SenticNet 4: A Semantic Resource for Sentiment Analysis Based on Conceptual Primitives

Proceedings of COLING 2016, the 26th International Conference on Computational Linguistics: Technical Papers, pages 2666–2677, Osaka, Japan, December 11-17 2016. SenticNet 4: A Semantic Resource for Sentiment Analysis Based on Conceptual Primitives Erik Cambria, Soujanya Poria, Rajiv Bajpai Nanyang Technological University 50 Nanyang Ave, Singapore 639798 {cambria,sporia rbajpai}@ntu.edu.sg Bj¨orn Schuller Imperial College London 180 Queens’ Gate, Huxley Bldg., London SW7 2AZ, UK bjoern.schuller@imperial.ac.uk Abstract An important difference between traditional AI systems and human intelligence is the human ability to harness commonsense knowledge gleaned from a lifetime of learning and experience to make informed decisions. This allows humans to adapt easily to novel situations where AI fails catastrophically due to a lack of situation-specific rules and generalization capabilities. Com- monsense knowledge also provides background information that enables humans to successfully operate in social situations where such knowledge is typically assumed. Since commonsense consists of information that humans take for granted, gathering it is an extremely difficult task. Previous versions of SenticNet were focused on collecting this kind of knowledge for sentiment analysis but they were heavily limited by their inability to generalize. SenticNet 4 overcomes such limitations by leveraging on conceptual primitives automatically generated by means of hierarchical clustering and dimensionality reduction. 1 Introduction The opportunity to capture the opinion of the general public has raised growing interest within both the scientific community as well as the business world, due to the remarkable benefits to be had from marketing and financial prediction, which has led to many exciting open challenges (Pang and Lee, 2008; Liu, 2012). Mining opinions and sentiments from natural language, however, is an extremely difficult task as it requires a deep understanding of most of the explicit and implicit, regular and irregular, syntactic and semantic rules of a language. Existing approaches to sentiment analysis mainly rely on parts of text in which opinions are explicitly expressed such as polarity terms, affect words, and their co-occurrence frequencies. However, opinions and sentiments are often conveyed implicitly through latent semantics, which make purely syntactic approaches ineffective. SenticNet (Cambria et al., 2014) captures such latent information in terms of semantics and sentics, i.e., the denotative and connotative information commonly associated with real-world objects, actions, events, and people. SenticNet steps away from blindly using keywords and word co-occurrence counts, and instead relies on the implicit meaning associated with commonsense concepts. Superior to purely syntactic techniques, SenticNet can detect subtly expressed sentiments by enabling the analysis of mul- tiword expressions that do not explicitly convey emotion, but are instead related to concepts that do so. The main limitation of SenticNet is that it is unable to generalize instances of concepts, e.g., eat pasta or slurp noodles: unless there is an exact match, SenticNet 3 raises a not-found error. In SenticNet 4, however, both verb and noun concepts are linked to primitives so that, for example, concepts such as eat pasta or slurp noodles are generalized as INGEST FOOD. In this way, most concept inflections can be captured by the knowledge base: verb concepts like …

Content truncated to 500 words