Recommending machine learning models and source codes for input datasets

Patent metadata

Field	Value
Publication number	US-11501191-B2
Application number	US-201816138566-A
Country	US
Kind code	B2
Filing date	Sep 21, 2018
Priority date	Sep 21, 2018
Publication date	Nov 15, 2022
Grant date	Nov 15, 2022

Abstract

Asset recommendation for a particular input dataset is provided. Candidate data analysis assets having a corresponding relatedness score associated with the particular input dataset greater than a defined relatedness score threshold value are selected. Those candidate data analysis assets having a corresponding relatedness score greater than the defined relatedness score threshold value are ranked by score. Those candidate data analysis assets having a corresponding relatedness score greater than the defined relatedness score threshold value are listed by rank from highest to lowest. A justification for each candidate data analysis asset is inserted in the ranked list of candidate data analysis assets. The ranked list of candidate data analysis assets along with each respective justification is outputted on a display device.

First claim

What is claimed is:

1. A computer-implemented method for asset recommendation for a particular input dataset, the computer-implemented method comprising:
selecting, by a computer, candidate machine learning models and source codes having a corresponding relatedness score to the particular input dataset greater than a defined relatedness score threshold value;
ranking, by the computer, those selected candidate machine learning models and source codes having a corresponding relatedness score greater than the defined relatedness score threshold value by score to form a ranked list of selected candidate machine learning models and source codes;
selecting, by the computer, one of a highest-ranking candidate machine learning model or a highest-ranking candidate source code in the ranked list of selected candidate machine learning models and source codes as a highest-ranking data analysis asset to classify the particular input dataset:
applying, by the computer, the highest-ranking data analysis asset to the particular input dataset to classify the particular input dataset for performing a task using the particular input dataset:
performing, by the computer, an analysis of the particular input dataset to determine semantics corresponding to the particular input dataset;
providing, by the computer, recommendations for the selected candidate machine learning models and source codes based on the semantics corresponding to the particular input dataset; and
training, by one of the candidate source codes, one of the candidate machine learning models using the particular input dataset.

2. The computer-implemented method of claim 1 further comprising:
calculating, by the computer, a relatedness score between the particular input dataset and the selected candidate machine learning models and source codes based on semantics corresponding to the particular input dataset and semantics corresponding to each candidate machine learning model and source code of the selected candidate machine learning models and source codes.

3. The computer-implemented method of claim 2 , wherein the relatedness score is a measure of strength of relationship between a particular candidate machine learning model and the particular input dataset and a particular candidate source code and the particular input dataset.

4. The computer-implemented method of claim 2 , wherein the semantics corresponding to the particular input dataset include at least one of structure and content of the particular input dataset, types of users who have used the particular input dataset previously, types of problems that users were trying to solve using the particular input dataset, types of data pattern analysis algorithms, data transformations, and source codes previously applied to the particular input dataset, and machine learning models previously trained using the particular input dataset.

5. The computer-implemented method of claim 2 , wherein the semantics corresponding to each candidate machine learning model and source code include domain, structure, and content of data used to train each particular candidate machine learning model and source code.

6. The computer-implemented method of claim 1 further comprising:
utilizing, by the computer, a user profile of a user to assist in providing the recommendations for the selected candidate machine learning models and source codes, wherein the user profile comprises a user type of the user and historical usage patterns that represent a pattern of dataset usage by the user.

7. The computer-implemented method of claim 1 further comprising:
determining, by the computer, whether a user input was received to open an existing dataset stored on the computer; and
responsive to the computer determining that a user input was received to open an existing dataset stored on the computer, retrieving, by the computer, a pre-computed set of semantics corresponding to the existing dataset from a storage device of the computer.

8. The computer-implemented method of claim 1 further comprising:
determining, by the computer, whether a user input was received to upload a new dataset to the computer;
responsive to the computer determining that a user input was received to upload a new dataset to the computer, computing, by the computer, a set of semantics corresponding to the new dataset; and
storing, by the computer, the new dataset and the set of semantics corresponding to the new dataset in a storage device of the computer.

9. The computer-implemented method of claim 1 further comprising:
displaying, by the computer, the recommendations for the candidate machine learning models and source codes to form displayed recommendations.

10. The computer-implemented method of claim 9 further comprising:
receiving, by the computer, a user selection of a particular machine learning model and source code in the displayed recommendations.

11. The computer-implemented method of claim 1 , wherein the computer is a data engineering platform located in a cloud environment.

12. A computer system for asset recommendation for a particular input dataset, the computer system comprising:
a bus system;
a storage device connected to the bus system, wherein the storage device stores program instructions; and
a processor connected to the bus system, wherein the processor executes the program instructions to:
select candidate machine learning models and source codes having a corresponding relatedness score to the particular input dataset greater than a defined relatedness score threshold value;
rank those selected candidate machine learning models and source codes having a corresponding relatedness score greater than the defined relatedness score threshold value by score to form a ranked list of selected candidate machine learning models and source codes;
select one of a highest-ranking candidate machine learning model or a highest- ranking candidate source code in the ranked list of selected candidate machine learning models and source codes as a highest-ranking data analysis asset to classify the particular input dataset;
apply the highest-ranking data analysis asset to the particular input dataset to classify the particular input dataset for performing a task using the particular input dataset;
perform an analysis of the particular input dataset to determine semantics corresponding to the particular input dataset;
providing recommendations for the selected candidate machine learning models and source codes based on the semantics corresponding to the particular input dataset; and
train, by one of the candidate source codes, one of the candidate machine learning models using the particular input dataset.

13. The computer system of claim 12 , wherein the processor further executes the program instructions to:
calculate a relatedness score between the particular input dataset and the selected candidate machine learning models and source codes based on semantics corresponding to the particular input dataset and semantics corresponding to each candidate machine learning model and source code of the plurality of selected candidate machine learning models and source codes.

14. A computer program product for asset recommendation for a particular input dataset, the computer program product comprising a computer readable storage medium having program instructions embodied therewith, the program instructions executable by a computer to cause the computer to perform a method comprising:
selecting, by the computer, candidate machine learning models and source codes having a corresponding relatedness score to the particular input dataset greater than a defined relatedn

Assignees

• IBM

Inventors

• Shaikh Samiulla
• Mehta Sameep
• Bhide Manish
• Lobig William B

Classifications

• G06N5/045 Primary

  Explanation of inference; Explainable artificial intelligence [XAI]; Interpretable artificial intelligence · CPC title

• G06N20/20 Primary

  Ensemble learning · CPC title

• G06F18/24

  Classification techniques · CPC title

• G06Q30/0201

  Market modelling; Market analysis; Collecting market data · CPC title

• G06F16/211

  Schema design and management · CPC title