Search

Health text ranging from patient-generated online forum posts to clinician-authored unstructured notes contain valuable information that can potentially improve healthcare service quality, patient experiences, and patient and population health outcomes. Health text data are also highly heterogeneous, produced in different contexts and serve different purposes, which require careful study design and methodological innovations to ensure study validity. However, the current practices of computational analysis on health text are often inconsistent and lack considerations of the contexts in which health text is produced.My dissertation includes three major studies that analyzed different types of health text including public-generated social media data and clinical notes of patients with rare diseases. In the first study, I conducted a systematic literature review that revealed multiple issues in the current practices of how computational sentiment analysis is applied on health-related social media data. I also comprehensively evaluated the commonly used sentiment analysis tools on several social media datasets and found that they failed to accurately label the sentiments conveyed in health-related social media data. In the second study, I developed and applied computer-assisted qualitative analysis pipelines to analyze health-related social media data including tweets and online physician reviews. The results identified public attitudes and concerns toward mask wearing during the COVID-19 pandemic and patient concerns around healthcare service quality. These insights contribute to better public health communication strategies and ways of enhancing patients’ experiences when interacting with healthcare systems. In the third study, I switched gears to develop a pipeline that extracts various clinical entities including diagnosis, environmental exposures, substance use, performance status, and staging from unstructured notes of patients with lymphoid malignancies. The pipe

[Abstract] People tend to consider social platforms as convenient media for expressing their concerns and emotional struggles. With their widespread use, researchers could access and analyze user-generated content related to mental states. Computational models that exploit that data show promising results in detecting at-risk users based on engineered features or deep learning models. However, recent works revealed that these approaches have a limited capacity for generalization and interpretation when considering clinical settings. Grounding the models' decisions on clinical and recognized symptoms can help to overcome these limitations. In this paper, we introduce BDI-Sen, a symptom-annotated sentence dataset for depressive disorder. BDI-Sen covers all the symptoms present in the Beck Depression Inventory-II (BDI-II), a reliable questionnaire used for detecting and measuring depression. The annotations in the collection reflect whether a statement about the specific symptom is informative (i.e., exposes traces about the individual's state regarding that symptom). We thoroughly analyze this resource and explore linguistic style, emotional attribution, and other psycholinguistic markers. Additionally, we conduct a series of experiments investigating the utility of BDI-Sen for various tasks, including the detection and severity classification of symptoms. We also examine their generalization when considering symptoms from other mental diseases. BDI-Sen may aid the development of future models that consider trustworthy and valuable depression markers.

[Context:] Society increasingly values sustainable corporate behaviour, impacting corporate reputation and customer trust. Hence, companies regularly publish sustainability reports to shed light on their impact on environmental, social, and governance (ESG) factors. [Problem:] Sustainability reports are written by companies and therefore considered a company-controlled source. Contrarily, studies reveal that non-corporate channels (e.g., media coverage) represent the main driver for ESG transparency. However, analysing media coverage regarding ESG factors is challenging since (1) the amount of published news articles grows daily, (2) media coverage data does not necessarily deal with an ESG-relevant topic, meaning that it must be carefully filtered, and (3) the majority of media coverage data is unstructured. [Research Goal:] We aim to automatically extract ESG-relevant information from textual media reactions to calculate an ESG score for a given company. Our goal is to reduce the cost of ESG data collection and make ESG information available to the general public. [Contribution:] Our contributions are three-fold: First, we publish a corpus of 432,411 news headlines annotated as being environmental-, governance-, social-related, or ESG-irrelevant. Second, we present our tool-supported approach called ESG-Miner, capable of automatically analysing and evaluating corporate ESG performance headlines. Third, we demonstrate the feasibility of our approach in an experiment and apply the ESG-Miner on 3000 manually labelled headlines. Our approach correctly processes 96.7% of the headlines and shows great performance in detecting environmental-related headlines and their correct sentiment. © 2023 IEEE.

The outbreak of the novel coronavirus, COVID-19, has become one of the most severe pandemics in human history. In this paper, we propose to leverage social media users as social sensors to simultaneously predict the pandemic trends and suggest potential risk factors for public health experts to understand spread situations and recommend proper interventions. More precisely, we develop novel deep learning models to recognize important entities and their relations over time, thereby establishing dynamic heterogeneous graphs to describe the observations of social media users. A dynamic graph neural network model can then forecast the trends (e.g. newly diagnosed cases and death rates) and identify high-risk events from social media. Based on the proposed computational method, we also develop a web-based system for domain experts without any computer science background to easily interact with. We conduct extensive experiments on large-scale datasets of COVID-19 related tweets provided by Twitter, which show that our method can precisely predict the new cases and death rates. We also demonstrate the robustness of our web-based pandemic surveillance system and its ability to retrieve essential knowledge and derive accurate predictions across a variety of circumstances. Our system is also available at http://scaiweb.cs.ucla.edu/covidsurveiller/. This article is part of the theme issue 'Data science approachs to infectious disease surveillance'.

[Resumen] Actualmente, las capacidades de interacción de los robots sociales son limitadas y, después de un tiempo de convivencia, los diálogos que pueden mantener son percibidos como predecibles, repetitivos y poco naturales. Esto puede llevar a una pérdida de interés de la persona en el robot. Si se quiere apostar por una convivencia exitosa y larga, es necesario dotar los robots de discursos más variados y que se adapten fácilmente a los cambios de necesidades que pueda haber en los usuarios. En esta contribución se propone una metodología que combina técnicas de minería de datos y de aprendizaje automático para, mediante el contenido publicado en las redes sociales, definir la comunicación verbal del robot de una forma dinámica. Se propone extraer información útil de las redes sociales para construir modelos de conocimiento basados en temas que son de interés para el usuario y en el contexto de la interacción con el robot. Además, se plantea mantener el modelo actualizado de acuerdo con la nueva información que se publique o cambios que pudieran ocurrir desde un punto de vista del usuario., [Abstract] Currently, the interaction capabilities of social robots are limited and, especially in long-term interactions, the human-robot dialogues can be perceived as predictable, repetitive, and unnatural. This situation can lead the user to lose interest in the robot. If we want to bet on a successful and long-lasting coexistence, it is necessary to provide robots with more varied discourses that can adapt to the changing needs of the users. This contribution proposes a methodology that combines data mining and machine learning techniques to dynamically define the robot’s verbal communication through the content published on social networks. We propose to extract useful information from social networks to build knowledge models based on the context of the interaction so that it changes according to the published information.

The detection of offensive, hateful and profane language has become a critical challenge since many users in social networks are exposed to cyberbullying activities on a daily basis. In this paper, we present an analysis of combining different textual features for the detection of hateful or offensive posts on Twitter. We provide a detailed experimental evaluation to understand the impact of each building block in a neural network architecture. The proposed architecture is evaluated on the English Subtask 1A: Identifying Hate, offensive and profane content from the post datasets of HASOC-2021 dataset under the team name TIB-VA. We compared different variants of the contextual word embeddings combined with the character level embeddings and the encoding of collected hate terms.

Analyzing social media posts and comments has become a critical task to prevent cyberbullying and hate speech. In this work we present a classification models based on the attention mechanism to analyze Arabic posts and filter out all kinds of inappropriate speech including Religious based hate speech, offensive and abusive content in different Arabic dialects. The attention-based models show promising results for four Arabic datasets. The results are presented and compared in terms of accuracy and training time.

Purpose: Social media has become a vital part of any institute’s marketing plan. Social networks benefit businesses by allowing them to interact with their clients, grow brand exposure through offers and promotions and find new leads. It also offers vital information concerning the general emotions and sentiments directly connected to the welfare and security of the online community involved with the brand. Big organizations can make use of their social media data to generate planned and operational decisions. This paper aims to look into the conversion of sentiments and emotions over time. Design/methodology/approach: In this work, a model called sentiment urgency emotion detection (SUED) from previous work will be applied on tweets from two different periods of time, one before the start of the COVID-19 pandemic and the other after it started to monitor the conversion of sentiments and emotions over time. The model has been trained to improve its accuracy and F1 score so that the precision and percentage of correctly predicted texts is high. This model will be tuned to improve results (Soussan and Trovati, 2020a; Soussan and Trovati, 2020b) and will be applied on a general business Twitter account of one of the largest chains of supermarkets in the UK to be able to see what sentiments and emotions can be detected and how urgent they are. Findings: This will show the effect of COVID-19 pandemic on the conversions of the sentiments, emotions and urgencies of the tweets. Originality/value: Sentiments will be compared between the two periods to evaluate how sentiments and emotions vary over time taking into consideration the COVID-19 as an affective factor. In addition, SUED will be tuned to enhance results and the knowledge that is mined when turning data into decisions is crucial because it will aid stakeholders handling the institute to evaluate the topics and issues that were mostly emphasized.

Finding relevant information in social media data to satisfy the user need presents unique challenges due to its nature (e.g. high volume, short length, sparseness). This thesis aims to discover informative feature representations that can help to capture user information needs when no annotated data is available. Using state-of-the-art techniques in text mining and information retrieval research, this research proposes novel methods to boost the user information need with representative information in a social media context. The experimental results show that the proposed models outperform baseline models on standard TREC 2011-2014 microblog datasets.

Owing to the peculiar nature of the discourse on Twitter, developing analytical frameworks to derive useful insights from Twitter remains challenging as evidenced by the poor performance at tasks such as reliable demographic inference, affect estimation, and event detection. One of the focal problems lies in analyzing short texts in general, and tweets in particular. The analysis is as such made difficult because of the vagaries of the linguistic expressions and Twitter further exacerbates this by enabling the use of emojis, hashtags, URLs, and embedded media. While the previous research has demonstrated ways of extracting useful information from individual tweet-texts to some extent, a detailed and thorough investigation of the role of metadata has not yet been systematically performed. Furthermore, a majority of the previous work has paid little or no attention to the emerging role of deep learning approaches in Twitter-based analytics. These observations motivate this thesis, which aims to enhance machine understanding of tweets towards deriving deeper insights from the public data on Twitter and inform the scientific objectives of this thesis. First, this thesis sets out to empirically investigate the impact and efficacy of deep learning approaches integrating message-text and metadata leveraging on the distributed semantic representations of textual entities. Second, the thesis contributes towards improving capturing enhanced semantics from tweets by harnessing external, open-sourced knowledge graphs and other crowd-sourced lexical resources. Third, the role of the user-created metadata, such as hashtags and URLs, in machine understanding of tweets is examined and quantified. At the same time, computational models are introduced to derive conversational, topical, and temporal contexts of tweets and utilize them in machine learning models to improve Twitter-based analytics. Validation of the proposed novel machine learning models integrating the diverse, Tiivistelmä Twitter-pohjainen analytiikka on noussut useiden tieteenalojen työkalupakkiin viime vuosina. Kuitenkin, järjestelmällisten analyysikokonaisuuksien kehitys on mikroblog-keskustelujen erityisluonteen vuoksi haastavaa. Analysointimenetelmien heikko suorituskyky on todettu useissa sovelluskohteissa, kuten kirjoittajien väestörakenne- ja tunnetila-analyyseissa taikka tehtävissä, joissa mikrobloggauksista pyritään havaitsemaan tärkeitä tapahtumia. Analyysit pitäisi suorittaa hyvin lyhyistä tekstipätkistä, tässä tutkimuksessa erityisesti mikroblogauksista. Omalaatuisten ja persoonallisten kielellisten ilmaisujen, mutta myös Twitterin emojien, metatietotagien, ulkoisten linkkien (url) ja upotettujen kuvien sekä videoiden käyttö monipuolistaa ongelmakenttää. Aikaisemmissa tutkimuksissa on onnistuttu johtamaan hyödyllistä tietoa yksittäisistä mikroblogauksista jossain määrin, mutta metatietojen roolia ja merkitystä ei ole vielä järjestelmällisesti eikä yksityiskohtaisesti tutkittu. Lisäksi syväoppimisen hyödyntämistä Twitter-pohjaisten datojen analyyseissa on tutkittu vähän tai ei ollenkaan. Tämän väitöskirjan tavoitteena on parantaa tietokoneiden valmiuksia käsitellä mikroblogauksia siten, että nykyistä parempi ja merkityksellisempi julkisten Twitter-aineistojen koneellinen ymmärtäminen olisi mahdollista. Ensinnäkin, tutkimuksessa testataan empiirisesti syväoppivan mallin vaikuttavuutta sekä tehokkuutta ym. tekstikokonaisuuksien hajautetun semanttisen esitysmuodon integroinnissa. Toiseksi, työssä parannetaan mikroblogauksien sisältöanalyysia ulkoisten, avoimen lähdekoodin tietograafien sekä muiden joukkoistettujen sanastojen avulla. Kolmanneksi tutkitaan ja kvantifioidaan käyttäjien luomien metadatojen, kuten metatietotagien ja ulkoisten linkkien roolit analyysikehikoissa. Työssä esitellään laskennalliset mallit mikroblogauksien keskusteluun, aihepiiriin sekä aikaan liittyvien asiayhteyksien päättelemiseksi ja käytetään näitä malleja koneoppimismallien suoritusky

Social media data represent an important resource for behavioral analysis of the aging population. This paper addresses the problem of age prediction from Twitter dataset, where the prediction issue is viewed as a classification task. For this purpose, an innovative model based on Convolutional Neural Network is devised. To this end, we rely on language-related features and social media specific metadata. More specifically, we introduce two features that have not been previously considered in the literature: the content of URLs and hashtags appearing in tweets. We also employ distributed representations of words and phrases present in tweets, hashtags and URLs, pre-trained on appropriate corpora in order to exploit their semantic information in age prediction. We show that our CNN-based classifier, when compared with baseline models, yields an improvement of up to 12.3% for Dutch dataset, 9.8% for English1 dataset, and 6.6% for English2 dataset in the micro-averaged F1 score.

Purpose: Social media has become a vital part of any institute’s marketing plan. Social networks benefit businesses by allowing them to interact with their clients, grow brand exposure through offers and promotions and find new leads. It also offers vital information concerning the general emotions and sentiments directly connected to the welfare and security of the online community involved with the brand. Big organizations can make use of their social media data to generate planned and operational decisions. This paper aims to look into the conversion of sentiments and emotions over time. Design/methodology/approach: In this work, a model called sentiment urgency emotion detection (SUED) from previous work will be applied on tweets from two different periods of time, one before the start of the COVID-19 pandemic and the other after it started to monitor the conversion of sentiments and emotions over time. The model has been trained to improve its accuracy and F1 score so that the precision and percentage of correctly predicted texts is high. This model will be tuned to improve results (Soussan and Trovati, 2020a; Soussan and Trovati, 2020b) and will be applied on a general business Twitter account of one of the largest chains of supermarkets in the UK to be able to see what sentiments and emotions can be detected and how urgent they are. Findings: This will show the effect of COVID-19 pandemic on the conversions of the sentiments, emotions and urgencies of the tweets. Originality/value: Sentiments will be compared between the two periods to evaluate how sentiments and emotions vary over time taking into consideration the COVID-19 as an affective factor. In addition, SUED will be tuned to enhance results and the knowledge that is mined when turning data into decisions is crucial because it will aid stakeholders handling the institute to evaluate the topics and issues that were mostly emphasized.

Social media data represent an important resource for behavioral analysis of the aging population. This paper addresses the problem of age prediction from Twitter dataset, where the prediction issue is viewed as a classification task. For this purpose, an innovative model based on Convolutional Neural Network is devised. To this end, we rely on language-related features and social media specific metadata. More specifically, we introduce two features that have not been previously considered in the literature: the content of URLs and hashtags appearing in tweets. We also employ distributed representations of words and phrases present in tweets, hashtags and URLs, pre-trained on appropriate corpora in order to exploit their semantic information in age prediction. We show that our CNN-based classifier, when compared with baseline models, yields an improvement of up to 12.3% for Dutch dataset, 9.8% for English1 dataset, and 6.6% for English2 dataset in the micro-averaged F1 score.

We contribute a system design and a generalized formal methodology to segment tourists based on their geolocated blogging behaviour according to their interests in identified tourist hotspots. Thus, it is possible to identify and target groups that are possibly interested in alternative destinations to relieve overtourism. A pilot application in a case study of Chinese travel in Switzerland by analysing Qyer travel blog data demonstrates the potential of our method. Accordingly, we contribute four conclusions supported by empirical data. First, our method can enable discovery of plausible geographical distributions of tourist hotspots, which validates the plausibility of the data and its collection. Second, our method discovered statistically significant stochastic dependencies that meaningfully differentiate the observed user base, which demonstrates its value for segmentation. Furthermore, the case study contributes two practical insights for tourism management. Third, Chinese independent travellers, which are the main target group of Qyer, are mainly interested in the discovered travel hotspots, similar to tourists on packaged tours, but also show interest in alternative places. Fourth, the proposed user segmentation revealed two clusters based on users’ social media activity level. For tourism research, users within the second cluster are of interest, which are defined by two segmentation attributes: they blogged about more than just one location, and they have followers. These tourists are significantly more likely to be interested in alternative destinations out of the hotspot axis. Knowing this can help define a target group for marketing activities to promote alternative destinations, + ID der Publikation: hslu_74107 + Art des Beitrages: Wissenschaftliche Medien + Jahrgang: 8 + Sprache: Englisch + Letzte Aktualisierung: 2020-06-25 16:01:57

We contribute a system design and a generalized formal methodology to segment tourists based on their geolocated blogging behaviour according to their interests in identified tourist hotspots. Thus, it is possible to identify and target groups that are possibly interested in alternative destinations to relieve overtourism. A pilot application in a case study of Chinese travel in Switzerland by analysing Qyer travel blog data demonstrates the potential of our method. Accordingly, we contribute four conclusions supported by empirical data. First, our method can enable discovery of plausible geographical distributions of tourist hotspots, which validates the plausibility of the data and its collection. Second, our method discovered statistically significant stochastic dependencies that meaningfully differentiate the observed user base, which demonstrates its value for segmentation. Furthermore, the case study contributes two practical insights for tourism management. Third, Chinese independent travellers, which are the main target group of Qyer, are mainly interested in the discovered travel hotspots, similar to tourists on packaged tours, but also show interest in alternative places. Fourth, the proposed user segmentation revealed two clusters based on users’ social media activity level. For tourism research, users within the second cluster are of interest, which are defined by two segmentation attributes: they blogged about more than just one location, and they have followers. These tourists are significantly more likely to be interested in alternative destinations out of the hotspot axis. Knowing this can help define a target group for marketing activities to promote alternative destinations, + ID der Publikation: hslu_74107 + Art des Beitrages: Wissenschaftliche Medien + Jahrgang: 8 + Sprache: Englisch + Letzte Aktualisierung: 2020-06-25 16:01:57

We contribute a system design and a generalized formal methodology to segment tourists based on their geolocated blogging behaviour according to their interests in identified tourist hotspots. Thus, it is possible to identify and target groups that are possibly interested in alternative destinations to relieve overtourism. A pilot application in a case study of Chinese travel in Switzerland by analysing Qyer travel blog data demonstrates the potential of our method. Accordingly, we contribute four conclusions supported by empirical data. First, our method can enable discovery of plausible geographical distributions of tourist hotspots, which validates the plausibility of the data and its collection. Second, our method discovered statistically significant stochastic dependencies that meaningfully differentiate the observed user base, which demonstrates its value for segmentation. Furthermore, the case study contributes two practical insights for tourism management. Third, Chinese independent travellers, which are the main target group of Qyer, are mainly interested in the discovered travel hotspots, similar to tourists on packaged tours, but also show interest in alternative places. Fourth, the proposed user segmentation revealed two clusters based on users’ social media activity level. For tourism research, users within the second cluster are of interest, which are defined by two segmentation attributes: they blogged about more than just one location, and they have followers. These tourists are significantly more likely to be interested in alternative destinations out of the hotspot axis. Knowing this can help define a target group for marketing activities to promote alternative destinations, + ID der Publikation: hslu_74107 + Art des Beitrages: Wissenschaftliche Medien + Jahrgang: 8 + Sprache: Englisch + Letzte Aktualisierung: 2020-06-25 16:01:57

We contribute a system design and a generalized formal methodology to segment tourists based on their geolocated blogging behaviour according to their interests in identified tourist hotspots. Thus, it is possible to identify and target groups that are possibly interested in alternative destinations to relieve overtourism. A pilot application in a case study of Chinese travel in Switzerland by analysing Qyer travel blog data demonstrates the potential of our method. Accordingly, we contribute four conclusions supported by empirical data. First, our method can enable discovery of plausible geographical distributions of tourist hotspots, which validates the plausibility of the data and its collection. Second, our method discovered statistically significant stochastic dependencies that meaningfully differentiate the observed user base, which demonstrates its value for segmentation. Furthermore, the case study contributes two practical insights for tourism management. Third, Chinese independent travellers, which are the main target group of Qyer, are mainly interested in the discovered travel hotspots, similar to tourists on packaged tours, but also show interest in alternative places. Fourth, the proposed user segmentation revealed two clusters based on users’ social media activity level. For tourism research, users within the second cluster are of interest, which are defined by two segmentation attributes: they blogged about more than just one location, and they have followers. These tourists are significantly more likely to be interested in alternative destinations out of the hotspot axis. Knowing this can help define a target group for marketing activities to promote alternative destinations, + ID der Publikation: hslu_74107 + Art des Beitrages: Wissenschaftliche Medien + Jahrgang: 8 + Sprache: Englisch + Letzte Aktualisierung: 2020-06-25 16:01:57

Inferring demographic intelligence from unlabeled social media data is an actively growing area of research, challenged by low availability of ground truth annotated training corpora. High-accuracy approaches for labeling demographic traits of social media users employ various heuristics that do not scale up and often discount non-English texts and marginalized users. First, we present a framework for inferring the demographic attributes of Twitter users from their profile pictures (avatars) using the Microsoft Azure Face API. Second, we measure the inter-rater agreement between annotations made using our framework against two pre-labeled samples of Twitter users (N1=1163; N2=659) whose age labels were manually annotated. Our results indicate that the strength of the inter-rater agreement (Gwet’s AC1=0.89; 0.90) between the gold standard and our approach is ‘very good’ for labelling the age group of users. The paper provides a use case of Computer Vision for enabling the development of large cross-sectional labeled datasets, and further advances novel solutions in the field of demographic inference from short social media texts.

We contribute a system design and a generalized formal methodology to segment tourists based on their geolocated blogging behaviour according to their interests in identified tourist hotspots. Thus, it is possible to identify and target groups that are possibly interested in alternative destinations to relieve overtourism. A pilot application in a case study of Chinese travel in Switzerland by analysing Qyer travel blog data demonstrates the potential of our method. Accordingly, we contribute four conclusions supported by empirical data. First, our method can enable discovery of plausible geographical distributions of tourist hotspots, which validates the plausibility of the data and its collection. Second, our method discovered statistically significant stochastic dependencies that meaningfully differentiate the observed user base, which demonstrates its value for segmentation. Furthermore, the case study contributes two practical insights for tourism management. Third, Chinese independent travellers, which are the main target group of Qyer, are mainly interested in the discovered travel hotspots, similar to tourists on packaged tours, but also show interest in alternative places. Fourth, the proposed user segmentation revealed two clusters based on users’ social media activity level. For tourism research, users within the second cluster are of interest, which are defined by two segmentation attributes: they blogged about more than just one location, and they have followers. These tourists are significantly more likely to be interested in alternative destinations out of the hotspot axis. Knowing this can help define a target group for marketing activities to promote alternative destinations, + ID der Publikation: hslu_74107 + Art des Beitrages: Wissenschaftliche Medien + Jahrgang: 8 + Sprache: Englisch + Letzte Aktualisierung: 2020-06-25 16:01:57

We contribute a system design and a generalized formal methodology to segment tourists based on their geolocated blogging behaviour according to their interests in identified tourist hotspots. Thus, it is possible to identify and target groups that are possibly interested in alternative destinations to relieve overtourism. A pilot application in a case study of Chinese travel in Switzerland by analysing Qyer travel blog data demonstrates the potential of our method. Accordingly, we contribute four conclusions supported by empirical data. First, our method can enable discovery of plausible geographical distributions of tourist hotspots, which validates the plausibility of the data and its collection. Second, our method discovered statistically significant stochastic dependencies that meaningfully differentiate the observed user base, which demonstrates its value for segmentation. Furthermore, the case study contributes two practical insights for tourism management. Third, Chinese independent travellers, which are the main target group of Qyer, are mainly interested in the discovered travel hotspots, similar to tourists on packaged tours, but also show interest in alternative places. Fourth, the proposed user segmentation revealed two clusters based on users’ social media activity level. For tourism research, users within the second cluster are of interest, which are defined by two segmentation attributes: they blogged about more than just one location, and they have followers. These tourists are significantly more likely to be interested in alternative destinations out of the hotspot axis. Knowing this can help define a target group for marketing activities to promote alternative destinations, + ID der Publikation: hslu_74107 + Art des Beitrages: Wissenschaftliche Medien + Jahrgang: 8 + Sprache: Englisch + Letzte Aktualisierung: 2020-06-25 16:01:57

We contribute a system design and a generalized formal methodology to segment tourists based on their geolocated blogging behaviour according to their interests in identified tourist hotspots. Thus, it is possible to identify and target groups that are possibly interested in alternative destinations to relieve overtourism. A pilot application in a case study of Chinese travel in Switzerland by analysing Qyer travel blog data demonstrates the potential of our method. Accordingly, we contribute four conclusions supported by empirical data. First, our method can enable discovery of plausible geographical distributions of tourist hotspots, which validates the plausibility of the data and its collection. Second, our method discovered statistically significant stochastic dependencies that meaningfully differentiate the observed user base, which demonstrates its value for segmentation. Furthermore, the case study contributes two practical insights for tourism management. Third, Chinese independent travellers, which are the main target group of Qyer, are mainly interested in the discovered travel hotspots, similar to tourists on packaged tours, but also show interest in alternative places. Fourth, the proposed user segmentation revealed two clusters based on users’ social media activity level. For tourism research, users within the second cluster are of interest, which are defined by two segmentation attributes: they blogged about more than just one location, and they have followers. These tourists are significantly more likely to be interested in alternative destinations out of the hotspot axis. Knowing this can help define a target group for marketing activities to promote alternative destinations, + ID der Publikation: hslu_74107 + Art des Beitrages: Wissenschaftliche Medien + Jahrgang: 8 + Sprache: Englisch + Letzte Aktualisierung: 2020-06-25 16:01:57

We contribute a system design and a generalized formal methodology to segment tourists based on their geolocated blogging behaviour according to their interests in identified tourist hotspots. Thus, it is possible to identify and target groups that are possibly interested in alternative destinations to relieve overtourism. A pilot application in a case study of Chinese travel in Switzerland by analysing Qyer travel blog data demonstrates the potential of our method. Accordingly, we contribute four conclusions supported by empirical data. First, our method can enable discovery of plausible geographical distributions of tourist hotspots, which validates the plausibility of the data and its collection. Second, our method discovered statistically significant stochastic dependencies that meaningfully differentiate the observed user base, which demonstrates its value for segmentation. Furthermore, the case study contributes two practical insights for tourism management. Third, Chinese independent travellers, which are the main target group of Qyer, are mainly interested in the discovered travel hotspots, similar to tourists on packaged tours, but also show interest in alternative places. Fourth, the proposed user segmentation revealed two clusters based on users’ social media activity level. For tourism research, users within the second cluster are of interest, which are defined by two segmentation attributes: they blogged about more than just one location, and they have followers. These tourists are significantly more likely to be interested in alternative destinations out of the hotspot axis. Knowing this can help define a target group for marketing activities to promote alternative destinations, + ID der Publikation: hslu_74107 + Art des Beitrages: Wissenschaftliche Medien + Jahrgang: 8 + Sprache: Englisch + Letzte Aktualisierung: 2020-06-25 16:01:57

Social networking and microblogging platforms such as Twitter, Facebook, Tumbler, Instagram etc., have become an integral part of day to day lives of people today for real-time information exchanging. Twitter has become one of the most popular microblogging avenues in the recent years. Today, it hosts more than 300 million monthly active users and generates more than 500 million tweets everyday. Twitter users both publish messages as well as search for messages. Current search results given by Twitter are chronologically ordered and often users have to manually scan through an overwhelming number of tweets to find the content of interest. This process can quickly become infeasible. Personalization techniques address this problem by learning the user interests and tailoring search results by matching them with the user’s interests. There has been a tremendous amount of work done in the domain of web search results personalization. However, research on personalization in microblogging environments such as Twitter, is very sparse. Microblogging environments differ from traditional web environments in several ways: microblogs are very short compared to web documents, they are noisy with informal language and they contain special entities such as hashtags and user mentions. Compared to the web domain, microblogging environments are rich in social interactions as well. Therefore, in this thesis, we propose novel approaches to personalized search in microblogging environments using the cutting-edge technologies used in the text/data mining research. We use Twitter as a specific use case. Our first approach is based on the use of topic modelling algorithms for search results personalization. Coping with sparsity is one of the major challenges when applying topic modelling algorithms on short text documents such as tweets. We conduct an in-depth investigation of how topic modelling algorithms can be applied on short text environments and propose a new tweet grouping scheme t

In this work, we consider the exploitation of social media data in the context of Remote Sensing and Spatial Information Sciences. To this end, we explore a way of augmenting and integrating information represented by geo-located feature vectors into a system for the classification of satellite images. For that purpose, we present a quite general data fusion framework based on Convolutional Neural Network (CNN) and an initial examination of our approach on features from geo-located social media postings on the Twitter and Sentinel images. For this examination, we selected six simple Twitter features derived from the metadata, which we believe could contain information for the spatial context. We present initial experiments using geotagged Twitter data from Washington DC and Sentinel images showing this area. The goal of classification is to determine local climate zones (LCZ). First, we test whether our selected feature maps are not correlated with the LCZ classification at the geo-tag position. We apply a simple boost tree classifier on this data. The result turns out not to be a mere random classifier. Therefore, this data can be correlated with LCZ. To show the improvement by our method, we compare classification with and without the Twitter feature maps. In our experiments, we apply a standard pixel-based CNN classification of the Sentinel data and use it as a baseline model. After that, we expand the input augmenting additional Twitter feature maps within the CNN and assess the contribution of these additional features to the overall F1-score of the classification, which we determine from spatial cross-validation.

Social media data represent an important resource for behavioral analysis of the ageing population. This paper addresses the problem of age prediction from Twitter dataset, where the prediction issue is viewed as a classification task. For this purpose, an innovative model based on Convolutional Neural Network is devised. To this end, we rely on language-related features and social media specific metadata. More specifically, we introduce two features that have not been previously considered in the literature: the content of URLs and hashtags appearing in tweets. We also employ distributed representations of words and phrases present in tweets, hashtags and URLs, pre-trained on appropriate corpora in order to exploit their semantic information in age prediction. We show that our CNN-based classifier, when compared with an SVM baseline model, yields an improvement of 12.3% and 6.6% in the micro-averaged F1 score on the Dutch and English datasets, respectively.

[Absctract]: The usage of part-of-day nouns, such as ‘night’, and their time-specific greetings (‘good night’), varies across languages and cultures. We show the possibilities that Twitter offers for studying the semantics of these terms and its variability between countries. We mine a worldwide sample of multilingual tweets with temporal greetings, and study how their frequencies vary in relation with local time. The results provide insights into the semantics of these temporal expressions and the cultural and sociological factors influencing their usage.

Due to the emergence of social media, there is a paradigm shift in the area of information production, processing and consumption. Hence, investigation in the utilization of open social media data is a relevant research topic. The openness of data, social media data, enables innovation and societal value creation. Social media analytics is an evolving research domain with interdisciplinary methods that are common in data mining such as text mining, social network analysis, trend analysis, and sentiment analysis. Also, social media analytics deals with development and evaluation of frameworks and informatics tools to process noisy and unstructured social media data. On the other hand, Internet of Things (IoT) enables the utilization of digital artifacts with well-established solutions and allows things to be connected regardless of location and time. However, a literature review about social media analytics and IoT integration is missing. In this paper, we conducted a systematic literature review of social media analytics and IoT integration. The literature review indicates that there are fewer research works done in the area of social media analytics and IoT compared to Data Mining and IoT. This paper facilitates discussion and elicits research potentials in social media analytics and IoT integration.

Social online media increasingly emerge as important informal information sources that can contribute to the detection of trends and early warnings in critical fields such as public health monitoring or emergency management. In the face of global environmental challenges the utilization of this information in ecological monitoring contexts has been called for, but examples remain sparse. This can be attributed to the significant technical challenges in processing this data and concerns about the quality, reliability and applicability of information mined from social media to the ecological domain. Here the strength and weaknesses of social media mining for ecological monitoring are assessed using the micro-blogging service Twitter and invasive alien species (IAS) monitoring as an example. The assessment is based on a manual analysis of 2842 Tweets sampled from Twitter data with potential direct or descriptive references to IAS impacting forest ecosystems, which was collected over a period of nearly three years. The results are presented as information topologies for Twitter messages of observational and non-observational character for three IAS with distinctive characteristics (Oak Processionary Moth, Emerald Ash Borer, Eastern Grey Squirrel). The results show that the social media channel Twitter is a rich source of primary and secondary observational biodiversity information. It also provides useful insights in the topical landscape of public communications on IAS as well as the public perception of IAS and IAS management. The analysis suggests broad application opportunities in IAS monitoring and management, and points at applications for related environmental questions. The results highlight that social media mining for ecological monitoring needs to be approached with the same best practices as ecological monitoring in general, requiring a good understanding of the monitored subjects and specific monitoring questions. The challenges in utilizing this informatio

There is need for an integrated event focused crawling system to collect Web data about key events. When an event occurs, many users try to locate the most up-to-date information about that event. Yet, there is little systematic collecting and archiving anywhere of information about events. We propose intelligent event focused crawling for automatic event tracking and archiving, as well as effective access. We extend the traditional focused (topical) crawling techniques in two directions, modeling and representing: events and webpage source importance. We developed an event model that can capture key event information (topical, spatial, and temporal). We incorporated that model into the focused crawler algorithm. For the focused crawler to leverage the event model in predicting a webpage's relevance, we developed a function that measures the similarity between two event representations, based on textual content. Although the textual content provides a rich set of features, we proposed an additional source of evidence that allows the focused crawler to better estimate the importance of a webpage by considering its website. We estimated webpage source importance by the ratio of number of relevant webpages to non-relevant webpages found during crawling a website. We combined the textual content information and source importance into a single relevance score. For the focused crawler to work well, it needs a diverse set of high quality seed URLs (URLs of relevant webpages that link to other relevant webpages). Although manual curation of seed URLs guarantees quality, it requires exhaustive manual labor. We proposed an automated approach for curating seed URLs using social media content. We leveraged the richness of social media content about events to extract URLs that can be used as seed URLs for further focused crawling. We evaluated our system through four series of experiments, using recent events: Orlando shooting, Ecuador earthquake, Panama papers, California shoot

In view of the substantial costs associated with classic monitoring networks, participatory data collection methods can be deemed a promising option to obtain complementary data. An emerging trend in this field is social media mining, i.e., harvesting of pre-existing, crowd-generated data from social media. Although this approach is participatory in a broader sense, the users are mostly not aware of their participation in research.Inspired by this novel development, we demonstrate in this study that it is possible to derive a water level time series from the analysis of multiple YouTube videos. As an example, we studied the recent water level rise in Dahl Hith, a Saudi Arabian cave. To do so, we screened 16 YouTube videos of the cave for suitable reference points (e.g., cave graffiti). Then, we visually estimated the distances between these points and the water level and traced their changes over time. To bridge YouTube hiatuses, we considered own photos taken during two site visits. For the time period 2013–2014, we estimate a rise of 9.5 m. The fact that this rise occurred at a somewhat constant rate of roughly 0.4 m per month points towards a new and permanent water source, possibly two nearby lakes formed from treated sewage effluent. An anomaly in the rising rate is noted for autumn 2013 (1.3 m per month). As this increased pace coincides with a cluster of rain events, we deem rapid groundwater recharge along preferential flow paths a likely cause. Despite the sacrifice in precision, we believe that YouTube harvesting may represent a viable option to gather historical water levels in data-scarce settings and that it could be adapted to other environments (e.g., flood extents). In certain areas, it might provide an additional tool for the monitoring toolbox, thereby possibly delivering hydrological data for water resources management.

Public libraries are increasingly using social media to connect their users in innovative ways. Librarians make use of social media as a tool for ‘being part of their communities’, promoting library services and events, and creating participatory services. However, little empirical investigation has explored the success of such social media use by libraries. In this paper, we study the role of Twitter for engaging users via a focus on public libraries. We use topic-modeling techniques to classify library user engagement strategies into four categories: literature exhibits, engaging topics, community building and library showcasing. These four engagement strategies are re-examined via a sentiment analysis of tweets collected from 10 public libraries over 3 months. Through data mining of tweets, we explore how user engagement strategies are used by libraries on Twitter, and suggest the best practices for libraries interested in pursuing social media initiatives to use to engage their users effectively.

Forest monitoring captures human impacts and other biotic and abiotic influences on forests and is a prerequisite for the sustainable use and protection of forest ecosystems. Forest inventories for example are a key tool to plan sustainable harvesting, whereas Forest Observational Studies provide the empirical basis for an improved understanding and long-term evaluation of forest ecosystem dynamics. To that end detailed data is collected at stand level, often integrated in larger forest observational networks, which feeds into forest ecosystem models. Forests exist however in a constantly changing societal context and the direct or indirect impact of human activity has become a crucial driver on all types of ecosystems. The Millenium Ecosystem Assessment underlines the linkage between social and ecological systems, highlighting the centrality of ecosystem services to human well-being and the requirement for ecosystem monitoring in the anthropocene to provide a holistic view of ecosystems as social-ecological systems. Framing information about the social context of a forest ecosystem, gaining the expertise and providing resources to collect this type of information is usually outside the scope of data collection for forest inventories and monitoring. Studies in other domains faced a similar challenge and turned to data mining informal online information sources to supplement traditional monitoring and data collection strategies. This paper explores how forest monitoring approaches especially Forest Observational Studies with their long-term and large-scale focus may be complemented by social media mining. We outline (a) how social media mining methods from other domains could be applied to forest monitoring, (b) discuss identification of stakeholders, events and demands on forest ecosystems as examples of social contextual information that could be obtained via this route and (c) explain how this information could be automatically mined from social media, online news, AuthorCount:3

Recent years have witnessed a surge of interest in computational methods for affect, ranging from opinion mining, to subjectivity detection, to sentiment and emotion analysis. This article presents a brief overview of the latest trends in the field and describes the manner in which the articles contained in the special issue contribute to the advancement of the area. Finally, we comment on the current challenges and envisaged developments of the subjectivity and sentiment analysis fields, as well as their application to other Natural Language Processing tasks and related domains.

Recent years have witnessed a surge of interest in computational methods for affect, ranging from opinion mining, to subjectivity detection, to sentiment and emotion analysis. This article presents a brief overview of the latest trends in the field and describes the manner in which the articles contained in the special issue contribute to the advancement of the area. Finally, we comment on the current challenges and envisaged developments of the subjectivity and sentiment analysis fields, as well as their application to other Natural Language Processing tasks and related domains.

Práce se věnuje důsledkům plně rozvinuté infrastruktury webu 2.0 a jeho obecnému celospolečenskému přijetí, jež jsou, v doprovodu dostupného výpočetního výkonu, základními faktory umožňujícími rozvoj masivního zpracování uživatelských dat. Práce Grid v této souvislosti využívá veřejně dostupných dat Facebooku o aktivitě uživatelů na stránkách českého pravicového populismu, která plně automatizovaně získává a dále statisticky analyzuje. Výsledek tohoto procesu je poté prezentován na webových stránkách designově i stylisticky podobných webům krajní pravice. Jejím příznivcům na první pohled ideově blízký web však prezentuje empirická data, která o mnohé slávě těchto hnutí, ani jejich přívrženců nevypovídají. Projekt Grid tak stojí na hranici amatérského sociologického výzkumu a umělecké intervence do mediálního prostoru., Work aims at consequences of fully developed infrastructure of web 2.0 and its wide social acceptance, which with cheap computing power leads to massive data interpretation. In relation to that this work uses public Facebook's data about user activity at right wing oriented Facebook pages which gathers and statisticaly analyzes. Result is then showed at web page which is designed to look close to the look of extreme right wing webs. Positive first sight impression of right wing sympathizers is then eroded with not so good empirical statistics. Under 100 users are commenting quite often so this right wing craziness looks like a small movement in the end. Therefore Grid stands somewhere between amateur sociol research and artistic intervention into public media space.

In the domain of information security, real-time information is critical if a decision-maker needs to assess systems and people to produce actionable intelligence. Twitter supplies a platform for fast – nearly real-time – distribution of information. Millions of Twitter users generate an overwhelming amount of data every single day. Assessing the value of information and extracting intelligence out of the data produced by social media users opens a lot of challenges. A recent example are fake-News and automated bots that try to influence the political views of millions of social media users. This study builds on an existing system that mines security related information from the social media platform Twitter. A first, rough analysis of the collected data revealed a possibly very high number of automatically created messages. Validating the information source and categorizing the social media account that distributes information, is a first step to assess whether the shared content is generated by automatically. Existing bot detection methodologies can be applied only on accounts satisfying certain conditions. Because bots get more sophisticated, the need of enhanced systems for supporting the process of bot detection is pressing. In contrast to prior research, we apply existing models on a specific domain. We answer the question of “Who shares with whom security information on Twitter?” Combining the results from bot detection with methodologies used in social network analysis helps us identifying the most influential users in the domain of security information. The final outcome of this thesis is an instantiation of methodologies and models for categorizing social media users of a specific domain. An additional contribution is structured analysis of the underlying social network. Beside the instantiation of bot detection we show how the influence of bots can be measured., von Severin Alton, B.Sc., Abweichender Titel laut Übersetzung der Verfasserin/des Verfassers, Masterarbeit Universität Innsbruck 2018

Práce se věnuje důsledkům plně rozvinuté infrastruktury webu 2.0 a jeho obecnému celospolečenskému přijetí, jež jsou, v doprovodu dostupného výpočetního výkonu, základními faktory umožňujícími rozvoj masivního zpracování uživatelských dat. Práce Grid v této souvislosti využívá veřejně dostupných dat Facebooku o aktivitě uživatelů na stránkách českého pravicového populismu, která plně automatizovaně získává a dále statisticky analyzuje. Výsledek tohoto procesu je poté prezentován na webových stránkách designově i stylisticky podobných webům krajní pravice. Jejím příznivcům na první pohled ideově blízký web však prezentuje empirická data, která o mnohé slávě těchto hnutí, ani jejich přívrženců nevypovídají. Projekt Grid tak stojí na hranici amatérského sociologického výzkumu a umělecké intervence do mediálního prostoru., Work aims at consequences of fully developed infrastructure of web 2.0 and its wide social acceptance, which with cheap computing power leads to massive data interpretation. In relation to that this work uses public Facebook's data about user activity at right wing oriented Facebook pages which gathers and statisticaly analyzes. Result is then showed at web page which is designed to look close to the look of extreme right wing webs. Positive first sight impression of right wing sympathizers is then eroded with not so good empirical statistics. Under 100 users are commenting quite often so this right wing craziness looks like a small movement in the end. Therefore Grid stands somewhere between amateur sociol research and artistic intervention into public media space.

Práce se věnuje důsledkům plně rozvinuté infrastruktury webu 2.0 a jeho obecnému celospolečenskému přijetí, jež jsou, v doprovodu dostupného výpočetního výkonu, základními faktory umožňujícími rozvoj masivního zpracování uživatelských dat. Práce Grid v této souvislosti využívá veřejně dostupných dat Facebooku o aktivitě uživatelů na stránkách českého pravicového populismu, která plně automatizovaně získává a dále statisticky analyzuje. Výsledek tohoto procesu je poté prezentován na webových stránkách designově i stylisticky podobných webům krajní pravice. Jejím příznivcům na první pohled ideově blízký web však prezentuje empirická data, která o mnohé slávě těchto hnutí, ani jejich přívrženců nevypovídají. Projekt Grid tak stojí na hranici amatérského sociologického výzkumu a umělecké intervence do mediálního prostoru., Work aims at consequences of fully developed infrastructure of web 2.0 and its wide social acceptance, which with cheap computing power leads to massive data interpretation. In relation to that this work uses public Facebook's data about user activity at right wing oriented Facebook pages which gathers and statisticaly analyzes. Result is then showed at web page which is designed to look close to the look of extreme right wing webs. Positive first sight impression of right wing sympathizers is then eroded with not so good empirical statistics. Under 100 users are commenting quite often so this right wing craziness looks like a small movement in the end. Therefore Grid stands somewhere between amateur sociol research and artistic intervention into public media space.

The analysis of sentiment contained in vast amounts of Twitter messages has reliably shown seasonal patterns of variation in multiple studies, a finding that can have great importance in the understanding of seasonal affective disorders, particularly if related with known seasonal variations in certain hormones. An important question, however, is that of directly linking the signals coming from Twitter with other sources of evidence about average mood changes. Specifically we compare Twitter signals relative to anxiety, sadness, anger, and fatigue with purchase of items related to anxiety, stress and fatigue at a major UK Health and Beauty retailer. Results show that all of these signals are highly correlated and strongly seasonal, being under-expressed in the summer and over-expressed in the other seasons, with interesting differences and similarities across them. Anxiety signals, extracted from both Twitter and from Health product purchases, peak in spring and autumn, and correlate also with the purchase of stress remedies, while Twitter sadness has a peak in the Winter, along with Twitter anger and remedies for fatigue. Surprisingly, purchase of remedies for fatigue do not match the Twitter fatigue, suggesting that perhaps the names we give to these indicators are only approximate indications of what they actually measure. This study contributes both to the clarification of the mood signals contained in social media, and more generally to our understanding of seasonal cycles in collective mood.

The analysis of sentiment contained in vast amounts of Twitter messages has reliably shown seasonal patterns of variation in multiple studies, a finding that can have great importance in the understanding of seasonal affective disorders, particularly if related with known seasonal variations in certain hormones. An important question, however, is that of directly linking the signals coming from Twitter with other sources of evidence about average mood changes. Specifically we compare Twitter signals relative to anxiety, sadness, anger, and fatigue with purchase of items related to anxiety, stress and fatigue at a major UK Health and Beauty retailer. Results show that all of these signals are highly correlated and strongly seasonal, being under-expressed in the summer and over-expressed in the other seasons, with interesting differences and similarities across them. Anxiety signals, extracted from both Twitter and from Health product purchases, peak in spring and autumn, and correlate also with the purchase of stress remedies, while Twitter sadness has a peak in the Winter, along with Twitter anger and remedies for fatigue. Surprisingly, purchase of remedies for fatigue do not match the Twitter fatigue, suggesting that perhaps the names we give to these indicators are only approximate indications of what they actually measure. This study contributes both to the clarification of the mood signals contained in social media, and more generally to our understanding of seasonal cycles in collective mood.

The analysis of sentiment contained in vast amounts of Twitter messages has reliably shown seasonal patterns of variation in multiple studies, a finding that can have great importance in the understanding of seasonal affective disorders, particularly if related with known seasonal variations in certain hormones. An important question, however, is that of directly linking the signals coming from Twitter with other sources of evidence about average mood changes. Specifically we compare Twitter signals relative to anxiety, sadness, anger, and fatigue with purchase of items related to anxiety, stress and fatigue at a major UK Health and Beauty retailer. Results show that all of these signals are highly correlated and strongly seasonal, being under-expressed in the summer and over-expressed in the other seasons, with interesting differences and similarities across them. Anxiety signals, extracted from both Twitter and from Health product purchases, peak in spring and autumn, and correlate also with the purchase of stress remedies, while Twitter sadness has a peak in the Winter, along with Twitter anger and remedies for fatigue. Surprisingly, purchase of remedies for fatigue do not match the Twitter fatigue, suggesting that perhaps the names we give to these indicators are only approximate indications of what they actually measure. This study contributes both to the clarification of the mood signals contained in social media, and more generally to our understanding of seasonal cycles in collective mood.

Today, the web is constantly growing, expanding global information space and more and more data is being processed and sourced online. The amount of electronically accessible and available online information is overwhelming. Increasingly, recommendation systems, which engage in some form of automated personalisation, are hugely prevalent on the web and have been extensively studied in the research literature. Several issues still remain unsolved including high sparsity situation and cold starts (how to recommend content to users who have had little or no prior interaction with the system). Recent work has demonstrated a potential solution in the form of cross-domain user modeling. This thesis will explore the design, implementation and testing of a cross-domain approach using social media data to model rich and effective user preferences and provide empirical evidence of the effectiveness of the approach based on direct real-world user feedback, deconstructing a cross-system news recommendation service where user models are generated via social media data. This will be accomplished by identifying the availability of a source domain from which to draw resources for recommendations and the availability of user profiles that capture a wide range of user interests from different domains. This thesis also demonstrates the viability of generating user models from social media data and evidences that the automated cross-domain approach can be superior to explicit filtering using self-declared preferences and can be further augmented when placing the user with the ability to maintain control over such models. The reasons for these results are qualitatively examined in order to understand why such effects occur, indicating that different models are capturing widely different areas within a user's preference space.