Recent research has shown that different tagging motivation and user behavior can effect the overall usefulness of social tagging systems for certain tasks. In this paper, we provide further evidence for this observation by demonstrating that tagging data obtained from certain types of users- so-called Categorizers- outperforms data from other users on a social classification task. We show that segmenting users based on their tagging behavior has significant impact on the performance of automated classification of tagged data by using (i) tagging data from two different social tagging systems, (ii) a Support Vector Machine as a classification mechanism and (iii) existing classification systems such as the Library of Congress Classification System as ground truth. Our results are relevant for scientists studying pragmatics and semantics of social tagging systems as well as for engineers interested in influencing emerging properties of deployed social tagging systems.

Support Vector Machines present an interesting and effective approach to solve automated classification tasks. Although it only handles binary and supervised problems by nature, it has been transformed into multiclass and semi-supervised approaches in several works. A previous study on supervised and semi-supervised SVM classification over binary taxonomies showed how the latter clearly outperforms the former, proving the suitability of unlabeled data for the learning phase in this kind of tasks. However, the suitability of unlabeled data for multiclass tasks using SVM has never been tested before. In this work, we present a study on whether unlabeled data could improve results for multiclass web page classification tasks using Support Vector Machines. As a conclusion, we encourage to rely only on labeled data, both for improving (or at least equaling) performance and for reducing the computational cost. 1

Ranking tweets is a fundamental task to make it easier to distill the vast amounts of information shared by users. In this paper, we explore the novel idea of ranking tweets on a topic using heterogeneous networks. We construct heterogeneous networks by harnessing cross-genre linkages between tweets and semantically-related web documents from formal genres, and inferring implicit links between tweets and users. To rank tweets effectively by capturing the semantics and importance of different linkages, we introduce Tri-HITS, a model to iteratively propagate ranking scores across heterogeneous networks. We show that integrating both formal genre and inferred social networks

Ranking tweets is a fundamental task to make it easier to distill the vast amounts of information shared by users. In this paper, we explore the novel idea of ranking tweets on a topic using heterogeneous networks. We construct heterogeneous networks by harnessing cross-genre linkages between tweets and semantically-related web documents from formal genres, and inferring implicit links between tweets and users. To rank tweets effectively by capturing the semantics and importance of different linkages, we introduce Tri-HITS, a model to iteratively propagate ranking scores across heterogeneous networks. We show that integrating both formal genre and inferred social networks

We present and evaluate several hybrid sys-tems for sentiment identification for Twit-ter, both at the phrase and document (tweet) level. Our approach has been to use a novel combination of lexica, traditional NLP and deep learning features. We also analyse tech-niques based on syntactic parsing and token-based association to handle topic specific sen-timent in subtask C. Our strategy has been to identify subphrases relevant to the designated topic/target and assign sentiment according to our subtask A classifier. Our submitted subtask A classifier ranked fourth in the Se-mEval official results while our BASELINE and µPARSE classifiers for subtask C would have ranked second. 1

While Twitter provides an unprecedented opportunity to learn about breaking news and current events as they happen, it often produces skepticism among users as not all the information is accurate but also hoaxes are sometimes spread. While avoiding the diffusion of hoaxes is a major concern during fast-paced events such as natural disasters, the study of how users trust and verify information from tweets in these contexts has received little attention so far. We survey users on credibility perceptions regarding witness pictures posted on Twitter related to Hurricane Sandy. By examining credibility perceptions on features suggested for information verification in the field of Epistemology, we evaluate their accuracy in determining whether pictures were real or fake compared to professional evaluations performed by experts. Our study unveils insight about tweet presentation, as well as features that users should look at when assessing the veracity of tweets in the context of fast-paced events. Some of our main findings include that while author details not readily available on Twitter feeds should be emphasized in order to facilitate verification of tweets, showing multiple tweets corroborating a fact misleads users to trusting what actually is a hoax. We contrast some of the behavioral patterns found on tweets with literature in Psychology research.

Resumen: Este art́ıculo presenta un resumen de la tarea conjunta que tuvo lugar en el marco del taller TweetMT celebrado junto con SEPLN 2015, que consiste en traducir diversas colecciones de tweets en varios lenguajes. El art́ıculo describe el proceso de recolección y anotación de datos, el desarrollo y evaluación de la tarea y los resultados obtenidos por los participantes. Palabras clave: Traducción Automática, Microblogs, Tuits, Social Media Abstract: This article presents an overview of the shared task that took place as part of the TweetMT workshop held at SEPLN 2015. The task consisted in translating collections of tweets from and to several languages. The article outlines the data collection and annotation process, the development and evaluation of the shared task, as well as the results achieved by the participants.

Resumen: Este art́ıculo presenta un resumen de la tarea compartida y taller TweetLID, organizado junto a SEPLN 2014. Resume brevemente el proceso de colección y anotación de datos, el desarrollo y evaluación de la tarea compartida, y por último, los resultados obtenidos por los participantes. Palabras clave: identificación de idioma, tuits, textos cortos, multilingüismo Abstract: This article presents a summary of the TweetLID shared task and workshop held at SEPLN 2014. It briefly summarizes the data collection and annotation process, the development and evaluation of the shared task, as well as the results achieved by the participants.

101 Reports on the Workshops Held at the Sixth International

Disambiguation to Wikipedia (D2W) is the task of linking mentions of concepts in text to their corresponding Wikipedia entries. Most previous work has focused on linking terms in formal texts (e.g. newswire) to Wikipedia. Linking terms in short informal texts (e.g. tweets) is difficult for systems and humans alike as they lack a rich disambiguation context. We first evaluate an existing Twitter dataset as well as the D2W task in general. We then test the effects of two tweet context expansion methods, based on tweet authorship and topic-based clustering, on a state-of-the-art D2W system and evaluate the results.