What is knowledge fusion

That is, merge two knowledge graphs (ontology/entity). The basic problem is how to fuse description information about the same entity or concept from multiple sources. Need to confirm: - equivalence instance - equivalence class/subclass - equivalence attribute/subattributeCopy the code

An example is shown in the figure above, where different colored circles represent different knowledge graph sources, where Rome in the dbpedia.org source and Roma in the Geoname.org source are the same entity, linked through two sameAs. Entity alignment between different knowledge graphs is the main work of KG fusion. In addition to entity alignment, there are conceptual level knowledge fusion, cross-language knowledge fusion and other work. It is worth mentioning here that knowledge fusion has different names in different literatures, such as ontology alignment, ontology matching, Record Linkage, Entity Resolution, Entity alignment, etc., but their essential work is the same. The main technical challenges of knowledge fusion are as follows: data quality challenges, such as fuzzy naming, data input errors, data loss, data format inconsistency, abbreviations, etc. And how to identify the two entities are the same entity, only the same entity can be fused. Data scale challenges: large amount of data (parallel computing), variety of data types, no longer just name matching, multiple relationships, more links, etc.Copy the code

Normalization, correlation, fusion

Normalization: Indicates whether the entities to be included are the same entity. Literally, many belong to one. If there are several entities to be included are the same entity, then these entities to be included are attributed to one entity. Association: Whether the entity to be included is the same entity as the entity in the core set. Fusion preference: it is the preference of entity attributes.Copy the code