Search

With the current exponential growth of video-based social networks, video retrieval using natural language is receiving ever-increasing attention. Most existing approaches tackle this task by extracting individual frame-level spatial features to represent the whole video, while ignoring visual pattern consistencies and intrinsic temporal relationships across different frames. Furthermore, the semantic correspondence between natural language queries and person-centric actions in videos has not been fully explored. To address these problems, we propose a novel binary representation learning framework, named Semantics-aware Spatial-temporal Binaries ( $\text{S}^{2}$ Bin), which simultaneously considers spatial-temporal context and semantic relationships for cross-modal video retrieval. By exploiting the semantic relationships between two modalities, $\text{S}^{2}$ Bin can efficiently and effectively generate binary codes for both videos and texts. In addition, we adopt an iterative optimization scheme to learn deep encoding functions with attribute-guided stochastic training. We evaluate our model on three video datasets and the experimental results demonstrate that $\text{S}^{2}$ Bin outperforms the state-of-the-art methods in terms of various cross-modal video retrieval tasks.

The aim of this paper is to sort and characterize natural and artificial mixed languages. These are also classified in an educational perspective. The article stressed the fact that natural mixed languages (e.g. pidgin) were not acquired in natural conditions (at home), whereas artificial languages (e.g. volapik) must be learnt at school. Is also discussed interlanguage as a phenomenon present in the process of learning foreign languages, as well as an effect of natural contact of languages. The article reflected on the matter of searching for a universal language, which could be a global means of communication. Furthermore, the essence of lingua franca is explained and the role of English (and its variants Basic English and Plain English) in the contemporary world is emphasized. Celem tekstu jest uporządkowanie i scharakteryzowanie naturalnych i sztucznych języków mieszanych, które podlegają również klasyfikacji „edukacyjnej”. W artykule zwraca się uwagę na to, że naturalne języki mieszane (np. pidgin) nie były poznawane w szkole, zaś sztuczne (np. volapük) można opanować tylko w wyniku uczenia się/nauczania. Omówiono interjęzyk jako zjawisko obecne w procesie dydaktycznym, ale i poza nim – jako efekt naturalnego kontaktu języków. Refleksji poddano kwestię poszukiwań języka uniwersalnego, który służyłby do globalnego porozumiewania się i podjęto próbę wyjaśnienia istoty lingua franca, akcentując rolę angielszczyzny (także jej odmian Basic English i Plain English) we współczesnym świecie.

Term weighting is one of the most commonly used approaches, which works by assigning weights to terms, that aims to improve the performance of information retrieval or text categorization tasks. In this paper, we present a novel term weighting technique, called discriminative feature spamming technique (DFST), which identifies distinctive terms, based on a term utility criteria (TUC), and then spams them to increase their discriminative power. The experimental results show that the DFST outperformed a set of time-tested term weighting schemes, from the information retrieval field. All the experiments were performed on the largest ever Roman Urdu (RU) dataset of 11000 reviews, which was collected and annotated for this work. In addition, a custom tokenizer was built, which further improved classification accuracy. A cross-scheme comparison was performed, which showed that the results obtained by using the newly proposed DFST, were statistically significant and better than previous approaches.

Requirements often specify the expected system behavior by using causal relations (e.g., If A, then B). Automatically extracting these relations supports, among others, two prominent RE use cases: Automatic test case derivation and dependency detection between requirements. However, existing tools fail to extract causality from natural language with reasonable performance. In this paper, we present our tool CiRA (Causality detection in Requirements Artifacts), which represents a first step towards automatic causality extraction from requirements. We evaluate CiRA on a publicly available data set of 61 acceptance criteria (causal: 32; non-causal: 29) describing the functionality of the German Corona-Warn-App. We achieve a macro1 score of 83 %, which corroborates the feasibility of our approach. © 2021 CEUR-WS. All rights reserved. open access

Sentiment analysis, or opinion mining, is a computational process to determine the polarity of a topic, opinion, emotion, or attitude. Most of the work done on sentiment analysis is for resource-rich languages, such as English and Chinese. However, only limited work has been done for Roman Urdu/Hindi, which is hence a resource-poor language. Developing a robust Sentiment analysis system for Roman Urdu/Hindi is necessitated due to two major reasons. First, Urdu/Hindi is the third largest spoken language in the world, with over 500 million speakers. Second, it is becoming increasingly used because people prefer to communicate on the web using Latin Script (26 English Alphabets), instead of typing using their language-specific keyboards. Since the work on Roman Urdu/Hindi sentiment analysis is still in its infancy stage, therefore an urgent development of new techniques and improvements in existing techniques is required. In particular, the development of an automated technique to address the problem of Roman Urdu/Hindi text normalization is necessary as that widely affects the performance of all Natural Language Processing applications, including Sentiment classification. The non-availability of an annotated dataset is another major issue towards building effective techniques for Roman Urdu/Hindi sentiment analysis. In this thesis, challenging issues hindering the development of effective Roman Urdu/Hindi sentiment classification have been addressed. First, the largest-ever dataset of 11000 Roman Urdu/Hindi reviews has been gathered from six different domains, using comprehensive annotation guidelines. Second, a machine learning-based Roman Urdu sentiment analysis is developed using different content-based features. Third, a novel feature selection technique, called Discriminative Feature Spamming Technique, has been developed for Roman Urdu/Hindi sentiment analysis. This technique identifies distinctive features based on a term utility criteria and then further increases their discriminative power by spamming them. The spelling variation problem inherent to Roman Urdu/Hindi adversely affects the performance of the machine learning algorithms. Therefore, in the next step, an open and hard problem of Roman Urdu/Hindi word normalization has been addressed by developing an automated lexical normalizer. The encoder maps differing spellings of a single Roman Urdu/Hindi word to a single common code, via a transliteration-based technique. This technique will have broad implications over different natural language processing applications. In addition, it will provide a concrete foundation to the research community to develop tools to automatically transliterate Urdu to Roman and vice-versa.

Resumen Este trabajo tuvo como propósito explorar entornos virtuales desde un sistema interactivo que permitiera interpretar los gestos manuales hacia una aproximación musical. La metodología implementada se realizó desde un enfoque cuantitativo, donde se establecieron tres fases de recolección de datos: secuencialidad de gestos, signos de Curwen y aproximación a la realidad. Este estudio fue dividido en tres etapas longitudinales conceptualización, diseño e implementación y evaluación para ampliar el espectro en la interpretación de los gestos. El método Kodály implementado en este trabajo, el cual ha sido adaptado en Colombia, utiliza los gestos manuales para asimilar con mayor rapidez conceptos musicales. La evaluación fue desarrollada con una muestra poblacional de 15 usuarios en cada fase para la recolección de datos. Los resultados se identificó una rápida adaptación en la interpretación de los gestos en relación con las notas musicales. Los tiempos y errores obtenidos permitieron medir aspectos en la ejecución de los gestos. Se adaptó un conjunto de gestos de la mano a un entorno virtual en el que los usuarios pueden tocar la melódica mediante la interpretación gestual, lo que demostró que el reconocimiento de gestos puede ayudar hacia una aproximación musical. Finalmente, los usuarios lograron un alto nivel de comprensión gracias al uso de movimientos naturales de la mano, lo cual fue un proceso significativo para los usuarios en una primera experiencia mediante el reconocimiento de gestos manuales. Abstract This article describes a virtual environment of an interactive computer system that can interpret hand gestures as pitches in a tonal scale. The methodology implemented here is based on a quantitative approach with three data collection stages: sequence of gestures, Curwen’s hand signs, and approximation to reality. This study was divided into three general stages conceptualization, design and implementation, and evaluation, to expand the scope of hand gesture recognition. The Kodály method implemented in this paper, which was adapted in Colombia, uses hand gesture to help students better understand musical concepts. The evaluation was conducted with a population sample of 15 users in each data collection stage. The results show that users adapted quickly to hand gesture recognition and its relationship with musical notes. Times and errors obtained in the tests were used to measure aspects of execution. A set of hand gestures was adapted to a virtual environment where users can play melodica by gestural interpretation, which showed that gesture recognition can help music instruction. Finally, users achieved a high level of understanding thanks to the use of natural hand movements, and it was a meaningful process for them because it was their first experience with a hand gesture recognition.

In the inter-domain routing infrastructure, a route leak is defined as a violation of the routing policy agreed between two Autonomous Systems (AS). Route leaks have resulted in large-scale outages on the Internet, taking down several services. Although route leaks seem a simple problem, the solution is complex because: (i) ASes consider -partially- routing policy private, (ii) lack of a formal and standard language to express routing policy and (iii) BGP lacks adequate cryptographic-based security. In this paper, we present an experimental analysis of a distributed ledger-based architecture that provides a solution to route leaks. Specifically, the routing policy is unambiguously expressed using a formal language, that is then stored in a blockchain. This decentralized architecture allows private policies and interfaces seamlessly with the current BGP infrastructure, requiring no changes to routers. We build a prototype to evaluate our proposed architecture using Hyperledger, we analyze its performance using a real-world BGP dataset. Our results show that our architecture scales linearly with relevant metrics. Additionally, we validate the architecture preventing an artificially introduced route leak in a realistic 10 AS topology. This work was supported by the Spanish MINECO under contract TEC2017-90034-C2-1-R (ALLIANCE) and the Catalan Institution for Research and Advanced Studies (ICREA).

The quality and maintainability of a knowledge graph are determined by the process in which it is created. There are different approaches to such processes; extraction or conversion of available data in the web (automated extraction of knowledge such as DBpedia from Wikipedia), community-created knowledge graphs, often by a group of experts, and hybrid approaches where humans maintain the knowledge graph alongside bots. We focus in this work on the hybrid approach of human edited knowledge graphs supported by automated tools. In particular, we analyse the editing of natural language data, i.e. labels. Labels are the entry point for humans to understand the information, and therefore need to be carefully maintained. We take a step toward the understanding of collaborative editing of humans and automated tools across languages in a knowledge graph. We use Wikidata as it has a large and active community of humans and bots working together covering over 300 languages. In this work, we analyse the different editor groups and how they interact with the different language data to understand the provenance of the current label data. © 2019 Copyright held by the owner/author(s). ACM ISBN 978-1-4503-6319-8/19/08.

The proliferation of smart spaces, such as smart buildings, is increasing opportunities for offenders to exploit the interplay between cyber and physical components, in order to trigger security incidents. Organizations are obliged to report security incidents to comply with recent data protection regulations. Organizations can also use incident reports to improve security of the smart spaces where they operate. Incident reporting is often documented in structured natural language. However, reports often do not capture relevant information about cyber and physical vulnerabilities present in a smart space that are exploited during an incident. Moreover, sharing information about security incidents can be difficult, or even impossible, since a report may contain sensitive information about an organization. In previous work, we provided a meta-model to represent security incidents in smart spaces. We also developed an automated approach to share incident knowledge across different organizations. In this paper we focus on incident reporting. We provide a System Editor to represent smart buildings where incidents can occur. Our editor allows us to represent cyber and physical components within a smart building and their interplay. We also propose an Incident Editor to represent the activities of an incident, including -for each activity- the target and the resources exploited, the location where the activity occurred, and the activity initiator. Building on our previous work, incidents represented using our editor can be shared across various organizations, and instantiated in different smart spaces to assess how they can re-occur. We also propose an Incident Filter component that allows viewing and prioritizing the most relevant incident instantiations, for example, involving a minimum number of activities. We assess the feasibility of our approach in assisting incident reporting using an example of a security incident that occurred in a research center. European Commission Horizon 2020 European Research Council Science Foundation Ireland

Dialog agents, like digital assistants and automated chat interfaces (e.g., chatbots), are becoming more and more popular as users adapt to conversing with their devices as they do with humans. In this paper, we present approaches and available tools for dialog management (DM), a component of dialog agents that handles dialog context and decides the next action for the agent to take. In this paper, we establish an overview of the field of DM, compare approaches and state-of-the-art tools in industry and research work on a set of dimensions, and identify directions for further research work.

By indicating semantic relations between a predicate and its associated participants in a sentence and identifying the role-bearing constituents, SRL provides an extensive dataset to understand natural languages and to enhance several NLP applications such as information retrieval, machine translation, information extraction, and question answering. The availability of large resources and the development of statistical machine learning methods have increased the studies in the field of SRL. One of the widely-used semantic resources applied for multiple languages is PropBank. In this paper, PropBanks applied for Turkish are compared by checking semantic roles in the frame files of matched verb senses. As this integrated lexical resource for Turkish is aimed to be used in a multilingual resource along with English, creation of an inclusive lexical resource for Turkish is of great importance. Publisher's Version

Exploiting the fact that natural languages are complex systems, the present exploratory article proposes a direct method based on frequency distributions that may be useful when making a decision on the status of problematic phonemes, an open problem in linguistics. The main notion is that natural languages, which can be considered from a complex outlook as information processing machines, and which somehow manage to set appropriate levels of redundancy, already "made the choice" whether a linguistic unit is a phoneme or not, and this would be reflected in a greater smoothness in a frequency versus rank graph. For the particular case we chose to study, we conclude that it is reasonable to consider the Spanish semiconsonant /w/ as a separate phoneme from its vowel counterpart /u/, on the one hand, and possibly also the semiconsonant /j/ as a separate phoneme from its vowel counterpart /i/, on the other. As language has been so central a topic in the study of complexity, this discussion grants us, in addition, an opportunity to gain insight into emerging properties in the broader complex systems debate., Comment: 24 pages, 10 figures

The structure of a sentence can be represented as a network where vertices are words and edges indicate syntactic dependencies. Interestingly, crossing syntactic dependencies have been observed to be infrequent in human languages. This leads to the question of whether the scarcity of crossings in languages arises from an independent and specific constraint on crossings. We provide statistical evidence suggesting that this is not the case, as the proportion of dependency crossings of sentences from a wide range of languages can be accurately estimated by a simple predictor based on a null hypothesis on the local probability that two dependencies cross given their lengths. The relative error of this predictor never exceeds 5% on average, whereas the error of a baseline predictor assuming a random ordering of the words of a sentence is at least 6 times greater. Our results suggest that the low frequency of crossings in natural languages is neither originated by hidden knowledge of language nor by the undesirability of crossings per se, but as a mere side effect of the principle of dependency length minimization., minor corrections; in press in Physical Review E

International audience; Retrieving information based on the users' preferences and profiles represent a challenging issue to overcome. Moreover, in the public transport field, this task becomes increasingly complex due to the heterogeneous data fetched from various sources. Though, ontologies have emerged in retrieving information field to reduce this complexity. This paper describes a visual framework aiming to search and extract the suitable itinerary that fits better with the user's information needs and based on the ontologies. The framework relies on three main components: a simple search component, an advanced search based on users' preferences and profile component and a Question-Answering search component represented in a natural language. Our framework has been implemented and evaluated.

Conversation Fillers (CFs), such as `um', `hmm', and `ah', may help to improve the human-robot interaction by smoothening the robot's responses. This paper presents the design and test of such CFs - alongside iconic pensive or acknowledging gestures - for Wizard of Oz (WoZ) controlled open-ended dialogues in child-robot interactions. A controlled experiment with 26 children showed that these CFs can improve the perceived speediness, aliveness, humanness, and likability of the robot, without decreasing perceptions of intelligence, trustworthiness, or autonomy.

The cornerstone of current Computational Linguistics research is the computation of semantic similarity between lexical items or some of their conceptualization in available semantic resources such as WordNet. However, measures for semantic similarity and/or relatedness usually work with numerical outputs, which are then used to solve tasks related to word disambiguation rather than information retrieval. In this paper, we start from the limitations of using numeric-based similarity measures, proposing a novel approach to provide explanations of similarity, even if still calculated through statistical and thus numerical analyses. This may allow a novel, fine-grained and context-based similarity reasoning over lexical entities. In this contribution, we define the concept of semantic similarity reasoning and a method of extraction from ConceptNet, a large common-sense resource. Finally, we present a number of hypotheses of how such shift of paradigm could represent a new building block of future natural language technologies.

[EN] Legal texts usually comprise many kinds of texts, such as contracts, patents and treaties. These texts usually include a huge quantity of unstructured information written in natural language. Thanks to automatic analysis and Information Retrieval (IR) techniques, it is possible to filter out information that is not relevant and, therefore, to reduce the amount of documents that users need to browse to find the information they are looking for. In this paper we adapted the JIRS passage retrieval system to work with three kinds of legal texts: treaties, patents and contracts, studying the issues related with the processing of this kind of information. In particular, we studied how a passage retrieval system might be linked up to automated analysis based on logic and algebraic programming for the detection of conflicts in contracts. In our set-up, a contract is translated into formal clauses, which are analysed by means of a model checking tool; then, the passage retrieval system is used to extract conflicting sentences from the original contract text. © 2011 Elsevier Inc. All rights reserved., We thank the MICINN (Plan I+D+i) TEXT-ENTERPRISE 2.0: (TIN2009-13391-C04-03) research project. The work of the second author has been possible thanks to a scholarship funded by Maat Gknowledge in the framework of the project with the Universidad Politécnica de Valencia Módulo de servicios semánticos de la plataforma G

En exploitant une représentation géométrique des phonèmes vocaliques, nous réalisons un modèle bidimensionnel dans lequel des voyelles sont des points et les distances entre ces points expriment des différences auditives. Ceci nous permettra de décrire le système vocalique d'une langue du point de vue d’une autre langue au moyen d'une partition d'un ensemble fini dont les propriétés combinatoires peuvent être explorées. Le concept de base que nous utilisons est celui du diagramme de Voronoï, qui a été largement utilisé dans d’autres domaines. Dans le cas présent, nous mettons en évidence quelques particularités combinatoires de partitions d’entiers qui décrivent des dissemblances entre des inventaires vocaliques de différentes langues et classons les relations possibles entre inventaires par le biais des graphes orientés appropriés, et, en particulier, parmi les diagrammes de Voronoï adéquats. Nous appliquons cette théorie à quelques langues réelles, en recherchant des améliorations pouvant faciliter la compréhension d’un inventaire vocalique par un auditeur dont la catégorisation auditive est différente. Enfin, nous décrivons des inventaires privilégiés, facilement compréhensibles dans beaucoup de langues en même temps. By exploiting a well-known geometrical representation of vowel phonemes, we devise a two-dimensional model in which vowels are points and distances between points express auditory discrepancies. This will allow us to describe the vowel system of a language, as seen by another language, by means of a set partition whose combinatorial properties can be explored. The basic concept we employ is that of Voronoi diagram, which has been, so far, extensively used for many other purposes. In the present framework we point out some combinatorial features of integers partitions which describe dissimilarities between vowel inventories of different languages. We classify the possible relations between inventories via suitable directed graphs related to point configurations and, in particular, to the pertinent Voronoi diagrams. We apply the above theory to some real languages, we also look for possible improvements that make a vowel inventory easier to understand by a listener whose auditory categorisation is different. Finally, we describe particular inventories, easily understandable in many languages at the same time.

Words in humans follow the so-called Zipf's law. More precisely, the word frequency spectrum follows a power function, whose typical exponent is ß ˜ 2, but significant variations are found. We hypothesize that the full range of variation reflects our ability to balance the goal of communication, i.e. maximizing the information transfer and the cost of communication, imposed by the limitations of the human brain. We show that the higher the importance of satisfying the goal of communication, the higher the exponent. Here, assuming that words are used according to their meaning we explain why variation in ß should be limited to a particular domain. From the one hand, we explain a non-trivial lower bound at about ß = 1.6 for communication systems neglecting the goal of the communication. From the other hand, we find a sudden divergence of ß if a certain critical balance is crossed. At the same time a sharp transition to maximum information transfer and unfortunately, maximum communication cost, is found. Consistently with the upper bound of real exponents, the maximum finite value predicted is about ß = 2.4. It is convenient, for human language not to cross the transition and remain in a domain where maximum information transfer is high but at a reasonable cost. Therefore, only a particular range of exponents should be found in human speakers. The exponent ß contains information about the balance between cost and communicative efficiency.

Although molecular biology, genetics, and related special disciplines represent a large amount of empirical data, a practical method for the evaluation and overview of current knowledge is far from being realized. The main concepts and narratives in these fields have remained nearly the same for decades and the more recent empirical data concerning the role of noncoding RNAs and persistent viruses and their defectives do not fit into this scenario. A more innovative approach such as applied biocommunication theory could translate empirical data into a coherent perspective on the functions within and between biological organisms and arguably lead to a sustainable integrative biology.

The lack of a software tool to automate the creation of a well-defined, Natural Language Software Requirements Specification (SRS) document is an obstacle to the process of efficient Requirements Engineering (RE). This paper provides an overview of a methodology and a novel software tool that attempt to formalize and automate the RE process, and it expands on the use of the SRS Documentation component of the tool that generates automatically a well-structured Natural Language SRS document. 2010 IEEE.

In this paper we present two kinds of procedural semantics for privative modification. We do this for three reasons. The first reason is to launch a tough test case to gauge the degree of substantial agreement between a constructivist and a realist interpretation of procedural semantics; the second is to extend Martin-L ̈f’s Constructive Type Theory to privative modification, which is characteristic of natural language; the third reason is to sketch a positive characterization of privation.

Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works. M. Fernández, V. López, M. Sabou, V. S. Uren, D. Vallet, E. Motta, and P. Castells, "Semantic search meets the Web", 2008 IEEE International Conference on Semantic Computing, Santa Clara (CA), 2008, pp. 253 - 260, While semantic search technologies have been proven to work well in specific domains, they still have to confront two main challenges to scale up to the Web in its entirety. In this work we address this issue with a novel semantic search system that a) provides the user with the capability to query Semantic Web information using natural language, by means of an ontology-based Question Answering (QA) system [14] and b) complements the specific answers retrieved during the QA process with a ranked list of documents from the Web [3]. Our results show that ontology-based semantic search capabilities can be used to complement and enhance keyword search technologies., This work was funded by the Spanish Ministry of Science and Education (TIN2005-068 and FPU program) and OpenKnowledge (FP6-027253) projects

International audience; Most of Information Retrieval Systems transform natural language users'queries into bags of words that are matched to documents also represented as bags of words. Through such process, the richness of the query is lost. In this paper we show that linguistic features of a query are good indicators to predict systems failure to answer it. The experiments are based on 42 systems or system variants and 50 TREC topics that consist of a descriptive part expressed in natural language.

We investigate the nature of written human language within the framework of complex network theory. In particular, we analyse the topology of Orwell's \textit{1984} focusing on the local properties of the network, such as the properties of the nearest neighbors and the clustering coefficient. We find a composite power law behavior for both the average nearest neighbor's degree and average clustering coefficient as a function of the vertex degree. This implies the existence of different functional classes of vertices. Furthermore we find that the second order vertex correlations are an essential component of the network architecture. To model our empirical results we extend a previously introduced model for language due to Dorogovtsev and Mendes. We propose an accelerated growing network model that contains three growth mechanisms: linear preferential attachment, local preferential attachment and the random growth of a pre-determined small finite subset of initial vertices. We find that with these elementary stochastic rules we are able to produce a network showing syntactic-like structures., Comment: 9 pages, 8 figures

Principled development techniques could greatly enhance the understandability of expert systems for both users and system developers. Current systems have limited explanatory capabilities and present maintenance problems because of a failure to explicitly represent the knowledge and reasoning that went into their design. This paper describes a paradigm for constructing expert systems which attempts to identify that tacit knowledge, provide means for capturing it in the knowledge bases of expert systems, and, apply it towards more perspicuous machine-generated explanations and more consistent and maintainable system organization.

This paper describes an experiment which attempted to subjectively measure the transfer of technology directly from the research arena to operational application. Areas of technology included recently developed computer software 'natural languages' based on artificial intelligence (AI) concepts, computer networking technology and automatic message handling systems. Exposure to this technology was accomplished over a 12 week period where the experimental group was provided with the tools and understanding to use the various research products. Data were collected from this experimental group prior to and after the 12 week period. Analysis of the data revealed interesting aspects concerning a user's perception of the value or utility of computer systems and computer software designed specifically to enhance the human-system interaction. Prepared for: Naval Electronic Systems Command Washington, D.C. 20360 Approved for public release; distribution is unlimited.

This paper describes a research activity aiming at extracting variability information from ambiguities and vagueness of generic requirement documents, written in Natural Language. The proposed activity continues a research stream focusing on techniques to extract variability information from requirement documents. Here, we study the introduction of a process able to distinguish structural from functional variability, both in the extracted variability model and in the derived lower-level requirements. The problem is stated with reference to an example, a solution proposal is sketched together with related research questions, and a validation path is envisaged.

In this paper, the new method of speech modeling which is called SYMPES is introduced and it is compared with the commercially available methods. It is shown that for the same compression ratio or better, SYMPES yields considerably better hearing quality over the coders such as G.726 at 16 Kbps and voice excited LPC-10E of 2.4Kbps. IEEE Publisher's Version Q3 WOS:000245413502072

In this paper, an extension of n-grams, called x-grams, is proposed. In this extension, the memory of the model (n) is not fixed a priori. Instead, large memories are accepted first, and merging criteria are then applied to reduce the complexity and to ensure reliable estimations. The results show how the perplexity obtained with x-grams is smaller than that of n-grams. Furthermore, the complexity is smaller than trigrams and can become close to bigrams.

Due to the huge availability of documents in digital form, and the deception possibility raise bound to the essence of digital documents and the way they are spread, the authorship attribution problem has constantly increased its relevance. Nowadays, authorship attribution,for both information retrieval and analysis, has gained great importance in the context of security, trust and copyright preservation. This work proposes an innovative multi-agent driven machine learning technique that has been developed for authorship attribution. By means of a preprocessing for word-grouping and time-period related analysis of the common lexicon, we determine a bias reference level for the recurrence frequency of the words within analysed texts, and then train a Radial Basis Neural Networks (RBPNN)-based classifier to identify the correct author. The main advantage of the proposed approach lies in the generality of the semantic analysis, which can be applied to different contexts and lexical domains, without requiring any modification. Moreover, the proposed system is able to incorporate an external input, meant to tune the classifier, and then self-adjust by means of continuous learning reinforcement., Published on: Proceedings of the XV Workshop "Dagli Oggetti agli Agenti" (WOA 2014), Catania, Italy, Sepember. 25-26, 2014

We present a method for generating conversations between human-like agents by proposing specific parameters for inter-agents messages with an approximation of virtual sound propagation. We are then be able to simulate appropriate human-like behaviours automatically. For instance, we demonstrate how to create proper reactions for agents that are not able to understand, but only to hear the sentences

This position paper presents the authors’ goals on advanced human computer interaction and 3D Web -- Previous work on speech, natural language processing and visual technologies has achieved the development of the BerbaTek language learning demonstrator, a 3D virtual tutor that supports Basque language students through spoken interaction -- Next steps consist on migrating all the system to multidevice web technologies -- This paper shows the architecture defined and the steps to be performed in the next months