As the adoption of sentiment analysis continues to spread across industries, from politics to PR, opinions about the field also run deep. That's especially true among practitioners, and a range of academic and vendor specialists weighed in at the Sentiment Analysis Symposium in New York last week. While the novelty factor begins to subside, clients are looking for more substance, and as befitting such a multifaceted topic, it's complicated. As a follow-up to yesterday's post that covered the analysis of visual images and facial coding, here the experts offered their perspectives on approaching 5 ongoing issues: The degree of accuracy issue is hard to answer, said Bing Liu, a University of Chicago computer science professor specializing in data mining. It depends on what you're measuring, the level of text you're analyzing, the number of data sets across domains and the voice sound quality of videos, among other variables.

Sentiment analysis is a powerful example of how machine learning can help developers build better products with unique features. In short, sentiment analysis is the automated process of understanding if text written in a natural language (English, Spanish, etc.) is positive, neutral, or negative about a given subject. Nowadays, we have many instances where people express opinions and sentiment: tweets, comments, reviews, articles, chats, emails and more. One popular example is Twitter, where real-time opinions from millions of users are expressed constantly. Companies use sentiment analysis on Twitter to discover insights about their products and services.

At first glance, it looks like Fox and Breitbart began to tank compared to the competition. But Corcoran says not to jump to conclusions. "I wouldn't put this decline down to the content of the sites themselves," the analyst told HuffPost. "In the case of Breitbart, their likes and comments actually increased between June 2015 and March 2016. In the case of Fox, those numbers include all local affiliates to the Fox network, so it's a broad coalition of sites, many of which may be seeing challenges in the way that their content is engaged with in the News Feed."

Our motive is to create a simple to integrate "Machine Learning" platform but yet powerful enough to provide high accuracy and low latency API. Such a system provides Data Mining, Machine Learning and Artificial Intelligence algorithms as a service. The system has ability to create training model for datasets uploaded as a training set and performs classification on similar datasets in the future using the saved models. "Sentiment analysis (also known as opinion mining) refers to the use of natural language processing, text analysis and computational linguistics to identify and extract subjective information in source materials." Download the sample "sentiment analysis" file Sentiment Analysis The first column should always be the label to be predicted.

This blog was originally published on our Text Analysis blog, the blog post set out to analyze and visualize 11 million tweets collected around the time of and during Apple Live 2014. Apple Live probably got off to the worst start possible earlier this year. Most of us who tried to log on to watch the much-anticipated launch were first, forced to watch the live feed in Safari and second, greeted with the TV Truck Schedule Screen... To add to this Apple also made a complete mess of the audio. We were left sitting refreshing the page, waiting for the stream to start while being subjected to an audio visual nightmare, described brilliantly by this "fan" below: To simulate the #applelive experience, open up several separate YouTube vids, play them simultaneously, minimize, stare at a test pattern. At AYLIEN, we gathered 11 million tweets mentioning'Apple', 'iPhone', 'iOS', 'iPad', 'Mac', 'iPod', 'Macbook', 'iCloud', 'OS X', 'iWatch' and '#AppleLive' from the 4th of September to the 10th of September with a view of analyzing the tweets to gain insight into the voice of Apple Followers.

The key to making a business case for any Analytics initiative, not just text analytics, is to identify specific business problems and pain points and use analytics to address them, instead of merely seeking insights. Companies find themselves in a world where an increasing number of their customers are using social media, and the one thing, people LOVE doing on social media is talk (tweet/post/blog/whatever...) They talk about their experiences in dealing with the company and its services or products, about its competitors and about how they really feel. So, as a company, you have all this customer feedback out there, in the form of text, just waiting to be gathered. The risk company management faces, for not capturing this customer feedback, is just too great to ignore. They face the risk of looking bad (think PR nightmare) and losing their competitive advantage, if they do nothing about it, which brings me to the single most important use case driving text analytics in the enterprise today, which is, the compelling need for Social Media engagement and analytics.

Event Extraction (EE) is a challenging Information Extraction task which aims to discover event triggers with specific types and their arguments. Most recent research on Event Extraction relies on pattern-based or feature-based approaches, trained on annotated corpora, to recognize combinations of event triggers, arguments, and other contextual information. These combinations may each appear in a variety of linguistic forms. Not all of these event expressions will have appeared in the training data, thus adversely affecting EE performance. In this paper, we demonstrate the overall effectiveness of Dependency Regularization techniques to generalize the patterns extracted from the training data to boost EE performance. We present experimental results on the ACE 2005 corpus, showing improvement over the baseline system, and consider the impact of the individual regularization rules.

We explore the relationship between negated text and negative sentiment in the task of sentiment classification. We propose a novel adjustment factor based on negation occurrences as a proxy for negative sentiment that can be applied to lexicon-based classifiers equipped with a negation detection pre-processing step. We performed an experiment on a multi-domain customer reviews dataset obtaining accuracy improvements over a baseline, and we further improved our results using out-of-domain data to calibrate the adjustment factor. We see future work possibilities in exploring negation detection refinements, and expanding the experiment to a broader spectrum of opinionated discourse, beyond that of customer reviews.

Embeddings are generic representations that are useful for many NLP tasks. In this paper, we introduce DENSIFIER, a method that learns an orthogonal transformation of the embedding space that focuses the information relevant for a task in an ultradense subspace of a dimensionality that is smaller by a factor of 100 than the original space. We show that ultradense embeddings generated by DENSIFIER reach state of the art on a lexicon creation task in which words are annotated with three types of lexical information - sentiment, concreteness and frequency. On the SemEval2015 10B sentiment analysis task we show that no information is lost when the ultradense subspace is used, but training is an order of magnitude more efficient due to the compactness of the ultradense space.

Tweet sentiment classification seeks to identify the emotional polarity of a tweet. One potential way to enhance classification performance is to include emoticons as features. Emoticons are representations of faces expressing various emotions in text. They are created through combinations of letters, punctuation marks and symbols, and are frequently found within tweets. While emoticons have been used as features for sentiment classification, the importance of their inclusion has not been directly measured. In this work, we seek to determine if the addition of emoticon features improves classifier performance. We also investigate how high dimensionality impacts the addition of emoticon features. We conducted experiments testing the impact of using emoticon features, both with and without feature selection. Classifiers are trained using four different learners and either emoticons, unigrams, or both as features. Feature selection was conducted using five filter based feature rankers with four feature subset sizes. Our results showed that the choice of feature set (emoticon, unigram or both) had no significant impact in our initial tests when using no feature selection; however, with any of the tested feature selection techniques, augmenting unigram features with emoticon features resulted in significantly better performance than unigrams alone. Additionally, we investigate how the addition of emoticons changes the top features selected by the rankers.