Book chapter
Mining for Opinions Across Domains: A Cross-Language Study
In book
The Semantic Evaluation (SemEval) series of workshops focuses on the evaluation and comparison of systems that can analyse diverse semantic phenomena in text with the aim of extending the current state of the art in semantic analysis and creating high quality annotated datasets in a range of increasingly challenging problems in natural language semantics. SemEval provides an exciting forum for researchers to propose challenging research problems in semantics and to build systems/techniques to address such research problems. SemEval-2016 is the tenth workshop in the series of International Workshops on Semantic Evaluation Exercises. The first three workshops, SensEval-1 (1998), SensEval-2 (2001), and SensEval-3 (2004), focused on word sense disambiguation, each time growing in the number of languages offered, in the number of tasks, and also in the number of participating teams. In 2007, the workshop was renamed to SemEval, and the subsequent SemEval workshops evolved to include semantic analysis tasks beyond word sense disambiguation. In 2012, SemEval turned into a yearly event. It currently runs every year, but on a two-year cycle, i.e., the tasks for SemEval-2016 were proposed in 2015. SemEval-2016 was co-located with the 2016 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL-HLT’2016) in San Diego, California. It included the following 14 shared tasks organized in five tracks: • Text Similarity and Question Answering Track – Task 1: Semantic Textual Similarity: A Unified Framework for Semantic Processing and Evaluation – Task 2: Interpretable Semantic Textual Similarity – Task 3: Community Question Answering • Sentiment Analysis Track – Task 4: Sentiment Analysis in Twitter – Task 5: Aspect-Based Sentiment Analysis – Task 6: Detecting Stance in Tweets – Task 7: Determining Sentiment Intensity of English and Arabic Phrases • Semantic Parsing Track – Task 8: Meaning Representation Parsing – Task 9: Chinese Semantic Dependency Parsing • Semantic Analysis Track – Task 10: Detecting Minimal Semantic Units and their Meanings – Task 11: Complex Word Identification – Task 12: Clinical TempEval iii • Semantic Taxonomy Track – Task 13: TExEval-2 – Taxonomy Extraction – Task 14: Semantic Taxonomy Enrichment This volume contains both Task Description papers that describe each of the above tasks and System Description papers that describe the systems that participated in the above tasks. A total of 14 task description papers and 198 system description papers are included in this volume. We are grateful to all task organisers as well as the large number of participants whose enthusiastic participation has made SemEval once again a successful event. We are thankful to the task organisers who also served as area chairs, and to task organisers and participants who reviewed paper submissions. These proceedings have greatly benefited from their detailed and thoughtful feedback. We also thank the NAACL 2016 conference organizers for their support. Finally, we most gratefully acknowledge the support of our sponsor, the ACL Special Interest Group on the Lexicon (SIGLEX). The SemEval-2016 organizers, Steven Bethard, Daniel Cer, Marine Carpuat, David Jurgens, Preslav Nakov and Torsten Zesch
In many areas, such as social science, politics or market research, people need to track sentiment and their changes over time. For sentiment analysis in this field it is more important to correctly estimate proportions of each sentiment expressed in the set of documents (quantification task) than to accurately estimate sentiment of a particular document (classification). Basically, our study was aimed to analyze the effectiveness of two iterative quantification techniques and to compare their effectiveness with baseline methods. All the techniques are evaluated using a set of synthesized data and the SemEval-2016 Task4 dataset. We made the quantification methods from this paper available as a Python open source library. The results of comparison and possible limitations of the quantification techniques are discussed.
The relevance of this study is due to the need to identify the degree of statistical interdependence of text messages, which are available on the Internet, and the financial results of companies. The main purpose of the work is to identify features of the text messages that can help to differentiate between successful organizations and organizations, which are in crisis.
In this paper, we consider opinion word extraction, one of the key problems in sentiment analysis. Sentiment analysis (or opinion mining) is an important research area within computational linguistics. Opinion words, which form an opinion lexicon, describe the attitude of the author towards certain opinion targets, i.e., entities and their attributes on which opinions have been expressed. Hence, the availability of a representative opinion lexicon can facilitate the extraction of opinions from texts. For this reason, opinion word mining is one of the key issues in sentiment analysis. We designed and implemented several methods for extracting opinion words. We evaluated these approaches by testing how well the resulting opinion lexicons help improve the accuracy of methods for determining the polarity of the reviews if the extracted opinion words are used as features. We used several machine learning methods: SVM, Logistic Regression, Naive Bayes, and KNN. By using the extracted opinion words as features we were able to improve over the baselines in some cases. Our experiments showed that, although opinion words are useful for polarity detection, they are not su fficient on their own and should be used only in combination with other features.
The relevance of this study is due to the need to identify the degree of statistical interdependence of text messages, which are available on the Internet, and the financial results of companies. The main purpose of the work is to identify features of the text messages that can help to differentiate between successful organizations and organizations, which are in crisis.
We formulate the problem of text integrity assessment as learning thediscourse structure of text given the dataset of texts with high integrity and lowintegrity. We use two approaches to formalizing the discourse structures, sentimentprofile and rhetoric structures, relying on sentence-level sentiment classifierand rhetoric structure parsers respectively. To learn discourse structures, weuse the graph-based nearest neighbor approach which allows for explicit featureengineering, and also SVM tree kernel–based learning. Both learning approachesoperate on the graphs (parse thickets) which are sets of parse trees with nodeswith either additional labels for sentiments, or additional arcs for rhetoric relationsbetween different sentences. Evaluation in the domain of valid vs invalidcustomer complains (those with argumentation flow, non-cohesive, indicating abad mood of a complainant) shows the stronger contribution of rhetoric structureinformation in comparison with the sentiment profile information. Both abovelearning approaches demonstrated that discourse structure as obtained by RSTparser is sufficient to conduct the text integrity assessment. At the same time,sentiment profile-based approach shows much weaker results and also does not complement strongly the rhetoric structure ones.
The 4th International Conference on Educational Data Mining (EDM 2011) brings together researchers from computer science, education, psychology, psychometrics, and statistics to analyze large datasets to answer educational research questions. The conference, held in Eindhoven, The Netherlands, July 6-9, 2011, follows the three previous editions (Pittsburgh 2010, Cordoba 2009 and Montreal 2008), and a series of workshops within the AAAI, AIED, EC-TEL, ICALT, ITS, and UM conferences. The increase of e-learning resources such as interactive learning environments, learning management systems, intelligent tutoring systems, and hypermedia systems, as well as the establishment of state databases of student test scores, has created large repositories of data that can be explored to understand how students learn. The EDM conference focuses on data mining techniques for using these data to address important educational questions.