Frequently Asked Questions about Lexicon¶
Alation Cloud Service Applies to Alation Cloud Service instances of Alation
Customer Managed Applies to customer-managed instances of Alation
Can the Lexicon job be run for one data source or schema?¶
No. The Lexicon will process metadata for all data objects in the catalog.
Can all the titles be confirmed at once?¶
All titles on the table or column pages cannot be confirmed at once. They can only be confirmed or discarded one by one.
Can I see a list of all abbreviations for an expansion?¶
No, currently you can only see a list of all expansions for each specific abbreviation.
Will auto-titling ever override titles confirmed by a user or provided by a user?¶
No. If a user added a title, it will always take precedence over the title computed by ALLIE. Guesses will only be filled in for blank titles. If a user confirmed a guess on an object then that guess will be treated as a human-given title.
How does auto-titling happen?¶
ALLIE attempts to best guess the title from abbreviations included in the name of a data object. It will adhere to the following conditions:
It will not override human-given or human-confirmed titles.
It will only provide a guess if it has some confidence about all parts of the name
If you reject expansions for abbreviations on the Lexicon page, it will not use those matches anymore when generating auto-titles
Why didn’t Alation guess the title of an object?¶
There are two main reasons for not guessing the title of an object.
The guess has not yet been computed and the Lexicon job needs to be run.
Alation is not confident in guessing the title of an object. The reasons are as follows:
The term that you have included has numbers or symbols. Alation does not guess well if the term has non-alphabetical characters.
Multiple expansions existing for one abbreviation may lead to insufficient confidence in a guess, and Lexicon will not be able to determine which expansion should be used. In such a case it will leave the object without an auto-title.
If the term has several parts, and if one of the parts in the term is uncertain, the guess will not be shown. Sometimes it is not obvious which of the two words a name can be expanded into, and this lowers the confidence score of the abbreviation-expansion match.
Will ALLIE/Lexicon learn from any confirmed titles of schemas, tables, columns from the catalog page?¶
In versions up to V R6 (5.10.x), no. The learning only happens directly on the Lexicon page (AFTER running the Lexicon job). When a user confirms or rejects a title on the catalog pages of schemas, tables, or columns, this confirmation/rejection only applies to this specific title found on this specific page. The confirmation or rejection is not passed on to the Lexicon dictionary. Starting with V R7 (5.12.x), the feedback users give to ALLIE on the catalog pages of data objects by confirming and rejecting the auto-suggested titles will now be applied system-wide when Lexicon is recalculated, helping to improve subsequent guesses.
How can I teach Alation to make better guesses?¶
Adding or confirming expansions for abbreviations on the Lexicon page will help ALLIE make better guesses. You can find this page only if you are an admin user (Admin Settings > Lexicon). When you make changes, they will not immediately impact guesses. The guesses are recomputed automatically once per week or you can run the Lexicon job manually on demand to apply the new changes.
What is in the initial Lexicon on a fresh Alation installation?¶
Nothing. The set of abbreviations in the Lexicon is based on data object names found in your specific Alation instance. You must first perform metadata extraction for data sources in your Alation instance for any abbreviations to appear on the Lexicon page.
How Do I Set Up Lexicon?¶
See: