Citation Context

... to classify the sentiment polarity of a tweet as positive, ∗ This work was done when the first and third authors were visiting Microsoft Research Asia. negative or neutral. The majority of existing approaches follow Pang et al. (2002) and employ machine learning algorithms to build classifiers from tweets with manually annotated sentiment polarity. Under this direction, most studies focus on designing effective features to obtain better classification performance. For example, Mohammad et al. (2013) build the top-performed system in the Twitter sentiment classification track of SemEval 2013 (Nakov et al., 2013), using diverse sentiment lexicons and a variety of hand-crafted features. Feature engineering is important but laborintensive. It is therefore desirable to discover explanatory factors from the data and make the learning algorithms less dependent on extensive feature engineering (Bengio, 2013). For the task of sentiment classification, an effective feature learning method is to compose the representation of a sentence (or document) from the representations of the words or phrases it contains (Socher et al., 2013b; Yessenalina and Cardie, 2011). Accordingly, it is a crucial step to learn the w...

Citation Context

... evaluation datasets for Twitter sentiment analysis have been made publicly available. The general evaluation dataset consists of a set of tweets, where each tweet is annotated with a sentiment label =-=[1,8,16,22]-=-. The most common sentiment labels are positive, negative and neutral, but some evaluation datasets consider additional sentiment labels such as mixed, other or irrelevant [1,23]. Instead of the final...

Citation Context

...nts. 4 Experiments We present a qualitative and quantitative analysis of the GB-RNN on a contextual sentiment classification task. The main dataset is provided by the SemEval 2013, Task 2 competition =-=[17]-=-. We outperform the winners of the 2013 challenge, as well as several baseline and model ablations. 4.1 Evaluation Dataset The SemEval competition dataset is composed of tweets labeled for 3 different...

Citation Context

...Dataset (SSTD) [23] 2289 1037 1252 6849 The Dialogue Earth Weather Dataset (WAB) [3] 5495 2580 2915 7485 The Dialogue Earth Gas Prices Dataset (GASP) [3] 6285 5235 1050 8128 Semeval Dataset (Semeval) =-=[14]-=- 7535 2186 5349 15851 4.2 Entity-Level Evaluation Setup In the second validation test, we evaluate the usefulness of SS-Patterns as features for entity-level sentiment analysis, i.e., detecting sentim...

Citation Context

...1) or big and proprietary such as in (Lin and Kolcz, 2012). Others rely on noisy labels obtained from emoticons and hashtags (Go et al., 2009). The SemEval-2014 task 9: Sentiment Analysis in Twitter (=-=Nakov et al., 2013-=-) provides a public dataset to be used to compare the accuracy of different approaches. In this paper, we propose to analyse tweet sentiment with the use of Adaptive Boosting (Freund and Schapire, 199...

Citation Context

...nctuation repetitions are language-independent emotion tokens. A different way of evaluating multilingual SA systems is posed by Balahur et al. (2014). They translate the English SemEval 2013 corpus (=-=Nakov et al., 2013-=-) into Spanish, Italian, French and German by means of machine translation (MT) systems. The resulting datasets were revised by non-native and native speakers independently, finding that the use of ma...

Citation Context

...cy and Emotional Contagion could be helpful for sentiment analysis. 3 Data and Resources 3.1 Labeled Data We used the data set provided in SemEval 2013 for subtask B of sentiment analysis in Twitter (=-=Nakov et al., 2013-=-). The participants have been provided with training tweets annotated positive, negative or neutral. We downloaded these tweets using the given script. We obtained 9646 tweets, the whole training data...

Citation Context

... relatively recent research topic that has attracted great interest, as reflected by a series of shared evaluation tasks, e.g., analysis of news headlines (Strapparava and Mihalcea, 2007) and tweets (=-=Nakov et al., 2013-=-). Relevant applications deal with numerous domains such as blogs (Balog et al., 2006), news stories (Lloyd et al., 2005), and product reviews (Hu and Liu, 2004). In (Turney and Littman, 2002), the af...

Citation Context

...support kernels, and is suitable for classification on large datasets with a large number of features. This is particularly useful for text classification, where the number of features is very large, which means that the data is likely to be linearly separable, and thus using kernels is not really necessary. We scaled the SVM input and we used L2-regularization during training. 6https://github.com/percyliang/ brown-cluster 592 3 Experiments, Results, Analysis 3.1 Experimental setup At development time, we trained on train-2013, tuned the C value of SVM on dev-2013, and evaluated on test-2013 (Nakov et al., 2013). For our submission, we trained on train-2013+dev-2013, and we evaluated on the 2014 test dataset provided by the organizers. This dataset contains two parts and a total of five datasets: (a) progress test (the Twitter and SMS test datasets for 2013), and (b) new test datasets (from Twitter, from Twitter with sarcasm, and from LiveJournal). We used C=0.012, which was best on development. 3.2 Official results Due to our very late entering in the competition, we have only managed to perform a small number of experiments, and we only participated in subtask B. We were ranked 20th out of 50 submi...

Citation Context

...nce evaluation. Two main strategies have been used to predict elections: buzz or number of tweets mentioning a given candidate or party and secondly the use of sentiment analysis. Different computational approaches have been explored to process sentiment in text, namely 2015 IEEE International Conference on Computer and Information Technology; Ubiquitous Computing and Communications; Dependable, Autonomic and Secure Computing; Pervasive Intelligence and Computing 978-1-5090-0154-5/15 $31.00 © 2015 IEEE DOI 10.1109/CIT/IUCC/DASC/PICOM.2015.228 1521 machine learning and linguistic based methods [21, 22, 23]. In practice, algorithms often combine both strategies. One particular challenge when using sentiment is how to aggregate opinions in a timely fashion that can be fed to the prediction method. The majority of the work using social media data applied to political scenarios consist in ad hoc studies where researchers collect data from a given social network during a pre-defined period of time and produce their specific analysis or predictions. The availability of open source research platforms in this area is scarse. Usually researchers use specific APIs and software modules to produce their st...