Belief Engine - Declarative is going towards the goal of a scientific quest on a Web of Evolving Declarative Beliefs for Human Being and Cognitive Robotics. It is a branch of Belief Engine, which is a Web of various declarative beliefs extracted and generated from various sources, mainly on the World Wide Web. It is designed to be used by various cognitive robotics, while the nature of Belief Engine enables it to be used in other knowledge processing and language understanding applications.
The current version of Belief Engine - Declarative is with two sources. One of them is declarative knowledge from our previous work CASIA-KB, a declarative semantic knowledge base built with extracted knowledge from Baidu Encyclopedia, Hudong Encyclopedia, and Wikipedia (with entity links among different knowledge sources). And the other is Conceptual Common Sense Declarative Belifes (conceptual common sense declarative knowledge with belief values) generated based on knowledge from CASIA-KB.
Please access the contents of "Belief Engine - Declarative" through this SPARQL ENDPOINT.
Our beliefs constitute a large part of our knowledge of the world. We have beliefs about objects, about culture, about the past, and about the future. We have beliefs about other people, and we believe that they have beliefs as well. We use beliefs to predict, to explain, to create, to console, to entertain. Many of our beliefs should be regarded as tentative and should be subject to revision (or elimination) based on rigorous evaluation -- just like scientific theories are.
-- Nils J. Nilsson. Understanding Beliefs, MIT Press, 2014.
Belief Engine shares the vision introduced above, and is aiming at creating a Web of beliefs for human being and cognitive robotics.
The declarative beliefs are designed to be represented in triples following the Resource Description Framework (RDF) and Ontology Web Language (OWL) standards. The current online beliefs are conceptual common sense knowledge with a belief value.
Conceptual Common Sense Beliefs (CCSB) are generated through conceptualization on subjects and objects of triples form knowledge extracted from various sources (e.g. Baidu Encyclopedia, Hudong Encyclopedia, and Wikipedia). Namely, each belief in the current Belief Engine is composed of two concepts (i.e. subject and object), one binary relation, and a belief value (ranging from 0 to 1). The belief values are based on statistical analysis of conceptual common sense knowledge generated through entity conceptualization. Instance level beliefs will also be released soon.