AI Query

AI Query explores how AI (Artificial Intelligence) can be used to perform searches on the NSIP data. The basic idea is that the user enters the query to filter the data, the query gets translated into a filter expression and that expression is then used to filter NSIP data, the results then get displayed to the user.

At this point AI Query is considered experimental and is only supported for Katahdin animals.

Concepts

AI Query understands various concepts that a user can include in their query. Most of them would be familiar to users while there are a few new and experimental concepts that might be useful to query on when exploring the data.

Animal Id

Id
Sire Id
Dam Id

AI Query supports exact matches, starts with, ends with and contains operations.

Flock

Flock where the animal was born.

AI Query will decide if a flock filter is required based on how the user writes the query. The default is to assume that the user is querying within their own flock. This can be changed by asking to query in other specific flocks or even query all flocks.

Age & Date of Birth

Date of Birth
Age in Days
Age in Months
Age in Years
Year Born

The age is calculated as it is at the time of querying. Animals that do not have a date of birth will not match any queries that utilize date of birth or age.

EBVs - EBV Percentile - Accuracies

AI Query knows about all Katahdin EBVs, their percentiles (where available) as well as their accuracies.

Test Status

AI Query knows if certain traits / trait categories have been tested for an animal. The test status flags do not always map directly to specific EBVs. The following test statuses are known:

Birth Weight Test Status
Weaning Weight Test Status
Early Post-Weaning Weight Test Status
Post-Weaning Weight Test Status
Yearling Weight Test Status
Hogget Weight Test Status
Fecal Test Status: Either WFEC or PFEC was submitted
Scrotal Circumference Test Status
Ultrasound Carcass Scan Test Status

Linked Status

Animals must have linkage for specific trait categories in order to get across flock EBVs. AI Query has knowledge of linkage for the following trait categories:

Reproductive
Fecal Egg Count
Weight
Carcass

Miscellaneous

Pedigree Known: The amount of NSIP pedigree that is known (a number from 0 to 100)
Gender: Ram/Male and Ewe/Female
Status: Status value of the animal (Current, Dead, Cull, Sold, etc.)
Last Year Active: The last year that an animal was active. An animal is considered active in a year if it was born in the year, sired lambs in the year or lambed in the year.
Dam Age: The age that the dam of the animal was when the animal was born.
Flock Count: The number of flocks that an animal has progeny in.
Progeny Count: The number of progeny that an animal has.
Genotyped: Indicates if an animal was genotyped or not.

Using AI Query

Enter you query then click the Go button. The AI Message screen explains how AI Query interpreted your query. Data retrieval starts when the AI Message screen aligns itself to the right of the screen. The AI Message screen automatically closes itself, you can re-open it by clicking on the message button to the right of the Go button.

Writing Queries

Queries must resolve to yes or no, yes - the animal should be included in the result, no - the animal should not be included. AI Query only supports filtering.

Here's a few queries to get you going:

Example:
Rams born this year in any flock with index in the 80th percentile or better and with a mwwt larger than 1.5

Example:
Ewes in my flock that were last active in 2024

Example:
Rams that were last active in 2024 that:
From any flock
Has 10 or offspring
Has index percentile of at least 80
PFEC or WFEC percentile of at least 80
Was tested for fecal egg count
Is linked for fecal egg count