What is the impact on the overall quality of the knowledge graph when entity disambiguation fails?

When entity disambiguation fails, the overall quality of the knowledge graph decreases significantly, mainly reflected in three aspects: data accuracy, relationship credibility, and downstream application effectiveness. Data layer: Incorrectly matched or unrecognized entities lead to redundant, conflicting, or missing entity nodes in the knowledge graph. For example, mistakenly associating "Apple (company)" with "Apple (fruit)" directly pollutes the basic data. Relationship layer: Failed entity disambiguation disrupts the semantic connections between entities, distorting the relationship network of the knowledge graph. For instance, incorrectly associating "Li Bai" with irrelevant attributes other than "Tang Dynasty poet" weakens the structured expression ability of the knowledge graph. Application layer: Downstream systems based on an erroneous knowledge graph (such as intelligent Q&A and recommendation engines) will output misleading results, reducing user trust. It is suggested to improve matching accuracy by optimizing entity disambiguation algorithms (such as combining contextual semantic analysis and cross-domain entity feature comparison) to maintain the reliability of the knowledge graph.