Los seres humanos han desarrollado especialmente su capacidad perceptiva para procesar caras y extraer información de las características faciales. Usando nuestra capacidad conductual para percibir rostros, hacemos atribuciones tales como personalidad, inteligencia o confiabilidad basadas en la apariencia facial que a menudo tienen un fuerte impacto en el comportamiento social en diferentes dominios. Por lo tanto, las caras desempeñan un papel fundamental en nuestras relaciones con otras personas y en nuestras decisiones cotidianas. Con la popularización de Internet, las personas participan en muchos tipos de interacciones virtuales, desde experiencias sociales, como juegos, citas o comunidades, hasta actividades profesionales, como e-commerce, e-learning, e-therapy o e-health. Estas interacciones virtuales manifiestan la necesidad de caras que representen a las personas reales que interactúan en el mundo digital: así surgió el concepto de avatar. Los avatares se utilizan para representar a los usuarios en diferentes escenarios y ámbitos, desde la vida personal hasta situaciones profesionales. En todos estos casos, la aparición del avatar puede tener un efecto no solo en la opinión y percepción de otra persona, sino en la autopercepción, que influye en la actitud y el comportamiento del sujeto. De hecho, los avatares a menudo se emplean para obtener impresiones o emociones a través de expresiones no verbales, y pueden mejorar las interacciones en línea o incluso son útiles para fines educativos o terapéuticos. Por lo tanto, la posibilidad de generar avatares de aspecto realista que provoquen un determinado conjunto de impresiones sociales supone una herramienta muy interesante y novedosa, útil en un amplio abanico de campos. Esta tesis propone un método novedoso para generar caras de aspecto realistas con un perfil social asociado que comprende 15 impresiones diferentes. Para este propósito, se completaron varios objetivos parciales. En primer lugar, las características faciales se extrajeron de una base de datos de caras reales y se agruparon por aspecto de una manera automática y objetiva empleando técnicas de reducción de dimensionalidad y agrupamiento. Esto produjo una taxonomía que permite codificar de manera sistemática y objetiva las caras de acuerdo con los grupos obtenidos previamente. Además, el uso del método propuesto no se limita a las características faciales, y se podría extender su uso para agrupar automáticamente cualquier otro tipo de imágenes por apariencia. En segundo lugar, se encontraron las relaciones existentes entre las diferentes características faciales y las impresiones sociales. Esto ayuda a saber en qué medida una determinada característica facial influye en la percepción de una determinada impresión social, lo que permite centrarse en la característica o características más importantes al diseñar rostros con una percepción social deseada. En tercer lugar, se implementó un método de edición de imágenes para generar una cara totalmente nueva y realista a partir de una definición de rostro utilizando la taxonomía de rasgos faciales antes mencionada. Finalmente, se desarrolló un sistema para generar caras realistas con un perfil de rasgo social asociado, lo cual cumple el objetivo principal de la presente tesis. La principal novedad de este trabajo reside en la capacidad de trabajar con varias dimensiones de rasgos a la vez en caras realistas. Por lo tanto, en contraste con los trabajos anteriores que usan imágenes con ruido, o caras de dibujos animados o sintéticas, el sistema desarrollado en esta tesis permite generar caras de aspecto realista eligiendo los niveles deseados de quince impresiones: Miedo, Enfado, Atractivo, Cara de niño, Disgustado, Dominante, Femenino, Feliz, Masculino, Prototípico, Triste, Sorprendido, Amenazante, Confiable e Inusual. Los prometedores resultados obtenidos permitirán investigar más a fondo cómo modelar l / Humans have specially developed their perceptual capacity to process faces and to extract information from facial features. Using our behavioral capacity to perceive faces, we make attributions such as personality, intelligence or trustworthiness based on facial appearance that often have a strong impact on social behavior in different domains. Therefore, faces play a central role in our relationships with other people and in our everyday decisions. With the popularization of the Internet, people participate in many kinds of virtual interactions, from social experiences, such as games, dating or communities, to professional activities, such as e-commerce, e-learning, e-therapy or e-health. These virtual interactions manifest the need for faces that represent the actual people interacting in the digital world: thus the concept of avatar emerged. Avatars are used to represent users in different scenarios and scopes, from personal life to professional situations. In all these cases, the appearance of the avatar may have an effect not only on other person's opinion and perception but on self-perception, influencing the subject's own attitude and behavior. In fact, avatars are often employed to elicit impressions or emotions through non-verbal expressions, and are able to improve online interactions or even useful for education purposes or therapy. Then, being able to generate realistic looking avatars which elicit a certain set of desired social impressions poses a very interesting and novel tool, useful in a wide range of fields. This thesis proposes a novel method for generating realistic looking faces with an associated social profile comprising 15 different impressions. For this purpose, several partial objectives were accomplished. First, facial features were extracted from a database of real faces and grouped by appearance in an automatic and objective manner employing dimensionality reduction and clustering techniques. This yielded a taxonomy which allows to systematically and objectively codify faces according to the previously obtained clusters. Furthermore, the use of the proposed method is not restricted to facial features, and it should be possible to extend its use to automatically group any other kind of images by appearance. Second, the existing relationships among the different facial features and the social impressions were found. This helps to know how much a certain facial feature influences the perception of a given social impression, allowing to focus on the most important feature or features when designing faces with a sought social perception. Third, an image editing method was implemented to generate a completely new, realistic face from just a face definition using the aforementioned facial feature taxonomy. Finally, a system to generate realistic faces with an associated social trait profile was developed, which fulfills the main objective of the present thesis. The main novelty of this work resides in the ability to work with several trait dimensions at a time on realistic faces. Thus, in contrast with the previous works that use noisy images, or cartoon-like or synthetic faces, the system developed in this thesis allows to generate realistic looking faces choosing the desired levels of fifteen impressions, namely Afraid, Angry, Attractive, Babyface, Disgusted, Dominant, Feminine, Happy, Masculine, Prototypical, Sad, Surprised, Threatening, Trustworthy and Unusual. The promising results obtained in this thesis will allow to further investigate how to model social perception in faces using a completely new approach. / Els sers humans han desenvolupat especialment la seua capacitat perceptiva per a processar cares i extraure informació de les característiques facials. Usant la nostra capacitat conductual per a percebre rostres, fem atribucions com ara personalitat, intel·ligència o confiabilitat basades en l'aparença facial que sovint tenen un fort impacte en el comportament social en diferents dominis. Per tant, les cares exercixen un paper fonamental en les nostres relacions amb altres persones i en les nostres decisions quotidianes. Amb la popularització d'Internet, les persones participen en molts tipus d'inter- accions virtuals, des d'experiències socials, com a jocs, cites o comunitats, fins a activitats professionals, com e-commerce, e-learning, e-therapy o e-health. Estes interaccions virtuals manifesten la necessitat de cares que representen a les persones reals que interactuen en el món digital: així va sorgir el concepte d'avatar. Els avatars s'utilitzen per a representar als usuaris en diferents escenaris i àmbits, des de la vida personal fins a situacions professionals. En tots estos casos, l'aparició de l'avatar pot tindre un efecte no sols en l'opinió i percepció d'una altra persona, sinó en l'autopercepció, que influïx en l'actitud i el comportament del subjecte. De fet, els avatars sovint s'empren per a obtindre impressions o emocions a través d'expressions no verbals, i poden millorar les interaccions en línia o inclús són útils per a fins educatius o terapèutics. Per tant, la possibilitat de generar avatars d'aspecte realista que provoquen un determinat conjunt d'impressions socials planteja una ferramenta molt interessant i nova, útil en un ampla varietat de camps. Esta tesi proposa un mètode nou per a generar cares d'aspecte realistes amb un perfil social associat que comprén 15 impressions diferents. Per a este propòsit, es van completar diversos objectius parcials. En primer lloc, les característiques facials es van extraure d'una base de dades de cares reals i es van agrupar per aspecte d'una manera automàtica i objectiva emprant tècniques de reducció de dimensionalidad i agrupament. Açò va produir una taxonomia que permet codificar de manera sistemàtica i objectiva les cares d'acord amb els grups obtinguts prèviament. A més, l'ús del mètode proposat no es limita a les característiques facials, i es podria estendre el seu ús per a agrupar automàticament qualsevol altre tipus d'imatges per aparença. En segon lloc, es van trobar les relacions existents entre les diferents característiques facials i les impressions socials. Açò ajuda a saber en quina mesura una determinada característica facial influïx en la percepció d'una determinada impressió social, la qual cosa permet centrar-se en la característica o característiques més importants al dissenyar rostres amb una percepció social desitjada. En tercer lloc, es va implementar un mètode d'edició d'imatges per a generar una cara totalment nova i realista a partir d'una definició de rostre utilitzant la taxonomia de trets facials abans mencionada. Finalment, es va desenrotllar un sistema per a generar cares realistes amb un perfil de tret social associat, la qual cosa complix l'objectiu principal de la present tesi. La principal novetat d'este treball residix en la capacitat de treballar amb diverses dimensions de trets al mateix temps en cares realistes. Per tant, en contrast amb els treballs anteriors que usen imatges amb soroll, o cares de dibuixos animats o sintètiques, el sistema desenrotllat en esta tesi permet generar cares d'aspecte realista triant els nivells desitjats de quinze impressions: Por, Enuig, Atractiu, Cara de xiquet, Disgustat, Dominant, Femení, Feliç, Masculí, Prototípic, Trist, Sorprés, Amenaçador, Confiable i Inusual. Els prometedors resultats obtinguts en esta tesi permetran investigar més a fons com modelar la percepció social en les cares utilitzant un enfocament complet / Fuentes Hurtado, FJ. (2018). A system for modeling social traits in realistic faces with artificial intelligence [Tesis doctoral no publicada]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/101943 / TESIS

Given the volatility of the stock market and the multitude of financial variables at play, forecasting the value of stocks can be a challenging task. Nonetheless, such prediction task presents a fascinating problem to solve using machine learning. The stock market can be affected by news events, social media posts, political changes, investor emotions, and the general economy among other factors. Predicting the stock value of a company by simply using financial stock data of its price may be insufficient to give an accurate prediction. Investors often openly express their attitudes towards various stocks on social medial platforms. Hence, combining sentiment analysis from social media and the financial stock value of a company may yield more accurate predictions. This thesis proposes a method to predict the stock market using sentiment analysis and financial stock data. To estimate the sentiment in social media posts, we use an ensemble-based model that leverages Multi-Layer Perceptron (MLP), Long Short-Term Memory (LSTM), and Convolutional Neural Network (CNN) models. We use an LSTM model for the financial stock prediction. The models are trained on the AAPL, CSCO, IBM, and MSFT stocks, utilizing a combination of the financial stock data and sentiment extracted from social media posts on Twitter between the years 2015-2019. Our experimental results show that the combination of the financial and sentiment information can improve the stock market prediction performance. The proposed solution has achieved a prediction performance of 74.3%.

Intresset och deltagandet på aktiemarknaden har ökat betydligt bland svenskar. En erkänd informationskälla om aktier är inlägg på sociala medier och speciellt på Twitter. Med hjälp av sentimentanalys av dessa inlägg, så kallade tweets, kan en allmän åsikt extraheras och användas för att förutsäga framtida resultat för ett företags aktiekurser. Syftet med denna studie är att ta fram en artefakt som kan extrahera sentiment från tweets om svenska mindre företag. Företagen valdes utifrån att de var relativt småskaliga jämfört med de företag som analyserats i liknande studier genomförda inom forskningsområdet. För denna studie har data samlats in från Twitter, analyserats och bearbetats. Olika metoder har testats för att extrahera sentiment ur tweets. Resultatet från sentimentanalys med framtagen artefakt är möjlig att använda i maskininlärningsmodeller som förutsäger aktieprisers rörelse. Resultatet från experimentet kan sammanfattas med att extrahering av sentiment från tweets är svår men möjlig. Vid analys av resultatet så framgår det att det maskininlärningsbaserade tillvägagångssättet ger en ökad prestanda jämfört med det lexikonbaserade på tweets likt de som använts i denna studie. / Interest and partaking on the stock market has increased significantly among Swedes. A recognized source of information about stocks is posts on social media and Twitter in particular. With the help of sentiment analysis on these social media posts called tweets, a public opinion can be extracted and perhaps predict the future performance of a company’s stock prices. This report is written in Swedish and the aim of the study is to produce an artefact that can extract sentiment out of tweets about minor Swedish companies. The companies were chosen on the basis that they were relatively small-scale in comparison to other studies conducted in related research. For this study data has been collected from Twitter, analyzed and processed. Different methodologies have been tested to extract sentiments out of tweets. Results of sentiment analysis with produced artefact is possible to use in machine learning models predicting stock movement. Results from conducted experiments conclude that extracting sentiment from tweets is difficult but possible. Through analysis of the results, a machine learning approach shows better performance than a lexicon based with tweets like the ones used in this study.

This thesis documents the grammatical processes and semantic impact of innovative ways to pejoratively reference individuals through adjectival nominalization. Research on nominalized adjectives suggests that when meanings shift from having one property (1) to becoming a kind with associated properties (2), the noun form often encodes stereotypical attributes: [1] "Her hair is blonde." (hair color); [2] "He married a blonde." (female, sexy, dumb). Likewise, the linguistic phenomenon of genericity refers to classes or kinds and different grammatical structures reflect properties in different ways. In 1 and 2 above, the shift from adjectival blonde to indefinite NP a blonde moves the focus from the definitional characteristic to the prototypical. Similarly, adjectival gay [3] is definitional, but the marked, nominal form [4] adds socially-based conceptions of the "average" gay (example from Twitter): [3] jesus christ i make a joke and now im a gay man? (sexuality) [constructed]; [4] jesus christ i make a joke and now im a gay? … (flamboyant, abnormal). To investigate innovative reference via nominalization, two corpus studies based in human judgment were conducted. In the first study, a subset of the corpus (N=121) was annotated for pejoration by five additional linguists following the same guidelines as the original annotator. In the second study, 800 instances were annotated by non-experts using crowd-sourcing. In both studies we find a correspondence between nominal status and pejorative meaning.

The goal of this thesis is to design and implement a system that analyses data from the web mentioning Czech grocery chain stores. Implemented system is able to download such data automatically, perform sentiment analysis of the data, extract locations and chain stores' names from the data and index the data. The system also includes a user interface showing results of the analyses. The first part of the thesis surveys the state of the art in collecting data from web, sentiment analysis and indexing documents. A description of the discussed system's design and its implementation follows. The last part of the thesis evaluates implemented system

Intergroup relations and the factors affecting them constitute a subject of recurring interest within the academic community. Social identity theory suggests that group membership and the value we assign to it drives the expression of in-group favoritism and outgroup prejudice, among other intergroup phenomena. The present study examines how (ir)religious identities are related to topic-sentiment polarization in the form of positive in-group and negative outgroup bias during interreligious debates in YouTube commentaries. Drawing from the propositions of social identity theory, six hypotheses were tested. The data for the study, a product of a natural experiment, are comments posted on YouTube commentary sections featuring videos of interreligious debates between (a) Christian and atheist or (b) Christian and Muslim speakers. Using topic-sentiment analysis, a multistage method of topic modeling with latent semantic analysis (LSA) and sentiment analysis, 52,607 comments, for the Christian - atheist debates, and 24,179 comments, for the Christian - Muslim debates, were analyzed. The results offer support (or partial support) to the hypotheses demonstrating identity-specific instances of topic-sentiment polarization to the predicted direction. The study offers valuable insights for the relevance of social identity theory in real-world interreligious interactions, while the successful application of topic-sentiment analysis lends support for the more systematic utilization of this method in the context of social identity theory.

The theme of the work is sentiment analysis, especially in terms of informatics (marginally from a linguistic point of view). The linguistic part discusses the term sentiment and language methods for its analysis, e.g. lemmatization, POS tagging, using the list of stopwords etc. More attention is paid to the structure of the sentiment analyzer which is based on some of the machine learning methods (support vector machines, Naive Bayes and maximum entropy classification). On the basis of the theoretical background, a functional analyzer is projected and implemented. The experiments are focused mainly on comparing the classification methods and on the benefits of using the individual preprocessing methods. The success rate of the constructed classifier reaches up to 84 % in the cross-validation.

This master's thesis is about the success of sharing economy in the accommodation segment in Prague. The thesis is based on theories conceptualizing sharing economy as a result of social and value change, not only as technological one. Using online review data, the user experience of shared accommodation via Airbnb and traditional via Booking are compared. Analysis is conducted with focus on users' satisfied needs and fulfilled values. For processing the data, text mining techniques (topic modelling and sentiment analysis) were employed. The major result is that in Prague the models of sharing economy accommodation meets the growing need in society to fulfil post-material values in the market much better than the models of traditional accommodation (hotels, hostels, boarding houses). In their experiences, Airbnb users reflect social and emotional values more often, even though most sharing economy accommodations in Prague do not involve any physical sharing with the host. The thesis thus brings a unique perspective on the Airbnb phenomenon in the Czech context and contributes to the discussion of why the market share of the sharing economy in the accommodation segment in Prague has been growing, while traditional models stagnated.

No description available.

The Swedish national bank has had shifting policies when it comes to publicity and confidentiality concerning publishing of texts within the bank. For some time, texts written by commissioners within the bank were decided to be published anonymously. Later they revoked the confidentiality policy, publishing all documents publicly again. This led to emerged interests in possible shifting attitudes toward topics discussed by the commissioners when writing anonymously versus publicly. On a request, based on the interests, there are ongoing analyses being conducted with the help of language technology where topics are extracted from the anonymous and public documents respectively. The aim is to find topics related to individual commissioners with the purpose of, as accurately as possible, identifying which of the anonymous documents is written by who. To discover unique relations between the commissioners and the generated topics, this thesis proposes hybrid topic based sentiment analysis as an author identification method to be able to use sentiments of topics as identifying features of commissioners. The results showed promise in the proposed approach. Though, further research is substantial, conducting comparisons with other acknowledged author identification methods, to confirm some level of efficacy, especially on documents containing close similarities among topics.