SenticNet

Abstract

The main focus of this chapter lies in investigating and analysing SenticNet. The methodology by which the SenticNet sentiment lexicon was compiled has been the subject of several academic papers published by the researcher who developed SenticNet. However, little technical documentation of what is “under the hood” of SenticNet is publicly available. With this in mind and the fact that the research question posed in this paper uses SenticNet as the platform on which the research is performed, an in-depth investigation and analysis of SenticNet was carried out. The investigation consisted of a review of SenticNet-related academic papers and of an analysis of SenticNet using descriptive statistics. The goal of the investigation was to evaluate both how SenticNet contributes to concept-based sentiment analysis in terms the origins of the sentiment information contained therein, and how SenticNet could be integrated into the research. The chapter consists of seven main sections. Section 3.1 describes the sources of the knowledge in SenticNet, with particular focus on the novel aspect of the knowledge extracted from these sources. Section 3.2 describes a seminal example of sentiment analysis research. Section 3.3 presents an overview of the techniques and methods used for producing SenticNet. Section 3.4 briefly describes the core processes involved in producing SenticNet. Section 3.5 describes how RDF/XML is used to encode knowledge in SenticNet. Section 3.6 describes how this knowledge can be accessed and retrieved. Section 3.7 presents a descriptive statistics analysis of SenticNet.