| 摘要 |
An assessment of a request for change is made by first receiving the request for change. The request for change is compared with a plurality of stored requests for change that are stored in a change database. Then, a computation is made, using learned similarity functions, of a similarity value for each of the stored requests for change with respect to the received request for change. Feedback is received with respect to the similarity values. A training set of data is updated based on the received feedback. The updated training set of data is applied to a learning process, to obtain learned similarity functions. The similarity calculator is updated with the learned similarity functions. The similarity functions can be used to evaluate the risk of changes by looking at the impact of similar past changes. |
