Draft:SenticNet

  • Comment: We can't accept AI submissions. ChrysGalley (talk) 10:01, 11 July 2026 (UTC)

SenticNet is an open-source knowledge base and neurosymbolic artificial intelligence framework used for concept-level sentiment analysis and natural language understanding. It is developed to support text analysis by shifting from word-frequency statistics to a representation model that matches the semantic and emotional meaning of multi-word expressions.[1]

The project is developed by a research group at the College of Computing and Data Science (CCDS) within Nanyang Technological University (NTU) in Singapore, along with external open-source contributors.[2]

History

SenticNet began in 2009 at the MIT Media Lab as part of a Cooperative Awards in Science and Engineering (CASE) project involving the MIT Media Lab, the University of Stirling, and Sitekit Solutions Ltd.[2] The resource has since undergone multiple updates to support applications in social data analytics, human–computer interaction, financial modeling, and healthcare.

Architecture and methodology

SenticNet is built automatically via graph-mining, multidimensional scaling, and representation learning techniques rather than relying on manual annotation. It extracts affective and commonsense data from foundational resources such as WordNet-Affect, Open Mind Common Sense (OMCS), and GECKA.[3]

Knowledge representation

Based on Marvin Minsky's panalogy principle—which suggests representing data in parallel formats to increase structural robustness—SenticNet organizes its knowledge base across three primary layouts:

  1. Semantic Network: A directed graph that maps contextual associations and relational links between concepts.
  2. Matrix Representation: Adjacency matrices that track the topological layout of the graph.
  3. Vector Space (AffectiveSpace): A high-dimensional vector space where multi-word concepts are positioned based on semantic similarity to allow geometric calculation and reasoning.[4]

Sentiment inference mechanisms

The extraction of semantic meaning and emotional metrics relies on several core components:

  • Spreading Activation: An algorithm that passes emotional valences along the nodes of the semantic graph to assign meaning to unlabelled multi-word concepts.
  • Sentic Neurons: Deep learning structures designed to model linguistic dependencies.
  • The Hourglass of Emotions: An emotion categorization model that maps emotional variables into four independent dimensions: Pleasantness, Attention, Sensitivity, and Aptitude.[3]

Version history

SenticNet has grown through sequential public releases, expanding its vocabulary size and transitioning its primary computational methods over time:

Version Year Primary Technical Features Concept Capacity
SenticNet 1 2010 Linear polarity mapping based on ConceptNet structures. ~6,000 concepts
SenticNet 2 2012 Integration of semantics with structured affective labels. ~13,000 concepts
SenticNet 3 2014 Application of energy-flow methods across semantic networks.[1] ~30,000 concepts
SenticNet 4 2016 Structural integration of conceptual semantic primitives. ~50,000 concepts
SenticNet 5 2018 Primitive inference optimized through recurrent neural networks (RNNs). ~100,000 concepts
SenticNet 6 2020 Ensemble models combining sub-symbolic transformers with symbolic logic. ~200,000 concepts
SenticNet 7 2022 Introduction of a graph-based neurosymbolic AI framework. ~300,000 concepts
SenticNet 8 2024 Integration of emotion AI models with enterprise commonsense graphs.[5] ~400,000 concepts
SenticNet 9 2026 Automated conceptual primitive discovery and time-shift mechanisms for generative emotion AI.[6] >400,000 concepts

Distributions and extensions

SenticNet is available in standard Semantic Web formats, including RDF/XML and OWL formats as formal web ontologies.

  • AffectiveSpace: A 100-dimensional vector space embedding representation of the underlying affective commonsense knowledge graph, used for similarity calculations.
  • PrimeNet: A connected subset of the knowledge base designed to map strict hierarchical hyponym-hypernym structural relationships.
  • BabelSenticNet: Localized variants of the knowledge base providing API support and semantic tools across 80 target languages.[2]

See also

References

  1. ^ a b Cambria, Erik; Olsher, Daniel; Rajagopal, Dhaval (2014). "SenticNet 3: A Common and Common-Sense Knowledge Base for Cognition-Driven Sentiment Analysis". Proceedings of the AAAI Conference on Artificial Intelligence. 28 (1).
  2. ^ a b c "SenticNet Project Home". SenticNet. Retrieved 11 July 2026.
  3. ^ a b Cambria, Erik; Wang, Haiwei; Mao, Rui (2018). "A Localization Toolkit for SenticNet". Proceedings of the IEEE International Conference on Data Mining.
  4. ^ Poria, Soujanya; Cambria, Erik; Bajpai, Devamanyu; Hussain, Amir (2017). "A Review of Affective Computing: From Unimodal Analysis to Multimodal Fusion". Information Fusion. 37: 98–125.
  5. ^ Cambria, Erik; Liu, Quanzhi; Decherchi, Sergio; Xing, Frank; Kwok, Kenneth (2024). "SenticNet 8: A Commonsense-based Neurosymbolic AI Framework for Explainable Sentiment Analysis". IEEE Transactions on Affective Computing.
  6. ^ Cambria, Erik; Mao, Rui; Zhang, Xulang; Xiao, Lin; Shen, Ting; Anand, Ashish (2026). "SenticNet 9: Generative Commonsense for Emotion AI via Conceptual Primitive Discovery and Time Shift Mechanism". IEEE Transactions on Computational Social Systems. 13.

Content Disclaimer

Informasi ini disarikan dari Wikipedia dan disajikan kembali untuk tujuan edukasi. Konten tersedia di bawah lisensi CC BY-SA 3.0. Kami tidak bertanggung jawab atas ketidakakuratan data yang bersumber dari kontribusi publik tersebut.

  1. The information displayed on this website is sourced in part or in whole from Wikipedia and has been adapted for the purpose of restating it. We strive to provide accurate and relevant information, however:
  2. There is no guarantee of absolute accuracy. Wikipedia is an open, collaborative project that can be edited by anyone, so information is subject to change.
  3. It is not intended to constitute professional advice. The content displayed is for informational and educational purposes only. For important decisions (e.g., medical, legal, or financial), please consult a professional.
  4. Content copyright. Wikipedia is licensed under the Creative Commons Attribution-ShareAlike License (CC BY-SA). This means that content may be reused with appropriate attribution and shared under a similar license.
  5. Responsible use. Any risk arising from the use of information from this website is entirely the responsibility of the user.