Model Execution

The input execute dataset must include all the descriptions from external sources we want to map into CCH tagetik descriptions, with the logic of concat and level fields we want to use for the mapping.

During this phase each row in the dataset is converted into an embedding vector, which is a numerical representation in a vector space.

Then, a measure of vector distance is computed between each possible pair of Train end Execute embeddings. Those distances define the matching rank between each input Execute row and all the input Train rows (the rank is given by the ascending order of the median level distances and concat distances, when median level distances are the same).