Matching configuration items with machine learning

A system includes persistent storage containing configuration items (CIs) representing discovered attribute values of computing resources associated with a managed network, and an application configured to perform operations, including obtaining test result data generated based on a third-party scanning system executing tests of a particular computing resource associated with the managed network. The test result data includes attribute values of the particular computing resource. The operations also include generating, by way of an embedding model and based on the attribute values, an embedding vector representing the attribute values, and comparing the embedding vector to a plurality of candidate embedding vectors, each representing the discovered attribute values of a corresponding CI of the CIs. The operations further include, based on comparing the embedding vector to the plurality of candidate embedding vectors, selecting a particular CI of the CIs, and updating the particular CI to additionally represent the test result data.