Improving Sentiment Classification Performance through Coaching Architectures.

Intelligent systems have been developed for years to solve specific tasks automatically. An important issue emerges when the information used by these systems exhibits a dynamic nature and evolves. This fact adds a level of complexity that makes these systems prone to a noticeable worsening of their performance. Thus, their capabilities have to be upgraded to address these new requirements. Furthermore, this problem is even more challenging when the information comes from human individuals and their interactions through language. This issue happens more easily and forcefully in the specific domain of Sentiment Analysis, where feelings and opinions of humans are in constant evolution. In this context, systems are trained with an enormous corpus of textual content, or they include an extensive set of words and their related sentiment values. These solutions are usually static and generic, making their manual upgrading almost unworkable. In this paper, an automatic and interactive coaching architecture is proposed. It includes a ML framework and a dictionary-based system both trained for a specific domain. These systems converse about the outcomes obtained during their respective learning stages by simulating human interactive coaching sessions. This leads to an Active Learning process where the dictionary-based system acquires new information and improves its performance. More than 800, 000 tweets have been gathered and processed for experiments. Outstanding results were obtained when the proposed architecture was used. Also, the lexicon was updated with the prior and new words related to the corpus used which is important to reach a better sentiment analysis classification.

Detecting Emotional Evolution on Twitter during the COVID-19 Pandemic Using Text Analysis.

Early in 2020, an unexpected and hazardous situation occurred threatening and challenging all of humankind. A new coronavirus called SARS-CoV-2 was first identified in Wuhan, China, and its related disease, called COVID-19, has induced one of the most dangerous crises at a global level since World War II. The ultra-fast transmission rate of the virus and the high mortality rate led the World Health Organization (WHO) to officially declare the situation a pandemic. Governments, for their part, were forced to implement unprecedented mobility restrictions and cease a large part of their economic activities. These facts triggered multiple reactions from people who expressed their feelings mainly through social networks (like Twitter), using them as vectors of information and opinion. In this paper, a study carried out in different Spanish speaking countries (Chile, Mexico, Peru, and Spain) is presented, which addresses the manner in which the evolution of the pandemic outbreak has affected the emotions expressed by individuals on Twitter over the last 13 months (from March 2020 to March 2021). We used a total of 3 million tweets to achieve this task. We made use of a well-known framework called EmoWeb to capture the dynamic variation in the sentimental value of pandemic-related words. The results reflect to what degree the pandemic and its derived problems have influenced and affected the population of the selected countries in different ways. The outcomes also illustrate the evolution over time of opinions published on Twitter regarding several topics related to COVID-19.

Social-Aware Driver Assistance Systems for City Traffic in Shared Spaces.

Shared spaces are gaining presence in cities, where a variety of players and mobility types (pedestrians, bicycles, motorcycles, and cars) move without specifically delimited areas. This makes the traffic they comprise challenging for automated systems. The information traditionally considered (e.g., streets, and obstacle positions and speeds) is not enough to build suitable models of the environment. The required explanatory and anticipation capabilities need additional information to improve them. Social aspects (e.g., goal of the displacement, companion, or available time) should be considered, as they have a strong influence on how people move and interact with the environment. This paper presents the Social-Aware Driver Assistance System (SADAS) approach to integrate this information into traffic systems. It relies on a domain-specific modelling language for social contexts and their changes. Specifications compliant with it describe social and system information, their links, and how to process them. Traffic social properties are the formalization within the language of relevant knowledge extracted from literature to interpret information. A multi-agent system architecture manages these specifications and additional processing resources. A SADAS can be connected to other parts of traffic systems by means of subscription-notification mechanisms. The case study to illustrate the approach applies social knowledge to predict people's movements. It considers a distributed system for obstacle detection and tracking, and the intelligent management of traffic signals.

Analysis of Intelligent Transportation Systems Using Model-Driven Simulations.

Intelligent Transportation Systems (ITSs) integrate information, sensor, control, and communication technologies to provide transport related services. Their users range from everyday commuters to policy makers and urban planners. Given the complexity of these systems and their environment, their study in real settings is frequently unfeasible. Simulations help to address this problem, but present their own issues: there can be unintended mistakes in the transition from models to code; their platforms frequently bias modeling; and it is difficult to compare works that use different models and tools. In order to overcome these problems, this paper proposes a framework for a model-driven development of these simulations. It is based on a specific modeling language that supports the integrated specification of the multiple facets of an ITS: people, their vehicles, and the external environment; and a network of sensors and actuators conveniently arranged and distributed that operates over them. The framework works with a model editor to generate specifications compliant with that language, and a code generator to produce code from them using platform specifications. There are also guidelines to help researchers in the application of this infrastructure. A case study on advanced management of traffic lights with cameras illustrates its use.