SSN_MLRG1 at SemEval-2017 Task 5: Fine-Grained Sentiment Analysis Using Multiple Kernel Gaussian Process Regression Model

Proceedings of the 11th International Workshop on Semantic Evaluations (SemEval-2017), pages 823–826, Vancouver, Canada, August 3 - 4, 2017. c⃝2017 Association for Computational Linguistics SSN MLRG1 at SemEval-2017 Task 5: Fine-Grained Sentiment Analysis Using Multiple Kernel Gaussian Process Regression Model Angel Deborah S, S Milton Rajendram, T T Mirnalinee SSN College of Engineering Kalavakkam 603 110, India angeldeboarahs@ssn.edu.in Abstract The system developed by the SSN MLRG1 team for Semeval-2017 task 5 on fine-grained sentiment analysis uses Multiple Kernel Gaussian Process for identifying the optimistic and pessimistic sentiments associated with companies and stocks. Since the comments on the same companies and stocks may display different emotions depending on time, their properities like smoothness and periodicity may vary. Our experiments show that while single Kernel Gaussian Process can learn some properties well, Multiple Kernel Gaussian Process are effective in learning the presence of different properties. 1 Introduction Sentiments have been widely studied as they play an important role in human intelligence, ratio- nal decision making, social interaction, percep- tion, memory, learning and creativity (Pang and Lee, 2008; Strapparava and Mihalcea, 2008; Maas et al., 2011; Li et al., 2015). The ability to dis- cern and understand human sentiments is criti- cal for making interactive human-like computer agents, and requires the use of machine learning approaches (Alm et al., 2005). 2 Gaussian Process Gaussian Process (GP) is a Bayesian non- parametric approach to machine learning. A Gaus- sian Process is a collection of random variables, any infinite number of which have a joint Gaus- sian distribution (Rasmussen and Williams, 2006). Using a Gaussian process, we can define a distri- bution over functions f(x), f(x) ∼ GP(m(x), k(x, x′)) (1) where m(x) is the mean function, usually defined to be zero, and k(x, x′) is the covariance function (or kernel function) that defines the prior prop- erties of the functions considered for inference. Gaussian Process has the following main advan- tages (Cohn and Specia, 2013; Cohn et al., 2014). • The kernel hyper-parameters can be learned via evidence maximization. • GP provides full probabilistic prediction, and an estimate of uncertainty in the prediction. • Compared to SVMs which need unbiased datasets for good performance, GPs do not usually suffer from this problem. • GP can be easily extended and incorporated into a hierarchical Bayesian model. • GP works really well when combined with kernel models. • GP works well for small datasets too. 2.1 Gaussian Process Regression The Gaussian Process regression framework as- sumes that, given an input x, output y is a noise corrupted version of a latent function evaluation. In a regression setting, we usually consider a Gaussian likelihood, which allows us to obtain a closed form solution for the test posterior (Ebden, 2008). Gaussian Process model, as they are ap- plied in machine learning, is an attractive way of doing non-parametric Bayesian modeling for a supervised learning problem. GP-based model- ing has the ability to learn hyper-parameters di- rectly from data by maximizing the marginal like- lihood. Like other kernel methods, …

Content truncated to 500 words