Natural Language Processing

We automatically extract the summaries of spoken class lectures. This paper presents a novel method for sentence extraction-based automatic speech summarization. We propose a technique that extracts "cue phrases for important sentences (CPs)" that often appear in important sentences. We formulate CP extraction as a labeling problem of word sequences and use Conditional Random Fields (CRF) for labeling. Automatic summarization using CP extraction results as features yields precisions of 0.603 and 0.556 when using manual transcriptions and Automatic Speech Recognition (ASR) results, respectively. Combining the features derived from the CPs and traditional features (including repeated words, words repeated in a slide text, and term frequency (tf), which are surface linguistic information, and speech power and duration, which are prosodic features) [2, 3], we obtained better summarization performance with a κ-value of 0.380, a F -measure of 0.539, and a Rouge-4 of 0.709.

If a dialog system were to respond to a user as naturally as a human, interaction would be smoother. Imitating the human prosodic behavior of utterances is important in computer-human natural conversations. In this paper, to develop a cooperative/friendly spoken dialog system, we analyzed the correlations between F0 synchrony tendency or overlap frequency and subjective measures: "liveliness," "familiarity," and "informality" in human-human dialogs. We also modeled the properties of these features and implemented the model on our dialog system that generated the response timing of aizuchi (back-channel), turn-taking based on a decision tree in real time, and dynamical F0 changes to realize chat-like conversations.

We propose a method of detecting "task incomplete" dialogs in spoken dialog systems using N-gram-based dialog models. We used a database created during a field test in which inexperienced users used a client-server music retrieval system with a spoken dialog interface on their own PCs. In this study, the dialog for a music retrieval task consisted of a sequence of user and system tags that related their utterances and behaviors. The dialogs were manually classified into two classes: the dialog either completed the music retrieval task or it didn't. We then detected dialogs that did not complete the task, using N-gram probability models or a Support Vector Machine with N-gram feature vectors trained using manually classified dialogs. Offline and on-line detection experiments were conducted on a large amount of real data, and the results show that our proposed method achieved good classification performance.

The aim of this research is to identify local Arabic dialects in texts from social media (Twitter) and link them to specific geographic areas. Dialect identification is studied as a subset of the task of language identification. The proposed method is based on unsupervised learning using simultaneously lexical and geographic distance. While this study focusses on Libyan dialects, the approach is general, and could produce resources to support human translators and interpreters when dealing with vernaculars rather than standard Arabic.

Background: Within the structural and grammatical bounds of a common language, all authors develop their own distinctive writing styles. Whether the relative occurrence of common words can be measured to produce accurate models of authorship is of particular interest. This work introduces a new score that helps to highlight such variations in word occurrence, and is applied to produce models of authorship of a large group of plays from the Shakespearean era. Methodology: A text corpus containing 55,055 unique words was generated from 168 plays from the Shakespearean era (16th and 17th centuries) of undisputed authorship. A new score, CM1, is introduced to measure variation patterns based on the frequency of occurrence of each word for the authors John Fletcher, Ben Jonson, Thomas Middleton and William Shakespeare, compared to the rest of the authors in the study (which provides a reference of relative word usage at that time). A total of 50 WEKA methods were applied for Fletcher, Jonson and Middleton, to identify those which were able to produce models yielding over 90% classification accuracy. This ensemble of WEKA methods was then applied to model Shakespearean authorship across all 168 plays, yielding a Matthews' correlation coefficient (MCC) performance of over 90%. Furthermore, the best model yielded an MCC of 99%. Our results suggest that different authors, while adhering to the structural and grammatical bounds of a common language, develop measurably distinct styles by the tendency to over-utilise or avoid particular common words and phrasings. Considering language and the potential of words as an abstract chaotic system with a high entropy, similarities can be drawn to the Maxwell's Demon thought experiment; authors subconsciously favour or filter certain words, modifying the probability profile in ways that could reflect their individuality and style.

In this paper we present the design and analysis of an intonation model for text-to-speech (TTS) synthesis applications using a combination of Relational Tree (RT ) and Fuzzy Logic (F L) technologies. The model is demonstrated using the Standard Yorùbá (SY ) language. In the proposed intonation model, phonological information extracted from text is converted into an RT. RT is a sophisticated data structure that represents the peaks and valleys as well as the spatial structure of a waveform symbolically in the form of trees. An initial approximation to the RT, called Skeletal Tree (ST), is first generated algorithmically. The exact numerical values of the peaks and valleys on the ST is then computed using FL. Quantitative analysis of the result gives RMSE of 0.56 and 0.71 for peak and valley respectively. Mean Opinion Scores (MOS) of 9.5 and 6.8, on a scale of 1 --10, was obtained for intelligibility and naturalness respectively.

Recent research on sociolinguistics has shown that while the Internet is potentially a multilingual environment where users from different parts of the world can communicate and share information, the way that the Internet is structured makes access more difficult for users who are unfamiliar with the Roman alphabet. To promote language equality within Cyberspace and to encourage such people to use the Internet, a proposal for internationalised domain names (IDN) has been developed. We examine how far this and other proposals meet the needs of all types of Internet users. In the light of our findings, we describe an alternative approach which uses an intelligent URL search engine to enable users to access the Internet using their own language. We have tested our implemented prototype on a small sample of data. The results suggest that our approach provides a viable means of encouraging language diversity on the Internet by overcoming the language barrier posed by conventional Intern...

The lexical representation of the Chinese word '4A: has been an issue of on-going debate. The lexical form suggested by Her seemed to provide a complete representation of the different syntactic behaviours of `4ht.' within a Lexical-Functional Grammar (LFG) account. However, when ...

In recent work on the Lexical-Functional Grammar (LFG) formalism, argument structure (a-structure) and lexical mapping theory have been used to explain many linguistic behaviours across languages. It has been suggested that the combina-tion of c-structure, f-structure ...

Research has shown that lexical databases are good sources of lexical knowledge for various Natural Language Processing (NLP) tasks. Over the years, several lexical databases have been developed, eg HowNet [1], WordNet [2], EuroWordNet [3] and CCD [4]. These knowledge ...

This study explores a large language model (LLM)-based system enhancing design methodology education, focusing on design for environment (DfE). By integrating GPT-4 with DfE guidelines and a dual-loop structured learning workflow, we developed an interactive system addressing persistent challenges in design method learning. Through personalised exercises and systematic feedback, our system guided beginners in applying DfE strategies while fostering self-directed learning. Experiments involving 30 participants revealed significant improvements across Bloom's taxonomy domains, particularly when learners demonstrated sustained or developing proactiveness. This study analysed how LLM-based systems can support education for various design for X methodologies through generalizable interaction frameworks.

In this paper we explore the effects of query and database size on news story classification performance. Memory Based Reasoning (MBR) (a k-nearest neighbor method) used as the classification method. There are 360 different possible codes. Close matches to a new story are found using an already coded training database of about 87,000 stories from the Dow Jones Press Release News Wire, and a Connection-Machine Document Retrieval system (CMDRS, [Stanfill]) that supports full text queries, as the underlying match engine. By combining the codes from the near matches, new stories are coded with a recall of about 80% and precision of about 70%, as reported in [Masand]. When the query size is varied from 10 terms to more than 200 terms (matched against the full-text of documents) the recallprecision product changes from 0.53 to 0.6. While this is a significant change, we find that moderate sized queries of 40-80 terms can suffice for finding relevant matches for classification. By changing the size of database from 10,000 stories to 80,000 we found that the recall-precision product changed from 0.22 to 0.57. This shows that with our current MBR approach the database size can't be reduced significantly without compromising performance. We also find that fewer number of retrieved matches are needed with larger query and database sizes.

The purpose of the present article is an analysis of the possible applications of translation to psycholinguistic research. It is argued that it offers valuable insights into foreign language processing, especially comprehension, the mental representation of words, the interaction of different kinds of information and, last but not least, affective states. However, the use of translation also has some limitations which should be taken into account and, if necessary, it should be combined with other research tools.

SUMMARY In this paper the LUSAS High Precision Moulding (HPM) integrated finite element environment for composite process simulation is introduced and validated against test components with single and double curvatures. Presented results show dependence on the laminate thickness, radius and in-plane dimensions, these agree well with the experimental data.

The objective of this descriptive research was to describe the students' ability in choosing the correct prepositional phrases for adverb phrase. The population of this research was the second year students of the English Study Program of Universitas Riau as big as 86 students of 2014/2015 academic year. Due to the population is very large, which consist of 3 classes, the cluster random sampling technique is used. The whole of number of the population is about 86 students. One of the class is taken as the sample as many as 30 students. The multiple choice test is used to collect the needed data of prepositional phrases for adverb phrase. The collected data were analyzed by calculating the scores of the students in answering the test and classified them to a certain level of ability in choosing the correct prepositional phrases for adverb phrase. The research finding show that the majority of the students passed in choosing the correct prepositional phrases for adverb phrase as b...

Recently, the approach towards mining various opinions on weblogs, forums and websites has gained attentions and interests of numerous researchers. In this regard, feature-based opinion mining has been extensively studied in English documents in order to identify implicit and explicit product features and relevant opinions. However, in case of texts written in Persian language, this task faces serious challenges. The objective of this research is to present an unsupervised method for feature-based opinion mining in Persian; an approach which does not require a labeled training dataset. The proposed method in this paper involves extracting explicit product features. Previous studies dealing with extraction of explicit features often focus on lexical roles of words; the approach which cannot be used in distinguishing between an adjective as a part of a noun or a sentiment word. In this study, in addition to lexical roles, syntactic roles are also considered to extract more relevant explicit features. The results demonstrate that the proposed method has got higher recall and precision values compared to prior studies.

Text mining of scientific literature has been essential for setting up large public biomedical databases, which are being widely used by the research community. In the biomedical domain, the existence of a large number of terminological resources and knowledge bases (KB) has enabled a myriad of machine learning methods for different text mining related tasks. Unfortunately, KBs have not been devised for text mining tasks but for human interpretation, thus performance of KB-based methods is usually lower when compared to supervised machine learning methods. The disadvantage of supervised methods though is they require labelled training data and therefore not useful for large scale biomedical text mining systems. KB-based methods do not have this limitation. In this paper, we describe a novel method to generate word-concept probabilities from a KB, which can serve as a basis for several text mining tasks.

Parallel corpora available for building machine translation (MT) models for dialectal Arabic (DA) are rather limited. The scarcity of resources has prompted the use of Modern Standard Arabic (MSA) abundant resources to complement the limited dialectal resource. However, clitics often differ between MSA and DA. This paper compares morphologyaware DA word segmentation to other word segmentation approaches like Byte Pair Encoding (BPE) and Sub-word Regularization (SR). A set of experiments conducted on Egyptian Arabic (EA), Levantine Arabic (LA), and Gulf Arabic (GA) show that a sufficiently accurate morphology-aware segmentation used in conjunction with BPE or SR outperforms the other word segmentation approaches.

In an era defined by globalization, language barriers persist as formidable obstacles to equitable participation in business, education, healthcare, and cultural exchange. This article explores how advancements in multilingual natural language processing (NLP) and voice recognition technologies are dismantling these barriers, enabling real-time translation, cross-cultural collaboration, and inclusive access to digital services. From neural machine translation models like mBERT and XLM-R to speech-to-text systems such as OpenAI's Whisper, these tools empower individuals and organizations to communicate seamlessly across linguistic divides. However, their transformative potential is tempered by ethical and technical challenges, including algorithmic bias, data scarcity for underrepresented languages, and infrastructure gaps in low-resource regions. Through case studies in crisis response, education, and cultural preservation, the article underscores the societal impact of multilingual AI while advocating for inclusive development, equitable resource distribution, and community-led innovation. By prioritizing linguistic diversity and ethical governance, stakeholders can harness these technologies to foster global empathy, drive innovation, and redefine communication as a universal human right.

Modern linguistics plays an important role in learning Arabic language skills. So this study aims to describe the role of modern linguistics in learning Arabic language skills. This study uses a qualitative approach with descriptive methods with the type of library research. The results of this study indicate that modern linguistics has an important role in learning Arabic language skills, namely in phonological analysis, understanding, analyzing, and applying phonological principles in studying sound systems in Arabic; analysis of morphology and word formation, to understand morphological structures, word formation rules, morpheme meanings, morphological variations, and morphological patterns in Arabic; analysis of syntax and sentence structure, to analyze syntax and sentence structure in more depth; analysis of semantics and word meanings, to understand word meanings and semantic structures in language; and pragmatic analysis and use of language to understand how language is used ...

The present study intended to investigate, first, the impact of semantic clustering on the recall and recognition of incidentally learned words in a new language, and second, how the interaction between semantic clustering and frequency of occurrence may modulate learning. To that end, Spanish university students watched an intentionally created video which contained Spanish target words that were either semantically related to others of the set, or not semantically linked at all. Furthermore, frequency of appearance changed among target words (1|4|8). All these words were paired with pseudowords that appeared as on-screen text during the videos. Participants were completely naive to the phases and the procedure of the experiment. After viewing the video, participants completed a recall test and a recognition test. Results showed that words presented in semantically unrelated categories were better recalled and better recognized than those presented in semantic clusters, especially ...

Most of the software companies needs to deal with large number of software bugs every day. Software bugs are unavoidable and fixing software bugs is an expensive task. The goal of effective bug triaging software is to assign potentially experienced developers to new-coming bug reports. To reduce time and cost of bug triaging, an automatic approach is proposed in this paper that predicts a developer with relevant experience to solve or fix the new coming bug report. In this paper, the five term selection methods on the accuracy of bug assignment are used. In addition, the load between developers based on their experience is re-balanced. The proposed system is built with intention to suggest or recommend the bug and not to automatically assign it. This allows a window to handle real time crisis that come up during project development lifecycle. Keywords— Mining software repositories, application of data pre-processing, data management in bug repositories, bug data reduction, feature s...

Recognition of Arabic-like scripts such as Persian and Urdu is more challenging than Latin-based scripts. This is due to the presence of a two-dimensional structure, context-dependent character shapes, spaces and overlaps, and placement of diacritics. Not much research exists for offline handwritten Urdu script which is the 10th most spoken language in the world. We present an attention based encoder-decoder model that learns to read Urdu in context. A novel localization penalty is introduced to encourage the model to attend only one location at a time when recognizing the next character. In addition, we comprehensively refine the only complete and publicly available handwritten Urdu dataset in terms of ground-truth annotations. We evaluate the model on both Urdu and Arabic datasets and show that contextual attention localization outperforms both simple attention and multi-directional LSTM models.

When the amount of training data is limited, the successful application of machine learning techniques typically hinges on the ability to identify useful features or abstractions. Expert knowledge often plays a crucial role in this feature engineering process. However, manual creation of such abstractions can be labor intensive and expensive. In this paper, we propose a feature learning framework that takes advantage of the vast amount of expert knowledge available in unstructured form on the Web. We explore the use of unsupervised learning techniques and non-Euclidean distance measures to automatically incorporate such expert knowledge when building feature representations. We demonstrate the utility of our proposed approach on the task of learning useful abstractions from a list of over two thousand patient medications. Applied to three clinically relevant patient risk stratification tasks, the classifiers built using the learned abstractions outperform several baselines including one based on a manually curated feature space.

In this paper, we introduce an approach to automatically convert simple modern standard Arabic children's stories to the best representative images that can efficiently illustrate the meaning of words. It is a kind of imitating the imaginative process when children read a story, yet a great challenge for a machine to achieve it. For simplification issues, we apply several techniques to find the images and we associate them with related words dynamically. First, we apply natural language processing techniques to analyze the text in stories and we extract keywords of all characters and events in each sentence. Second, we apply an image captioning process through a pre-trained deep learning model for all retrieved images from our multimedia database as well as the Google search engine. Third, using sentence similarities, most significant images are retrieved back by selecting top-k highest similarity values. It is worth mentioning that using the captioning process, to rank top-k images, has shown reasonable precision values as per our preliminary results. The option to refine or validate the ranked images to compose the final visualization for each story is also provided to ensure a flexible and safe learning environment. INDEX TERMS Robust machine learning, automated Arabic text illustration, mapping text to multimedia, visualization, deep learning model.

This paper discusses the debatable hypotheses of “Translation Universals”, i. e. the recurring common features of translated texts in relation to original utterances. We propose that, if translational language does have some distinctive linguistic features in contrast to non-translated writings in the same language, those differences should be statistically significant, consistently distributed and systematically co-occurring across registers and genres. Based on the balanced Corpus of Translational English (COTE) and its non-translated English counterpart, the Freiburg-LOB corpus of British English (FLOB), and by deploying a multi-feature statistical analysis on 96 lexical, syntactic and textual features, we try to pinpoint those distinctive features in translated English texts. We also propose that the stylo-statistical model developed in this study will be effective not only in analysing the translational variation of English but also be capable of clustering those variational fe...

This paper offers a novel account of polysemous copredicative sentences. The solution, which it is argued enjoys a number of advantages over the alternative accounts currently on the market, is inspired by Donald Davidson’s first attempt to deal with ambiguity. Specifically, the account involves mapping ambiguities in the object language (in this case polysemous singular terms) onto ambiguities in the metalanguage. If this account is coherent and superior to its rivals, it tells us something important about logical form: the value of logical form does not lie in the elimination all lexical ambiguity.

В статье рассматривается взаимосвязь романа Б. Акунина «Алмазная колесница» с другими текстами русской и мировой литературы: на уровне собственно интертекстуальности - присутствия цитат, аллюзий, реминисценций в одном тексте, паратекстуальности (отношение текста романа к названию первого тома «Ловец стрекоз»), метатекстуальности (сопоставление с рассказом А. Куприна «Штабс-капитан Рыбников»). Интертекстуальная поэтика включает актуализацию культурных кодов – русского (составной частью которого является пушкинский код) и японского, отражая тем самым противостояние Востока и Запада.The article is devoted to the interconnection of the novel by B. Akunin "The Diamond Chariot" with other texts of Russian and world literature: at the level of actual intertextuality - the presence of a quotation, allusions, reminiscences in one text, paratextuality (the relation of the text of the novel to the title of the first volume "The Dragonfly Catcher"), metatextuality (compariso...

This report presents the work carried out at NLE Lab for the QA@CLEF-2009 competition. We used the JIRS passage retrieval system, which is based on redundancy, with the assumption that it is possible to find the response to a question in a large enough document collection. The retrieved passages are ranked depending on the number, length and position of the question n-gram structures found in the passages. The best results were obtained in monolingual English, while the worst results were obtained for French. We suppose the difference is due to the question style that varies considerably from one language to another.

The aim of this paper is to investigate the role of Luxembourgish adjectives in expressing sentiments in user comments written at the web presence of rtl.lu (RTL is the abbreviation for Radio Television Lëtzebuerg). Alongside many textual features or representations, adjectives could be used in order to detect sentiment, even on a sentence or comment level. In fact, they are also by themselves one of the best ways to describe a sentiment, despite the fact that other word classes such as nouns, verbs, adverbs or conjunctions can also be utilized for this purpose. The empirical part of this study focuses on a list of adjectives which were extracted from an annotated corpus. The corpus contains the part of speech tags of individual words and sentiment annotation on the adjective, sentence and comment level. Suffixes of Luxembourgish adjectives like -esch, -eg, -lech, -al, -el, -iv, -ent, -los, -bar and the prefix on-were explicitly investigated, especially by paying attention to their role in regards to building a model by applying classical machine learning techniques. We also considered the interaction of adjectives with other grammatical means, especially other part of speeches, e.g. negations, which can completely reverse the meaning, thus the sentiment of an utterance.

Despite some recent work [17, p. 5], the ongoing research for the processing of Luxembourgish is still largely in its infancy. While a rich variety of linguistic processing tools exist, especially for English, these software tools offer little scope for the Luxembourgish language. LuNa (a Tool for Luxembourgish National Corpus) is an Open Toolbox that allows researchers to annotate a text corpus written in Luxembourgish language and to build/query an annotated corpus. The aim of the paper is to demonstrate the components of the system and its usage for Machine Learning applications like Topic Modelling and Sentiment Detection. Overall, LuNa bases on a XML-database to store the data and to define the XML scheme, it offers a Graphical User Interface (GUI) for a linguistic data preparation such as tokenization, Part-Of-Speech tagging, and morphological analysis -just to name a few.

Information is commonly reflected in news articles. However, texts are unstructured and thus demanding to analyze automatically. To identify and capture the facts in a news story we propose a novel approach, which utilizes natural language engineering. A combination of selected linguistic and statistical criteria enables the identification of grammatical units such as noun, verb, adjective, and adverb phrases. In literature, these entities are presumed to carry the meaning and the information expressed in English texts. In our study, we focus on determining multi-word features in articles related to the monetary policy conducted by the central bank in the USA, FED. The features are composed as attribute-value pairs, where the attributes represent grammatical units, which quantify the major event characteristics. The corresponding values are conditional expressions, which vary over time as facts evolve. The final set is aggregated over the corpus by the application of heuristic and syntax-based rules. Financial experts contributed to the project by providing expertise for the document interpretation.

Part of Speech Tagging automatically tags the word of a text by labels that can be used to determine the structure of sentence. In this paper we propose an approach to the problem that is inspired from human behavior. We used a combination of rule based and dictionary based approach to tackle this problem. Our goal in this paper is to design a simple yet effective system to POS tagging that also helps us in more effective understanding of human behavior.

Le traitement automatique de la langue quechua (TALQ) ne dispose pas actuellement d'un dictionnaire électronique des verbes, du français-quechua. Pourtant, un projet visant la traduction automatique nécessite au préalable, entre autres, cette importante ressource.Cette thèse consiste à élaborer un dictionnaire électronique français-quechua des verbes. La réalisation d'un tel dictionnaire peut ouvrir également de nouvelles perspectives dans l'enseignement à distance, dans les domaines de l'accès multilingue aux informations, l'annotation/l'indexation des documents, la correction orthographique et pour le TAL en général. La première difficulté consiste à sélectionner un dictionnaire français comme base de travail. Parmi les nombreux dictionnaires français, il en existe très peu en format électronique, et moins encore ceux dont les sources soient en libre accès au public. Parmi ces derniers, l'ouvrage Les verbes français (LVF), contenant 25 610 sens verbaux,...

Le traitement automatique de la langue quechua (TALQ) ne dispose pas actuellement d'un dictionnaire electronique des verbes, du francais-quechua. Pourtant, un projet visant la traduction automatique necessite au prealable, entre autres, cette importante ressource.Cette these consiste a elaborer un dictionnaire electronique francais-quechua des verbes. La realisation d'un tel dictionnaire peut ouvrir egalement de nouvelles perspectives dans l'enseignement a distance, dans les domaines de l'acces multilingue aux informations, l'annotation/l'indexation des documents, la correction orthographique et pour le TAL en general. La premiere difficulte consiste a selectionner un dictionnaire francais comme base de travail. Parmi les nombreux dictionnaires francais, il en existe tres peu en format electronique, et moins encore ceux dont les sources soient en libre acces au public. Parmi ces derniers, l'ouvrage Les verbes francais (LVF), contenant 25 610 sens verbaux,...

Scopi: rilevazioni di frasi fisse o parti fisse di una frase nel testo omerico « Iliade-Odissea» Analisi dei tratti e rilevazioni: meccanismi retorici linguistici fissi :unità lessicali superiore, frasi idiomatiche metafora OBIETTIVI Individuare le forme fisse o parzialmente fisse nella struttura omerica Riprodurre ed implementare in codici linguistici e binarie le forme fisse Valorizzazione delle unità lessicali del DPH Formulazione dei tratti prescelti in codice formale linguistico (lessico-Grammatica) Analisi trasformazionale e distribuzionale in ambiente linguistico ad alto calcolo (Nooj)