One of the core tasks in Opinion Mining consists of estimating the polarity of the opinionated documents found. In some scenarios (e.g. blogs), this estimation is severely affected by sentences that are off-topic or that simply do not express any opinion. In fact, the key sentiments in a blog post often appear in specific locations of the text. In this paper we propose several effective and robust polarity detection methods based on different sentence features. We show that we can successfully determine the polarity of documents guided by a sentence-level analysis that takes into account topicality and the location in the blog post of the subjective sentences. Our experimental results show that some of our proposed variants are both highly effective and computationally-lightweight.

No abstract

Polarity estimation in large-scale and multi-topic domains is a difficult issue. Most state-of-the-art solutions essentially rely on frequencies of sentiment-carrying words (e.g., taken from a lexicon) when analyzing the sentiment conveyed by natural language text. These approaches ignore the structural aspects of a document, which contain valuable information. Rhetorical Structure Theory (RST) provides important information about the relative importance of the different text spans in a document. This knowledge could be useful for sentiment analysis and polarity classification. However, RST has only been studied for polarity classification problems in constrained and small scale scenarios. The main objective of this paper is to explore the usefulness of RST in largescale polarity ranking of blog posts. We apply sentence-level methods to select the key sentences that convey the overall on-topic sentiment of a blog post. Then, we apply RST analysis to these core sentences in order to guide the classification of their polarity and thus to generate an overall estimation of the document's polarity with respect to a specific topic. Our results show that RST provides valuable information about the discourse structure of the texts that can be used to make a more accurate ranking of documents in terms of their estimated sentiment in multi-topic blogs.

Sentiment Analysis tools often rely on counts of sentiment-carrying words, ignoring structural aspects of content. Natural Language Processing has been fruitfully exploited in Text Mining, but advanced discourse processing is still non pervasive for mining opinions. Some studies, however, extracted opinions based on the discursive role of text segments. The merits of such computationally intensive analyses have thus far been assessed in very specific, small-scale scenarios. In this paper, we investigate the usefulness of Rhetorical Structure Theory in various Sentiment Analysis tasks on different types of information sources. First, we demonstrate how to perform a large-scale ranking of individual blog posts in terms of their overall polarity, by exploiting the rhetorical structure of a few key evaluative sentences. In order to further validate our findings, we additionally explore the potential of Rhetorical Structure Theory in sentence-level polarity classification of news and product reviews. Our most valuable polarity classification features turn out to capture the way in which polar terms are used, rather than the sentimentcarrying words per se.

In the blogosphere, different actors express their opinions about multiple topics. Users, companies or editors socially interact by commenting, recommending and linking blogs and posts. These social media contents are increasingly growing. As a matter of fact, the size of the blogosphere is estimated to double every six months. In this context, the problem of finding a topically relevant blog to subscribe to becomes a Big Data challenge. Moreover, combining multiple types of evidence is essential for this search task. In this paper we propose a group of textual and social-based signals, and apply different Information Fusion algorithms for a Blog Distillation Search task.Information fusion through the combination of the different types of evidence requires optimisation for appropriately weighting each source of evidence. To this end, we analyse well-established population-based search methods. Namely, global search (Particle Swarm Optimisation and Differential Evolution) and a local search method (Line Search) that has been effective in various Information Retrieval tasks. Moreover, we propose hybrid combinations between the global search and the local search method and compare all the alternatives following a standard methodology. Efficiency is an imperative here and, therefore, we focus not only on achieving high search effectiveness but also on designing efficient solutions.