The integration of Artificial Intelligence (AI) in the software development industry has experienced a notable increase, with the emergence of Large Language Models (LLM). The adoption of LLM-Based Coding Assistance Tools (LBCAT) is still limited in the Software Maintenance (SM) field and the literature focuses on the initial stages of software development, showing inconclusive results. In this work, we identify categories of SM that can be supported by LBCAT and explore the critical factors that influence their use. We employed a mixed-method approach, beginning with a qualitative analysis using content analysis to categorize software developers’ perspectives on the applicability of LBCAT in SM. In this initial stage, we identified five potential dimensions. In the second stage, we conducted a survey to evaluate which of these dimensions contributed to better performance in SM: (1) Organizational Context and Business Strategy, (2) Innovation and technological evolution, (3) Security and Testing, (4) Technical aspects of maintenance, and (5) Human Factors Management. With these dimensions, we conducted a survey where 55

Digital transformation (DT) has become a strategic priority for public administrations, particularly due to the need to deliver more efficient and citizen-centered services and respond to societal expectations, ESG (Environmental, Social, and Governance) criteria, and the United Nations Sustainable Development Goals (UN SDGs). In this context, the main objective of this study is to propose an innovative methodology to automatically evaluate the level of digital transformation (DT) in public sector organizations. The proposed approach combines traditional assessment methods with Artificial Intelligence (AI) techniques. The methodology follows a dual approach: on the one hand, surveys are conducted using specialized staff from various public entities; on the other, AI-based models (including neural networks and transformer architectures) are used to estimate the DT level of the organizations automatically. Our approach has been applied to a real-world case study involving local public administrations in the Valencian Community (Spain) and shown effective performance in assessing DT. While the proposed methodology has been validated in a specific local context, its modular structure and dual-source data foundation support its international scalability, acknowledging that administrative, regulatory, and DT maturity factors may condition its broader applicability. The experiments carried out in this work include (i) the creation of a domain-specific corpus derived from the surveys and websites of several organizations, used to train the proposed models; (ii) the use and comparison of diverse AI methods; and (iii) the validation of our approach using real data. Based on the deficiencies identified, the study concludes that the integration of technologies such as the Internet of Things (IoT), sensor networks, and AI-based analytics can significantly support resilient, agile urban environments and the transition towards more effective and sustainable Smart City models.

Generative Artificial Intelligence has experienced exponential growth largely due to the advent of Large Language Models (LLMs). This expansion is fueled by the impressive performance of deep learning methods used in Natural Language Processing (NLP) and its subfield, Natural Language Generation (NLG), which is the focus of this paper. Popular LLMs, such as GPT-4, Bard, and tools such as ChatGPT have set benchmarks for addressing various NLG tasks. This scenario raises critical questions regarding the future of NLG and its adaptation to emerging challenges in the LLM era. To explore these issues, the present paper reviews a representative sample of recent NLG surveys, thereby providing the scientific community with a research roadmap to identify NLG aspects that remain inadequately addressed and to suggest areas warranting further in-depth exploration in NLG.

There was an error in the original publication [...]

Generative Artificial Intelligence has grown exponentially as a result of Large Language Models (LLMs). This has been possible because of the impressive performance of deep learning methods created within the field of Natural Language Processing (NLP) and its subfield Natural Language Generation (NLG), which is the focus of this paper. Within the growing LLM family are the popular GPT-4, Bard and more specifically, tools such as ChatGPT have become a benchmark for other LLMs when solving most of the tasks involved in NLG research. This scenario poses new questions about the next steps for NLG and how the field can adapt and evolve to deal with new challenges in the era of LLMs. To address this, the present paper conducts a review of a representative sample of surveys recently published in NLG. By doing so, we aim to provide the scientific community with a research roadmap to identify which NLG aspects are still not suitably addressed by LLMs, as well as suggest future lines of research that should be addressed going forward.

In this article, we present CuentosIE (TalesEI: chatbot of tales with a message to develop Emotional Intelligence), an educational chatbot on emotions that also provides teachers and psychologists with a tool to monitor their students/patients through indicators and data compiled by CuentosIE. The use of “tales with a message” is justified by their simplicity and easy understanding, thanks to their moral or associated metaphors. The main contributions of CuentosIE are the selection, collection, and classification of a set of highly specialized tales, as well as the provision of tools (searching, reading comprehension, chatting, recommending, and classifying) that are useful for both educating users about emotions and monitoring their emotional development. The preliminary evaluation of the tool has obtained encouraging results, which provides an affirmative answer to the question posed in the title of the article.

Autism Spectrum Disorder (ASD) is a developmental disability primarily characterized by challenges in social interaction and communication. Due to the unknown etiology of ASD, numerous computational psychiatry research studies have been carried out to identify pertinent features and uncover hidden correlations to detect this type of disability at an early stage. The aim of this ongoing project is to present the initial tests carried out on autistic children by analysing their conversations or writings to assess their social skills in order to find indicators for the most personalised intervention possible. This model would consist of the most advanced machine learning algorithms and Natural Language Processing techniques (e.g. Transformers or ChatGPT). The paper concludes by presenting a case study that utilized autism data to verify the efficacy of our proposed model, demonstrating remarkably promising findings.

The scientific and technological evolution that has taken place in the field of human activity recognition, while enormous, is possibly only the tip of the iceberg of the possibilities that we are currently facing and will continue to experience in the foreseeable future. Much of this innovation has also been driven by the rise of IoT technology at the personal device level which has enabled convenient application in the fields of assisting living, ambient intelligence, and e-health. This paper presents a proposal that is part of an ongoing project for the deployment of sensors in supervised housing, with the challenges and opportunities that this implies. At the moment, different technical possibilities are being assessed for the implementation of the schemes, both at macro and micro level, which are described in the proposed architecture detailed in the solutions presented in this paper. The special characteristics of the individuals, usually with different types of disabilities, who live in these homes make this exciting project, marked with a very high social component, a very big challenge for their inclusion in society.

Ministerio de Ciencia, Innovacion y Universidades del Gobierno de Espana, programa PROMETEO de la Generalitat Valenciana, COST Action Distant Reading for European Literary History (CA16204 - Distant-Reading)

Different fields such as linguistics, teaching, and computing have demonstrated special interest in the study of sign languages (SL). However, the processes of teaching and learning these languages turn complex since it is unusual to find people teaching these languages that are fluent in both SL and the native language of the students. The teachings from deaf individuals become unique. Nonetheless, it is important for the student to lean on supportive mechanisms while being in the process of learning an SL. Bidirectional communication between deaf and hearing people through SL is a hot topic to achieve a higher level of inclusion. However, all the processes that convey teaching and learning SL turn difficult and complex since it is unusual to find SL teachers that are fluent also in the native language of the students, making it harder to provide computer teaching tools for different SL. Moreover, the main aspects that a second language learner of an SL finds difficult are phonology, non-manual components, and the use of space (the latter two are specific to SL, not to spoken languages). This proposal appears to be the first of the kind to favor the Costa Rican Sign Language (LESCO, for its Spanish acronym), as well as any other SL. Our research focus stands on reinforcing the learning process of final-user hearing people through a modular architectural design of a learning environment, relying on the concept of phonological proximity within a graphical tool with a high degree of usability. The aim of incorporating phonological proximity is to assist individuals in learning signs with similar handshapes. This architecture separates the logic and processing aspects from those associated with the access and generation of data, which makes it portable to other SL in the future. The methodology used consisted of defining 26 phonological parameters (13 for each hand), thus characterizing each sign appropriately. Then, a similarity formula was applied to compare each pair of signs. With these pre-calculations, the tool displays each sign and its top ten most similar signs. A SUS usability test and an open qualitative question were applied, as well as a numerical evaluation to a group of learners, to validate the proposal. In order to reach our research aims, we have analyzed previous work on proposals for teaching tools meant for the student to practice SL, as well as previous work on the importance of phonological proximity in this teaching process. This previous work justifies the necessity of our proposal, whose benefits have been proved through the experimentation conducted by different users on the usability and usefulness of the tool. To meet these needs, homonymous words (signs with the same starting handshape) and paronyms (signs with highly similar handshape), have been included to explore their impact on learning. It allows the possibility to apply the same perspective of our existing line of research to other SL in the future.

The study of sign languages (SL) has generated great interest in the fields of linguistics, computing and teaching. Teaching and learning SL are complex tasks, since they are rarely taught by people who are fluent both in SL and the native language of the students. These lessons taught by deaf people are irreplaceable, but it is appropriate for the student to have support mechanisms in the process. This paper presents a SL learning reinforcement tool that uses phonological proximity to improve the obtained results. The methodology used consisted of mapping the phonological parameters of each sign in the lexicon to numerical values and then constructing a matrix where each sign is compared with the others by means of a classic measure of similarity. As far as the authors know, this is the first time that this type of proposal is made for Costa Rican SL (LESCO, for its Spanish acronym), incorporating phonological proximity in a reinforcement tool. The operation of a graphical software tool is explained, showing grouped concepts and reproducing the signs, interfacing with the computerized avatar of the already operational International Platform for Sign Language Edition (PIELS, for its Spanish acronym), expanding its current functionality by learning signs with similar handshapes. For these purposes, homonyms, paronyms, and polysemy are explored, to the extent that these concepts apply to SL. The incorporation of phonological proximity to the tool is explored, in order to reinforce LESCO learning, offering the possibility of using the same approach in any other SL.

The study of phonological proximity makes it possible to establish a basis for future decision-making in the treatment of sign languages. Knowing how close a set of signs are allows the interested party to decide more easily its study by clustering, as well as the teaching of the language to third parties based on similarities. In addition, it lays the foundation for strengthening disambiguation modules in automatic recognition systems. To the best of our knowledge, this is the first study of its kind for Costa Rican Sign Language (LESCO, for its Spanish acronym), and forms the basis for one of the modules of the already operational system of sign and speech editing called the International Platform for Sign Language Edition (PIELS). A database of 2665 signs, grouped into eight contexts, is used, and a comparison of similarity measures is made, using standard statistical formulas to measure their degree of correlation. This corpus will be especially useful in machine learning approaches. In this work, we have proposed an analysis of different similarity measures between signs in order to find out the phonological proximity between them. After analyzing the results obtained, we can conclude that LESCO is a sign language with high levels of phonological proximity, particularly in the orientation and location components, but they are noticeably lower in the form component. We have also concluded as an outstanding contribution of our research that automatic recognition systems can take as a basis for their first prototypes the contexts or sign domains that map to clusters with lower levels of similarity. As mentioned, the results obtained have multiple applications such as in the teaching area or the Natural Language Processing area for automatic recognition tasks.

New information and communication technologies have contributed to the development of the smart city concept. On a physical level, this paradigm is characterized by deploying a substantial number of different devices that can sense their surroundings and generate a large amount of data. The most typical case is image and video acquisition sensors. Recently, these types of sensors are found in abundance in urban spaces and are responsible for producing a large volume of multimedia data. The advanced computer vision methods for this type of multimedia information means that many aspects can be dynamically monitored, which can help implement value-added applications in the city. However, obtaining more elaborate semantic information from these data poses significant challenges related to a large amount of data generated and the processing capabilities required. This paper aims to address these issues by using a combination of cloud computing technologies and mobile computing techniques to design a three-layer distributed architecture for intensive urban computing. The approach consists of distributing the processing tasks among a city's multimedia acquisition devices, a middle computing layer, known as a cloudlet, and a cloud-computing infrastructure. As a result, each part of the architecture can now focus on a small number of tasks for which they are specially designed, and data transmission communication needs are significantly reduced. To this end, the cloud server can hold and centralize the multimedia analysis of the processed results from the lower layers. Finally, a case study on smart lighting is described to illustrate the benefits of using the proposed model in smart city environments.

[This corrects the article DOI: 10.1371/journal.pone.0215288.].

Nowadays, firms have realized the importance of Big Data, highlighting the need for understanding the current state of marketing practice with respect to Big Data analytics. Among the different sources of Big Data, User-Generated Content (UGC) is one of the most important ones. From blogs to social media and online reviews, consumers generate huge amounts of brand related information that have a decisive potential business value in targeted advertising, customer engagement or brand communication, among others. In the same line, previous empirical findings show that UGC has significant effects on brand images, purchase intentions, and sales. It plays an important role for customers' potential buying decisions. Thus, mining and analysing UGC data such as comments and sentiments might be useful for firms. Particularly, brand management can be one area of interest, as online reviews might have an influence on brand image and brand positioning. Within this context, as well as the quantitative star score usual in this UGC, in which the buyers rate the product, a recent stream of research employs Sentiment Analysis (SA) tools with the aim of examining the textual content of the review and categorizing buyers' opinions. While certain SA split the comments into two classes (negative or positive), other incorporate more sentiment classes. However, the review can have phrases with different polarities because the user can have different experiences and sentiments about each feature of the product. Finding the polarity of each feature can be interesting for the decision makers of a product. In this paper, we consider that although these two scores (star and sentiment) are related, the sentiment score highlights extra information not detailed in the star score, which is crucial to be extracted in order to have better criteria of comparison between products. Moreover, we mine the positive and negative features of the products analysing the sentiment.

[This corrects the article DOI: 10.1371/journal.pone.0215288.].

User-generated content about brands is an important source of big data that can be transformed into valuable information. A huge number of items are reviewed and rated by consumers on a daily basis, and managers have a keen interest in real-time monitoring of this information to improve decision-making. The main challenge is to mine reliable textual consumer opinions, and automatically use them to rate the best products or brands. We propose a framework to automatically analyse these reviews, transforming negative and positive user opinions in a quantitative score. Sentiment analysis was employed to analyse online reviews on Amazon. The Fake Review Detection Framework—FRDF— detects and removes fake reviews using Natural Language Processing technology. The FRDF was tested on reviews of products from high-tech industries. Brands were rated according to consumer sentiment. The findings demonstrate that brand managers and consumers would find this tool useful, in combination with the 5-Star score, for more comprehensive decision-making. For instance, the FRDF ranks the best products by price alongside their respective sentiment value and the 5-Star score.

Automatic word sense disambiguation (WSD) from text is a task of great importance in various applications of natural language processing, for example, in machine translation, question answering, automatic summarization or sentiment analysis. There are different approaches to finding the meaning of a word within a context, whether using supervised, unsupervised, semi-supervised or knowledge-based methods. Several studies have been conducted to automatically translate from text to sign language, reproducing the result of the translation with a signing avatar, in a way that deaf users have access to informative contents that otherwise are highly inaccessible, because sign language is their mother tongue. The many proposals that have been made look forward to minimize these informative and communicative barriers. Sign languages, however, do not have as many words as the spoken languages, so an automatic translation must be as accurate and free of ambiguities as possible. In this paper, we propose to evaluate the use of public access big data resources, as well as appropriate techniques to access this type of resources for WSD tasks, illustrating their effects in a translation system from text in Spanish to Costa Rican Sign Language (LESCO). The architecture of the actual system incorporates the use of a folksonomy, from which the disambiguation process will benefit. When an exact word is not found for a given detected sense in the source text, the ontology will be fed back with a new relationship of hyperonymy, to alert the curator on the need to propose a new sign in that category, thus promoting an enrichment in a key component of the architecture. As a result of the evaluation, the most appropriate big data public resources and techniques for WSD for sign language will be elucidated.