# US 2024/0202798 A1 · Solving sparse data problems in a recommendation system with freezing start

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

- Number: US 2024/0202798 A1
- Document ID: US20240202798A1
- Status: Published application
- Kind: A1
- Application number: 18/083,373
- Filed: 2022-12-16
- Published: 2024-06-20
- Earliest claimed: 2022-12-16
- Assignees: Sap SE
- Applicants: Sap SE
- Inventors: Nick Pendar; Seyed Farhad Faghihi
- Web page: https://patentgrep.com/patent/US20240202798A1

## CPC

- G06Q 30/0631: {Recommending goods or services}
- G06Q 30/0202: Market predictions or forecasting for commercial activities
- G06Q: INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR

## Abstract

Recommendation services typically struggle with sparse data scenarios. A freezing generated item start technique can use a matrix of external generated items to find a linking generated item. Embeddings can be used to determine distance between items. The technologies are useful for providing recommendations even in scenarios involving little or no transaction data.

## Claims

### Claim 1

A computer-implemented method comprising:
building a generated-item-by-generated-item matrix representing correlations between external generated items based on observed transactions for the external generated items;
embedding representations of the external generated items into a warm embedded vector space;
for a plurality of freezing generated items, embedding representations of the plurality of freezing embedded items into a freezing embedded vector space;
for a freezing generated item out of the freezing generated items, determining a location of the freezing generated item in the warm embedded vector space;
finding a nearest neighbor external generated item r of the freezing generated item within a complementary threshold distance of the freezing generated item in the warm embedded vector space;
for a given linking external generated item out of a plurality of possible linking external generated items appearing in the matrix for r, determining whether a linked freezing nearest neighbor generated item exists for the given linking external generated item in the freezing embedded vector space;
responsive to determining that a linked freezing nearest neighbor generated item exists, associating the freezing generated item with the linked freezing nearest neighbor generated item for recommendation; and
outputting a recommendation based on the association.

### Claim 2 (depends on 1)

The method of claim 1, further comprising:
receiving a request for a recommendation for the freezing generated item; and
responsive to the request for the recommendation and based on association between the freezing generated item and the linked freezing nearest neighbor generated item, responding with the recommendation, wherein the recommendation comprises the linked freezing nearest neighbor generated item.

### Claim 3 (depends on 1)

The method of claim 1, further comprising:
filtering the linked freezing nearest neighbor item with a category tree.

### Claim 4 (depends on 1)

The method of claim 1, further comprising:
finding at least one additional linking generated item out of the plurality of possible linking external generated items; and
associating at least one additional linked freezing nearest neighbor item with the freezing generated item.

### Claim 5 (depends on 1)

The method of claim 1 wherein:
the freezing embedded vector space and the warm embedded vector space are a same vector space.

### Claim 6 (depends on 5)

The method of claim 5 wherein:
the freezing embedded vector space and the warm embedded vector space are represented by different indexes;
finding a nearest neighbor in the warm embedded vector space comprises searching a first index; and
determining whether a linked freezing nearest neighbor exists comprises searching a second index different from the first index.

### Claim 7 (depends on 1)

The method of claim 1 wherein:
determining whether a linked freezing nearest neighbor generated item exists incorporates a threshold.

### Claim 8 (depends on 1)

The method of claim 1 wherein:
determining whether a linked freezing nearest neighbor generated item exists comprises searching for an approximate nearest neighbor.

### Claim 9 (depends on 1)

The method of claim 1, further comprising:
pruning the generated-item-by-generated-item matrix;
wherein:
the pruning comprises removing any non-warm generated items from the generated-item-by-generated-item matrix; and
the pruned matrix represents warm generated items.

### Claim 10 (depends on 1)

The method of claim 1, further comprising:
normalizing the generated-item-by-generated-item matrix based on observed transactions per represented generated item.

### Claim 11 (depends on 1)

The method of claim 1, further comprising:
representing a number of observed co-occurrence transactions for a generated item in a diagonal of the generated-item-by-generated-item matrix.

### Claim 12 (depends on 1)

The method of claim 1, wherein:
embedding representations of the external generated items into the warm embedded vector space comprises:
for a given item out of the external generated items, calculating vector values of text describing the given item.

### Claim 13 (depends on 12)

The method of claim 12, wherein:
the text describing the given item comprises a title of the given item.

### Claim 14 (depends on 12)

The method of claim 12, wherein:
calculating vector values of text describing the given item comprises incorporating context of words in the text.

### Claim 15

A system comprising:
a generated-item-by-generated-item matrix stored in one or more non-transitory computer-readable media and representing observed co-occurrences within transactions of external generated items represented in the matrix;
an internal representation of a high-dimensional embedded vector space in which a freezing generated item is represented along with a plurality of warm generated items out of the external generated items; and
a recommendation extension service configured to identify a warm approximate nearest neighbor of the freezing generated item in the high-dimensional embedded vector space, search candidate linking items found in matrix entries of the warm approximate nearest neighbor for linked freezing items within a threshold distance of the candidate linking items in the high-dimensional embedded vector space, and responsive to finding a linked freezing generated item, associating the freezing generated item with the linked freezing generated item.

### Claim 16 (depends on 15)

The system of claim 15 further comprising:
a recommendation service configured to receive a request for a generated item recommendation for an input generated item, wherein the recommendation service provides a recommendation comprising the linked freezing generated item based on an association between the freezing generated item and the linked freezing generated item.

### Claim 17 (depends on 15)

The system of claim 15 wherein:
the external generated items are represented in the high-dimensional vector space based on respective textual descriptions of the external generated items.

### Claim 18 (depends on 15)

The system of claim 15 further comprising:
an internal representation of a hierarchical category tree;
wherein the linked freezing generated item is filtered with fallback category filtering.

### Claim 19 (depends on 15)

The system of claim 15 wherein:
the matrix is pruned to remove items having less than a threshold number of co-occurrences; and
the matrix is normalized.

### Claim 20

One or more non-transitory computer-readable media having stored therein computer-executable instructions that when executed by a computing system, cause the computing system to perform a method of outputting a generated item recommendation from a plurality of possible generated items, the method comprising:
building a generated-item-by-generated-item matrix representing correlations between external generated items based on observed transactions for the external generated items;
pruning the generated-item-by-generated-item matrix;
normalizing the generated-item-by-generated-item matrix;
embedding representations of the external generated items into a warm embedded high-dimensional vector space based on titles of the external generated items;
for a plurality of freezing generated items, embedding the plurality of freezing generated items into a freezing embedded high-dimensional vector space based on titles of the freezing generated items;
for a freezing generated item out of the freezing generated items, determining a location of the freezing generated item in the warm embedded high-dimensional vector space;
finding a nearest neighbor external generated item r of the freezing generated item within a complementary threshold distance of the freezing generated item in the warm embedded high-dimensional vector space;
for a given linking generated item out of a plurality of possible linking generated items appearing in rows and columns for r within the matrix, determining whether a linked freezing nearest neighbor generated item exists for the given linking generated item in the freezing embedded high-dimensional vector space; and
responsive to determining that a linked freezing nearest neighbor generated item exists, associating the freezing generated item with the linked freezing nearest neighbor generated item for recommendation.
