61 research outputs found
Comparative Experiments Using Supervised Learning and Machine Translation for Multilingual Sentiment Analysis
No full textSentiment analysis is the Natural Language Processing task dealing with sentiment detection and classification from texts.
In recent years, due to the growth in the quantity and fast spreading of user-generated contents online and the impact such information has on events, people and companies worldwide,
this task has been approached in an important body of research in the field.
Despite different methods having been proposed for distinct types of text, the research community has concentrated less on developing methods for languages other than English.
In the abovementioned context, the present work studies the possibility to employ Machine Translation systems and supervised methods to build models able to detect and classify sentiment in languages for which less/no resources are available for this task.
Our extensive evaluation scenarios show that MT systems are approaching a good level of maturity and that they can, in combination to appropriate machine learning algorithms and carefully chosen features, be used to build sentiment analysis systems that can obtain comparable performances to the one obtained for English.JRC.G.2 - Global security and crisis managemen
Multilingual Sentiment Analysis using Machine Translation?
No full textThe past years have shown a steady growth
in interest in the Natural Language Processing
task of sentiment analysis. The research
community in this field has actively proposed
and improved methods to detect and classify
the opinions and sentiments expressed in different
types of text - from traditional press articles,
to blogs, reviews, fora or tweets. A less
explored aspect has remained, however, the
issue of dealing with sentiment expressed in
texts in languages other than English. To this
aim, the present article deals with the problem
of sentiment detection in three different
languages - French, German and Spanish - using
three distinct Machine Translation (MT)
systems - Bing, Google and Moses. Our extensive
evaluation scenarios show that SMT
systems are mature enough to be reliably employed
to obtain training data for languages
other than English and that sentiment analysis
systems can obtain comparable performances
to the one obtained for English.JRC.G.2 - Global security and crisis managemen
Improving Sentiment Analysis in Twitter Using Multilingual Machine Translated Data
No full textSentiment analysis is currently a very dynamic field in Computational Linguistics. Research herein has concentrated on the development
of methods and resources for different types of texts and various languages. Nonetheless, the implementation of a multilingual system
that is able to classify sentiment expressed in various languages has not been approached so far. The main challenge this paper addresses is sentiment analysis from tweets in a multilingual setting. We first build a simple sentiment analysis system for tweets in English. Subsequently, we translate the data from English to four other languages - Italian, Spanish, French and German - using a standard machine translation system. Further on, we manually correct the test data and create Gold Standards for each of the target languages. Finally, we test the performance of the sentiment analysis classifiers for the different languages concerned
and show that the joint use of training data from multiple languages (especially those pertaining to the same family of languages) significantly improves the results of the sentiment classification.JRC.G.2 - Global security and crisis managemen
Comparative Experiments for Multilingual Sentiment Analysis using Machine Translation
No full textSentiment analysis is the Natural Language Processing task dealing
with sentiment detection and classification. In the past few years, there has been
a steady increase in the interest towards this task, for which different methods and
resources have been proposed. Sentiment analysis has been studied in the context
of traditional media, but also the new social media. Nevertheless, the research
community has concentrated less on developing methods for languages other than
English.Motivated by this fact, the present article deals with the problem of sentiment
detection in three different languages - French, German and Spanish - using
three distinct Machine Translation (MT) systems - Bing, Google and Moses, using
supervised methods with different combinations of features. Our extensive
evaluation scenarios show that SMT systems are approaching a good level of maturity
and can start to be employed to obtain training data for languages other than
English and that sentiment analysis systems can obtain comparable performances
to the one obtained for English.JRC.G.2 - Global security and crisis managemen
The Challenge of Processing Opinions in Online Contents in the Social Web Era
No full textIn the past years, the NLP community has been increasingly interested in the field of opinion mining (also known as sentiment analysis), whose aim is to retrieve and classify the opinions expressed in text. Online reputation management, as a related task, is more focused on opinions on individuals and other entities. Additionally, the computational task of online reputation management also considers the analysis of facts that influence the status quo of these entities. The problem in this context is much more difficult to solve, as entities, as opposed to products, are related to different events and topics and there is no fixed set of “attributes” that are commented on by persons expressing opinions on these entities. There is only one freely accessible system performing such as a task - Lydia (Skiena et al., 2007), which gathers news from portals and blogs and classifies opinions on different entities. However, both this system, as well as different approaches that have been presented for this problem in the research literature, show that the issue of entity-centered opinion mining and, additionally, the correlation of the results with facts over events where these entities are involved are not trivial (Balahur and Steinberger, 2009; Zhang and Skiena, 2010). The present position paper studies the challenges related to the field of online reputation management and suggests possible solutions.JRC.G.2 - Global security and crisis managemen
IBEREVAL OM: Minería de opiniones en los nuevos géneros textuales
No full textThe increasing amount of subjective data on the Web is creating the need to develop effective Question Answering systems able to discriminate such information from factual data, and subsequently process it with specific methods. The participants in the IBEREVAL OM tasks will be given a set of opinion questions (in Spanish and English). Optionally, they will also be able to receive the same set of opinion questions, in which the source, target and expected polarity, as well as the time span the question is referring to are given. They will also be provided with a collection of blog posts, extracted using the Technorati blog search engine (in Spanish and English), in which the answers to the opinion questions should be found The gold standard for this blog posts collection will previously be annotated using the EmotiBlog scheme, by a number of 3 annotators. The EmotiBlog corpus and the set of questions presented in (Balahur et al., 2009) – in their present state will be provided for system training. The participants will be able to participate in two subtasks : 1) in the first one, they will be asked to provide the list of answers to each of the questions (in the same language as the questions, or in the other language); 2) in the second one, they will be asked to provide a summary of the question answers – the top x% of the most important answers, in a non-redundant manner. The Gold Standard for the summaries will be automatically extracted from the manual annotations, taking into account the “intensity” parameter of the opinions expressed.Con el grande aumento de la información subjetiva en la Web, hay una importante necesidad de desarrollar sistemas de Question Answering que sean eficientes y capaces de discriminar entre datos objetivos y subjetivos. Los participantes tendrán una colección de preguntas de opinión (Español e Inglés) en las cuales se deberán encontrar las respuestas. El Gold Standard será anotado previamente con el esquema de anotación EmotiBlog por 3 anotadores. El corpus EmotiBlog y la colección de preguntas presentados en (Balahur et al. 2009) se pondrá a disposición para el entrenamiento del sistema. Los participantes deberán devolver un listado de respuestas para cada una de las preguntas, (en el mismo idioma que la pregunta o en otro), un resumen de las respuestas –de las x% de las respuestas más importantes, de una manera no redundante, el Gold Standard para los resúmenes será extraído automáticamente de las anotaciones manuales teniendo en consideración el parámetro de “intensidad” de la opinión expresada.This evaluation task proposal has been partially supported by Ministerio de Ciencia e Innovación - Spanish Government (grant no. TIN2009-13391-C04-01), and Conselleria d'Educació - Generalitat Valenciana (grant no. PROMETEO/2009/119 and ACOMP/2010/288
Sentiment Analysis in Spanish
No full textTesis doctoral elaborada por E. Martínez Cámara en la Universidad de Jaén bajo la dirección de los doctores D. L. Alfonso Ureña López y Da. M. Teresa Martín Valdivia. La defensa tuvo lugar el 26 de octubre de 2015 en Jaén ante el tribunal formado por la doctora Da. María Teresa Taboada Gómez de la Universidad Simon Fraser (Canadá) como presidenta, por el doctor D. José Manuel Perea Ortega de la Universidad de Extremadura (España) como secretario y por la doctora Da. Alexandra Balahur Dobrescu del Joint Research Centre (Italia) de la Comisión Europea como vocal. La tesis obtuvo la mención Internacional y logró una calificación de Sobresaliente Cum Laude.Ph.D. thesis written by Eugenio Martínez Cámara at the University of Jaén under the supervision of the Ph.D. L. Alfonso Ureña López and the Ph.D. M. Teresa Martín Valdivia. The author was examined on 26st October 2015 by a pannel composed by the Ph.D. María Teresa Taboada Gómez from the Simon Fraser University (Canada) as president of the pannel, the Ph.D. José Manuel Perea Ortega from the University of Extremadura (Spain) as secretary of the pannel and the Ph.D. Alexandra Balahur Dobrescu from the Joint Research Centre (Italy) of the European Comission as a panel member. The Ph.D. was awared Summa cum laude and it obtained the International mention.Este trabajo de investigación ha sido parcialmente financiado por el Fondo Europeo de Desarrollo Regional (FEDER), el proyecto FIRST FP7-287607 del Séptimo Programa Marco para el Desarrollo de la Investigación y la Tecnología de la Comisión Europea; el proyecto ATTOS TIN2012-38536-C03-0 del Ministerio de Economía y Competitividad y el proyecto AROESCU P11-TIC-7684 MO de Excelencia de la Junta de Andalucía
OPAL at SemEval Task 4: the Challenge of Porting a Sentiment Analysis System to the "Real" World
No full textSentiment analysis has become a well-established task in Natural Language Pro-cessing. As such, a high variety of methods have been proposed to tackle it, for different types of texts, text levels, languages, domains and formality levels. Although state-of-the-art systems have obtained promising results, a big challenge that still remains is to port the systems to the “real world” – i.e. to implement systems that are running around the clock, dealing with information of heterogeneous na-ture, from different domains, written in differ-ent styles and diverse in formality levels. The present paper describes our efforts to imple-ment such a system, using a variety of strate-gies to homogenize the input and comparing various approaches to tackle the task. Specifi-cally, we are tackling the task using two dif-ferent approaches: a) one that is unsu-pervised, based on dictionaries of sentiment-bearing words and heuristics to compute final polarity of the text considered; b) the second, supervised, trained on previously annotated data from different domains. For both ap-proaches, the data is first normalized and the slang is replaced with its expanded version.JRC.E.1 - Disaster Risk Managemen
WASSA 2012 - Proceedings of the 3rd Workshop in Computational Approaches to Subjectivity and Sentiment Analysis
No full textIn the past years, the quantity of contents generated by users on the Web, in social networking sites,
fora and microblogs has reached an unprecedented level. All this data adds on to the contents generated
in traditional media, such as newspapers, bringing additional factual, as well as a high quantity of
opinionated and subjective information. In the context of the society in which we live, where sifting
through the immense quantities of information to gather knowledge has become a must, the challenge
of processing opinionated and subjective information is becoming more and more a focus to the Natural
Language Processing (NLP) research communities worldwide.
In the past decade, the interest in proposing computational methods to deal with subjectivity and
sentiment in text has grown constantly from the NLP community. However, although the subjectivity
and sentiment analysis research fields have been highly dynamic in this period, much remains still to
be done, so that systems dealing with subjectivity, sentiment and, more generally, affect in text, can be
reliably used in critical decision-making environments. Moreover, the new means of communication
and user connection, in microblogs and social networks, become more and more relevant to these two
tasks, as the contexts (internal and external) of the information communication process bring about new
challenges and applications to be explored.
Inspired by the above-mentioned issues and the objectives we aimed at in the first two editions of
the Workshop on Computational Approaches to Subjectivity Analysis (WASSA 2010 and WASSA
2.011), the purpose of the third edition of the Workshop on Computational Approaches to Subjectivity
and Sentiment Analysis (WASSA 2012) was to create a framework for presenting and discussing the
challenges related to subjectivity and sentiment analysis in NLP and its applications, in traditional
and Social Media contexts, from an interdisciplinary theoretical and practical perspective. WASSA
2012 was organized in conjunction to the 50th Annual Meeting of the Association for Computational
Linguistics, on July 12, 2012, in Jeju, Korea.JRC.G.2 - Global security and crisis managemen
Extending the EmotiNet Knowledge Base to Improve the Automatic Detection of Implicitly Expressed Emotions from Text
No full textSentiment analysis is one of the recent, highly dynamic fields in Natural
Language Processing. Most existing approaches are based on word-level
analysis of texts and are mostly able to detect only explicit expressions of
sentiment. However, in many cases, emotions are not expressed by using
words with an affective meaning (e.g. happy), but by describing real-life
situations, which readers (based on their commonsense knowledge) detect
as being related to a specic emotion. Given the challenges of detecting
emotions from contexts in which no lexical clue is present, in this article we
present a comparative analysis between the performance of well-established
methods for emotion detection (supervised and lexical knowledge-based) and
a method we propose and extend, which is based on commonsense knowledge
stored in the EmotiNet knowledge base. Our extensive evaluations show
that, in the context of this task, the approach based on EmotiNet is the
most appropriate.JRC.G.2 - Global security and crisis managemen
- …
