# US 2026/0310735 A1 · Recommending data sources based on new activities of a distributed system

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2026/0310735 A1
- Document ID: US20260310735A1
- Status: Published application
- Kind: A1
- Application number: 19/098,860
- Filed: 2025-04-02
- Published: 2026-10-08
- Earliest claimed: 2025-04-02
- Assignees: Dell Products L.P.
- Applicants: Dell Products L.P.
- Inventors: ZIJIA WANG; XIAOCHEN CHEN; ASHUTOSH P. NANEKAR; NING LI
- Web page: https://patentgrep.com/patent/US20260310735A1

## CPC

- G06F 16/953: Querying, e.g. by the use of web search engines
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

Methods and systems for desired computer implemented services using data stored in data sources of a distributed system are disclosed. The desired computer implemented services may be provided by recommending data sources based on new activities of a user and/or the data sources. The recommended data sources may be provided by measuring first levels of relevancy and second levels of uncertainty between the new activities and/or the data sources. The recommended data sources may be used with a request template to generate an information request that is processable by information extraction frameworks. From processing the information request, information may be obtained that is used to facilitate a provisioning of the desired computer implemented services.

## Claims

### Claim 1

A method for providing computer implemented services using data stored in data sources of a distributed system, the method performed by a data processing configured as a management system that manages the data sources of the distributed system and comprising:
obtaining a data request from a user for a first computer implemented services of the computer implemented services, the data request being serviceable using a portion of the data stored in data sources;
based on the data request, performing a real-time data source analysis process to obtain recommended data sources from which the portion of the data should be obtained by at least:
performing, by the management system, an adaptive learning process to detect in real time all new activities that have occurred within the distributed system since at least the obtaining of the data request, the new activities comprising at least a data change in characteristic of the data stored in the data sources;
based on the new activities detected in the real time:
computing, for each of the data sources, a level of uncertainty of relevancy of the data sources with respect to the data request;
computing, for each of the data sources, an undiscounted level of the relevancy of the data sources with respect to the data request;
computing, for each of the data sources, a discounted level of the relevancy of the data sources with respect to the data request using the undiscounted level and the level of uncertainty for each of the data sources;
obtaining, using the discounted level of the relevancy for each of the data sources, recommended data sources to service the data request;
generating, using the recommended data sources, at least one information request by replacing one or more placeholders in a request template with data source identifiers of the recommend data sources;
obtaining, using the at least one information request, the portion of the data; and
using the portion of the data to provide the first computer implemented services, wherein obtaining the portion of the data from the recommend data sources increases a likelihood of the first computer implemented services being provided to the user as compared to obtaining the portion of the data directly from the data sources without using the recommend data sources.

### Claim 2 (depends on 1)

The method of claim 1, wherein obtaining the portion of the data comprises initiating servicing, by information extraction frameworks that manage access to the data stored in the data sources, the information request to obtain the portion of the data from the data sources.

### Claim 3 (depends on 2)

The method of claim 2, wherein the data request is incompatible with the information extraction frameworks for obtaining of the portion of the data.

### Claim 4 (depends on 3)

The method of claim 3, wherein the information request is a query usable to obtain the portion of the data, the information extraction frameworks comprising query execution engines compatible with the query.

### Claim 5 (depends on 1)

The method of claim 1, wherein the level of uncertainty is computed using, at least in part, a graph data structure based on the data sources, the graph data structure comprising:
nodes corresponding to each of the data sources; and
edges between the nodes corresponding to relationships between the data sources.

### Claim 6 (depends on 5)

The method of claim 5, wherein for one of the data sources, the level of uncertainty is based on embeddings for the one of the data sources.

### Claim 7 (depends on 6)

The method of claim 6, wherein computing the undiscounted level of the relevancy of the data sources comprises updating the embeddings associated with the nodes by propagating information between a first node of the nodes and a second node of the nodes based on an importance level between the first node and the second node.

### Claim 8 (depends on 6)

The method of claim 6, wherein the level of the uncertainty of the relevancy for the one of the data sources is based on historical changes to the embeddings associated with the new activities occurring within the distributed system, an embedding of the embeddings being based at least on a first activity among the new activities detected in the real time with respect to the one of the data sources, and the embedding being updated based on changes in the first activity.

### Claim 9 (depends on 8)

The method of claim 8, further comprising:
before computing the level of the uncertainty of the relevancy:
identifying the first activity with respect to the data sources that is not taken into account by the embeddings; and
performing an update to at least one embedding of the embeddings based on the first activity to obtain updated embeddings.

### Claim 10 (depends on 9)

The method of claim 9, wherein the first activity comprises at least one selected from a group consisting of:
a new interaction of the user with respect to the data sources; and
the data change in characteristic of at least the portion of the data stored in the data sources.

### Claim 11 (depends on 9)

The method of claim 9, wherein for the one of the data sources, computing the level of the uncertainty of the relevancy of the data sources comprises:
generating, using a probability distribution and the updated embeddings, a probabilistic vector, the probability distribution comprising a type of variability expected of the at least one embedding of the embeddings and the probabilistic vector comprising at least one fitted parameter of the probability distribution that is fitted to the at least one embedding;
computing, using the probabilistic vector, an entropy of the probability distribution; and
obtaining, using the entropy of the probability distribution, the level of the uncertainty of the relevancy of the one of the data sources.

### Claim 12 (depends on 11)

The method of claim 11, wherein computing the discounted level of the relevancy of the data sources comprises performing at least a mathematical operation using at least the undiscounted level of the relevancy and the level of the uncertainty of the relevancy to generate the discounted level of the relevancy, the mathematical operation at least in part discounting the undiscounted level of the relevancy based on the level of the uncertainty of the relevancy.

### Claim 13

A non-transitory machine-readable medium having instructions stored therein, which when executed by a processor of a management system that manages data sources of a distributed system, cause the processor to perform operations for providing computer implemented services using data stored in the data sources, the operations comprising:
obtaining a data request from a user for a first computer implemented services of the computer implemented services, the data request being serviceable using a portion of the data stored in data sources;
based on the data request, performing a real-time data source analysis process to obtain recommended data sources from which the portion of the data should be obtained by at least:
performing, by the management system, an adaptive learning process to detect in real time all new activities that have occurred within the distributed system since at least the obtaining of the data request, the new activities comprising at least a data change in characteristic of the data stored in the data sources;
based on the new activities detected in the real time:
computing, for each of the data sources, a level of uncertainty of relevancy of the data sources with respect to the data request;
computing, for each of the data sources, an undiscounted level of the relevancy of the data sources with respect to the data request;
computing, for each of the data sources, a discounted level of the relevancy of the data sources with respect to the data request using the undiscounted level and the level of uncertainty for each of the data sources;
obtaining, using the discounted level of the relevancy for each of the data sources, recommended data sources to service the data request;
generating, using the recommended data sources, at least one information request by replacing one or more placeholders in a request template with data source identifiers of the recommend data sources;
obtaining, using the at least one information request, the portion of the data; and
using the portion of the data to provide the first computer implemented services, wherein obtaining the portion of the data from the recommend data sources increases a likelihood of the first computer implemented services being provided to the user as compared to obtaining the portion of the data directly from the data sources without using the recommend data sources.

### Claim 14 (depends on 13)

The non-transitory machine-readable medium of claim 13, wherein obtaining the portion of the data comprises initiating servicing, by information extraction frameworks that manage access to the data stored in the data sources, the information request to obtain the portion of the data from the data sources.

### Claim 15 (depends on 14)

The non-transitory machine-readable medium of claim 14, wherein the data request is incompatible with the information extraction frameworks for obtaining of the portion of the data.

### Claim 16 (depends on 15)

The non-transitory machine-readable medium of claim 15, wherein the information request is a query usable to obtain the portion of the data, the information extraction frameworks comprising query execution engines compatible with the query.

### Claim 17

A data processing system, comprising:
a processor; and
a memory coupled to the processor to store instructions, which when executed by the processor, cause the processor to perform operations for providing computer implemented services using data stored in data sources of a distributed system that includes the data processing system, the operations comprising:
obtaining a data request from a user for a first computer implemented services of the computer implemented services, the data request being serviceable using a portion of the data stored in data sources;
based on the data request, performing a real-time data source analysis process to obtain recommended data sources from which the portion of the data should be obtained by at least:
performing, by the data processing system, an adaptive learning process to detect in real time all new activities that have occurred within the distributed system since at least the obtaining of the data request, the new activities comprising at least a data change in characteristic of the data stored in the data sources;
based on the new activities detected in the real time:
computing, for each of the data sources, a level of uncertainty of relevancy of the data sources with respect to the data request;
computing, for each of the data sources, an undiscounted level of the relevancy of the data sources with respect to the data request;
computing, for each of the data sources, a discounted level of the relevancy of the data sources with respect to the data request using the undiscounted level and the level of uncertainty for each of the data sources;
obtaining, using the discounted level of the relevancy for each of the data sources, recommended data sources to service the data request;
generating, using the recommended data sources, at least one information request by replacing one or more placeholders in a request template with data source identifiers of the recommend data sources;
obtaining, using the at least one information request, the portion of the data; and
using the portion of the data to provide the first computer implemented services, wherein obtaining the portion of the data from the recommend data sources increases a likelihood of the first computer implemented services being provided to the user as compared to obtaining the portion of the data directly from the data sources without using the recommend data sources.

### Claim 18 (depends on 17)

The data processing system of claim 17, wherein obtaining the portion of the data comprises initiating servicing, by information extraction frameworks that manage access to the data stored in the data sources, the information request to obtain the portion of the data from the data sources.

### Claim 19

(canceled)

### Claim 20

(canceled)

### Claim 21 (depends on 1)

The method of claim 1, wherein obtaining the recommended data sources comprises:
performing a rank-ordering of the data sources based on the discounted levels of the relevancy of the data sources; and
filtering the rank-ordering of the data sources using a threshold discounted level of relevancy and a highest n number of the data sources to obtain the recommend data sources.

### Claim 22 (depends on 21)

The method of claim 21, wherein the undiscounted level of the relevancy of the data sources is computed using at least a scalar matrix generated by a trained machine learning model, the scalar matrix being input by the trained machine learning model into an activation function to generate the undiscounted level of the relevancy of the data sources.
