an enabler for downstream applications such as language learning and text to speech. Diacritics come in two varieties, namely: core-word diacritics and case endings. In this paper we introduce a highly effective morphologically informed character-level model that can recover both types of diacritics simultaneously. The model uses a Recurrent Neural Network (RNN) based architecture that takes in text as a sequence of characters, with markers for morphological segmentation, and outputs a sequence of diacritics. We also introduce a character-based morphological segmentation model that we train for Modern Standard Arabic (MSA) and dialectal Arabic. We demonstrate the efficacy of our diacritization model on Classical Arabic, MSA, and two dialectal (Moroccan and Tunisian) texts. We achieve the lowest reported word-level diacritization error rate for MSA (3.4%), match the best results for Classical Arabic (5.4%), and report competitive results for dialectal Arabic.

Data availability is crucial for advancing artificial intelligence applications, including voice-based technologies. As content creation, particularly in social media, experiences increasing demand, translation and text-to-speech (TTS) technologies have become essential tools. Notably, the performance of these TTS technologies is highly dependent on the quality of the training data, emphasizing the mutual dependence of data availability and technological progress. This paper introduces an end-to-end tool to generate high-quality datasets for text-to-speech (TTS) models to address this critical need for high-quality data. The contributions of this work are manifold and include: the integration of language-specific phoneme distribution into sample selection, automation of the recording process, automated and human-in-the-loop quality assurance of recordings, and processing of recordings to meet specified formats. The proposed application aims to streamline the dataset creation process for TTS models through these features, thereby facilitating advancements in voice-based technologies.

Arabic is among the most challenging languages in the world. Unfortunately, the scarcity of Arabic datasets makes studies in Arabic speech technology demanding. This paper introduces SADA, the Saudi Audio Dataset for Arabic, with 668 hours of high-quality audio suitable for supervised training. The audio recordings were sourced from 57 television shows provided by the Saudi Broadcasting Authority. The audio covers both read and spontaneous speaking styles in various genres. The National Center for Artificial Intelligence in Saudi Arabia transcribed and prepared the data for training and processing. The recordings are in Arabic. Most are in Saudi dialects, while other Arabic dialects include Yemeni, Egyptian, and Levantine. The dataset is split into training, validation, and testing sets to enhance its usage. The validation and testing sets contain 10 hours of audio segments each. Besides giving a detailed description of the dataset, wide range of speech recognition experiments using standard tools are also presented.

Emotion detection can provide us with a window into understanding human behavior. Due to the complex dynamics of human emotions, however, constructing annotated datasets to train automated models can be expensive. Thus, we explore the efficacy of cross-lingual approaches that would use data from a source language to build models for emotion detection in a target language. We compare three approaches, namely: i) using inherently multilingual models; ii) translating training data into the target language; and iii) using an automatically tagged parallel corpus. In our study, we consider English as the source language with Arabic and Spanish as target languages. We study the effectiveness of different classification models such as BERT and SVMs trained with different features. Our BERT-based monolingual models that are trained on target language data surpass state-of-the-art (SOTA) by 4% and 5% absolute Jaccard score for Arabic and Spanish respectively. Next, we show that using cross-lingual approaches with English data alone, we can achieve more than 90% and 80% relative effectiveness of the Arabic and Spanish BERT models respectively. Lastly, we use LIME to analyze the challenges of training cross-lingual models for different language pairs.

The performance of Machine Translation (MT) systems varies significantly with inputs of diverging features such as topics, genres, and surface properties. Though there are many MT evaluation metrics that generally correlate with human judgments, they are not directly useful in identifying specific shortcomings of MT systems. In this demo, we present a benchmarking interface that enables improved evaluation of specific MT systems in isolation or multiple MT systems collectively by quantitatively evaluating their performance on many tasks across multiple domains and evaluation metrics. Further, it facilitates effective debugging and error analysis of MT output via the use of dynamic filters that help users hone in on problem sentences with specific properties, such as genre, topic, sentence length, etc. The interface can be extended to include additional filters such as lexical, morphological, and syntactic features. Aside from helping debug MT output, it can also help in identifying problems in reference translations and evaluation metrics.

User stance detection entails ascertaining the position of a user towards a target, such as an entity, topic, or claim.Recent work that employs unsupervised classification has shown that performing stance detection on vocal Twitter users, who have many tweets on a target, can be highly accurate (+98%).However, such methods perform poorly or fail completely for less vocal users, who may have authored only a few tweets about a target.In this paper, we tackle stance detection for such users using two approaches.In the first approach, we improve user-level stance detection by representing tweets using contextualized embeddings, which capture latent meanings of words in context.We show that this approach outperforms two strong baselines and achieves 89.6% accuracy and 91.3% macro F-measure on eight controversial topics.In the second approach, we expand the tweets of a given user using their Twitter timeline tweets, which may not be topically relevant, and then we perform unsupervised classification of the user, which entails clustering a user with other users in the training set.This approach achieves 95.6% accuracy and 93.1% macro F-measure.

Diacritics (short vowels) are typically omitted when writing Arabic text, and readers have to reintroduce them to correctly pronounce words. There are two types of Arabic diacritics: The first are core-word diacritics (CW), which specify the lexical selection, and the second are case endings (CE), which typically appear at the end of word stems and generally specify their syntactic roles. Recovering CEs is relatively harder than recovering core-word diacritics due to inter-word dependencies, which are often distant. In this article, we use feature-rich recurrent neural network model that use a variety of linguistic and surface-level features to recover both core word diacritics and case endings. Our model surpasses all previous state-of-the-art systems with a CW error rate (CWER) of 2.9% and a CE error rate (CEER) of 3.7% for Modern Standard Arabic (MSA) and CWER of 2.2% and CEER of 2.5% for Classical Arabic (CA). When combining diacritized word cores with case endings, the resultant word error rates are 6.0% and 4.3% for MSA and CA, respectively. This highlights the effectiveness of feature engineering for such deep neural models.

Discovering the stances of media outlets and influential people on current, debatable topics is important for social statisticians and policy makers.Many supervised solutions exist for determining viewpoints, but manually annotating training data is costly.In this paper, we propose a cascaded method that uses unsupervised learning to ascertain the stance of Twitter users with respect to a polarizing topic by leveraging their retweet behavior; then, it uses supervised learning based on user labels to characterize both the general political leaning of online media and of popular Twitter users, as well as their stance with respect to the target polarizing topic.We evaluate the model by comparing its predictions to gold labels from the Media Bias/Fact Check website, achieving 82.6% accuracy.

Abstract This work introduces robust multi-dialectal part of speech tagging trained on an annotated data set of Arabic tweets in four major dialect groups: Egyptian, Levantine, Gulf, and Maghrebi. We implement two different sequence tagging approaches. The first uses conditional random fields (CRFs), while the second combines word- and character-based representations in a deep neural network with stacked layers of convolutional and recurrent networks with a CRF output layer. We successfully exploit a variety of features that help generalize our models, such as Brown clusters and stem templates. Also, we develop robust joint models that tag multi-dialectal tweets and outperform uni-dialectal taggers. We achieve a combined accuracy of 92.4% across all dialects, with per dialect results ranging between 90.2% and 95.4%. We obtained the results using a train/dev/test split of 70/10/20 for a data set of 350 tweets per dialect.

Predicting the stance of social media users on a topic can be challenging, particularly for users who never express explicit stances. Earlier work has shown that using users' historical or non-relevant tweets can be used to predict stance. We build on prior work by making use of users' interaction elements, such as retweeted accounts and mentioned hashtags, to compute the similarities between users and to classify new users in a user similarity feature space. We show that this approach significantly improves stance prediction on two datasets that differ in terms of language, topic, and cultural background.

Younes Samih, Mohamed Eldesouki, Mohammed Attia, Kareem Darwish, Ahmed Abdelali, Hamdy Mubarak, Laura Kallmeyer. Proceedings of the 21st Conference on Computational Natural Language Learning (CoNLL 2017). 2017.

There is little doubt about whether social networks play a role in modern protests. This agreement has triggered an entire research avenue, in which social structure and content analysis have been central - but are typically exploited separately.

Yuan Zhang, Chengtao Li, Regina Barzilay, Kareem Darwish. Proceedings of the 2015 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. 2015.

Several recent papers on Arabic dialect identification have hinted that using a word unigram model is sufficient and effective for the task.However, most previous work was done on a standard fairly homogeneous dataset of dialectal user comments.In this paper, we show that training on the standard dataset does not generalize, because a unigram model may be tuned to topics in the comments and does not capture the distinguishing features of dialects.We show that effective dialect identification requires that we account for the distinguishing lexical, morphological, and phonological phenomena of dialects.We show that accounting for such can improve dialect detection accuracy by nearly 10% absolute.

The use of social media has profoundly affected social and political dynamics in the Arab world. In this paper, we explore the Arabic microblogs retrieval. We illustrate some of the challenges associated with Arabic microblog retrieval, which mainly stem from the use of different Arabic dialects that vary in lexical selection, morphology, and phonetics and lack orthographic and spelling conventions. We present some of the required processing for effective retrieval such as improved letter normalization, elongated word handling, stopword removal, and stemming

Social media streams such as Twitter are regarded as faster first-hand sources of information generated by massive users. The content diffused through this channel, although noisy, provides important complement and sometimes even a substitute to the traditional news media reporting. In this paper, we propose a novel unsupervised approach based on topic modeling to summarize trending subjects by jointly discovering the representative and complementary information from news and tweets. Our method captures the content that enriches the subject matter by reinforcing the identification of complementary sentence-tweet pairs. To valuate the complementarity of a pair, we leverage topic modeling formalism by combining a two-dimensional topic-aspect model and a cross-collection approach in the multi-document summarization literature. The final summaries are generated by co-ranking the news sentences and tweets in both sides simultaneously. Experiments give promising results as compared to state-of-the-art baselines.

Searching social content in general and microblogs (aka tweets) in particular has been basic and limited, especially for time-sensitive topics. The currently implemented microblog search on sites such as Twitter is based on simple word matching and retrieves the most recent microblogs that match a given query. Furthermore, a user may obtain hundreds or perhaps thousands of microblogs in response to a given query, leading to information overload. We present a new multidimensional microblog search tool that generates a comprehensive report from microblogs instead of a flat list of recent/relevant microblogs for a given query. Reports may include tag-clouds, topic time series, and most popular and funny microblogs, etc. The tool can be configured for monitoring time-sensitive topics using a set of predefined queries. We demonstrate our system on Arabic and English microblog collections. Additionally, we show a special configuration of the system for monitoring the 2012 Egyptian presidential elections.

Relevance feedback (RFB) involves requesting some user judgments for an initial set of search results and then using these judgments to improve search results. Typical queries may have multiple possible interpretations or facets, only one of which is relevant to a user's need, but top search results may be dominated by one interpretation or facet. Thus, if the user is only given the top results to inspect, none of them may be relevant. One way to solve this is to intentionally diversify the top few results to cover multiple interpretations. This paper proposes the use of density-based clustering for the purpose of results diversification in the context of RFB. Other traditional clustering algorithms are also used for a comparative study. Clustering is compared to a baseline that nominates the top results from an initial ranked list and compared to using the top results after re-ranking using Maximal Marginal Relevance. The results show that density-based clustering achieves the best results with a statistically significant improvement of 12% over the baseline.

Recent library digitization projects attempt to provide large collections of printed material from varying sources in a searchable format. The scanned documents are typically processed using Optical Character Recognition (OCR), which typically introduces errors in the text. This paper proposes a technique for correction of OCR degraded text that is independent of character-level OCR errors, and hence independent of scanned document source. It is based on language modeling in conjunction with a uniform character model that uses edit distance only. The technique compares well to state-of-the-art correction techniques that are based on language modeling and source-specific character error models. Although the proposed technique yielded lower correction effectiveness, its impact on retrieval effectiveness is statistically significant and at par with state-of-the-art correction techniques. The main requirement of the proposed technique is the training of a “good” language model matching genre, style, and temporal coverage. The advantage of being independent of character level errors is clear in applications were printed documents vary in source, font, and degradation level.

Due to the existence of large numbers of legacy documents (such as old books and newspapers), improving retrieval effectiveness for OCR'ed documents continues to be an important problem. This article compares the effect of OCR error correction with and without language modeling and the effect of query garbling with weighted structured queries on the retrieval of OCR degraded Arabic documents. The results suggest that moderate error correction does not yield statistically significant improvement in retrieval effectiveness when indexing and searching using n-grams. Also, reversing error correction models to perform query garbling in conjunction with weighted structured queries yields improved retrieval effectiveness. Lastly, using very good error correction that utilizes language modeling yields the best improvement in retrieval effectiveness.

Structured methods for query term replacement rely on separate estimates of term tes of replacement probabilities. Statistically significantfrequency and document frequency to compute a weight for each query term. This paper reviews prior work on structured query techniques and introduces three new variants that leverage estima improvements in retrieval effectiveness are demonstrated for cross-language retrieval and for retrieval based on optical character recognition when replacement probabilities are used to estimate both term frequency and document frequency.

Searching is inherently a user-centered process; people pose the questions for which machines seek answers, and ultimately people judge the degree to which retrieved documents meet their needs. Rapid development of interactive systems that use queries expressed in one language to search documents written in another poses five key challenges: (1) interaction design, (2) query formulation, (3) cross-language search, (4) construction of translated summaries, and (5) machine translation. This article describes the design of MIRACLE, an easily extensible system based on English queries that has previously been used to search French, German, and Spanish documents, and explains how the capabilities of MIRACLE were rapidly extended to accommodate Cebuano and Hindi. Evaluation results for the cross-language search component are presented for both languages, along with results from a brief full-system interactive experiment with Hindi. The article concludes with some observations on directions for further research on interactive cross-language information retrieval.

Since many Arabic documents are available only in print, automating retrieval from collections of scanned Arabic document images using Optical Character Recognition (OCR) is an interesting problem. Arabic combines rich morphology with a writing system that presents unique challenges to OCR systems. These factors must be considered when selecting terms for automatic indexing. In this paper, alternative choices of indexing terms are explored using both an existing electronic text collection and a newly developed collection built from images of actual printed Arabic documents. Character n-grams or lightly stemmed words were found to typically yield near-optimal retrieval effectiveness, and combining both types of terms resulted in robust performance across a broad range of conditions.