CauseNet is a large-scale, open-domain causality graph that compiles over 11 million claimed causal relations between causal concepts extracted from various semi-structured and unstructured web sources with an estimated precision of 83%. The project aims to create a comprehensive causal knowledge base separate from mere causal beliefs to advance artificial intelligence and enable large-scale causal inference research. CauseNet comprises causal concepts connected by causal relations, each with detailed provenance data indicating where and how the relation was extracted. Sources include ClueWeb12 sentences, Wikipedia sentences, Wikipedia lists, and Wikipedia infoboxes. Examples demonstrate how causal relations link a cause and an effect concept, along with rich metadata such as sentence context, page references, timestamps, and extraction patterns. The dataset is available in three versions: - CauseNet-Full (11.6 million relations, 12.1 million concepts, 1.8 GB) - CauseNet-Precision (higher precision subset with 199,806 relations, 80,223 concepts, 135 MB) - CauseNet-Sample (small sample dataset with 264 relations, 524 concepts, 54 KB) Sample code is provided to load CauseNet into the Neo4j graph database. The developers also provide training and evaluation datasets for causal concept spotting, supporting identification of multi-word causal concepts from text. CauseNet is the basis for the CIKM 2020 paper titled "CauseNet: Towards a Causality Graph Extracted from the Web," and is available under open licenses (MIT for code, Creative Commons Attribution 4.0 for data). Contact information for the development team at Paderborn University, Technical University of Munich, and Leipzig University is provided for inquiries. In summary, CauseNet offers a valuable resource for causal reasoning, computational argumentation, multi-hop question answering, and AI research dependent on causal knowledge.