Sequential Recommendation Patents: Who Leads, Where the Gaps Are 2026
Sequential and Session-Based Recommendation Patents
Concentrated at the top. The five most active filers account for 75.0% of all 44 records in scope, with a long tail below them. Filings were negligible before 2021. From 2021 to 2024 — the last year that can be treated as complete — recorded activity grew from 4 to 12, a +200% increase, with 2024 standing as the peak year so far. 2025 and 2026 show fewer records, but that reflects the roughly 18-month gap between filing and publication, not a slowdown in the field.
- 1SALESFORCE INC10
- 2VISA INTERNATIONAL SERVICE ASSOCIATION7
- 3HOME DEPOT INTERNATIONAL INC7
- 4SAMSUNG ELECTRONICS CO LTD7
- 5ADOBE INC2
See the full sequential and session based recommendation analysis in Eureka
- The complete ranking, not just the top five
- Every IPC branch with its share of the corpus
- The most-cited records, and where claim space is still thin
Frequently asked questions
Who holds the most patents in sequential and session-based recommendation?
The ranked leader holds 10 of the 44 records in scope, and the top 5 assignees together account for 75.0% of all records. Filing activity drops sharply after that group, with a long tail of companies and universities holding only one or two records each. The full ranking covers 16 companies, so this is a concentrated field at the top with wide dispersion below it.
Is patent filing in sequential recommendation still growing?
Recorded filings grew from 4 in 2021 to 12 in 2024, a +200% increase over that span, with 2024 the peak year in the dataset so far. The lower counts visible in 2025 and 2026 reflect the roughly 18-month lag between filing and publication rather than a real decline. Any assessment of momentum should treat 2024 as the most recent complete data point.
What does US20230252345A1 actually cover?
US20230252345A1, assigned to Salesforce.com, Inc. and published 2023-08-10, describes training a sequential recommendation model by encoding user behavior sequences into interest representations, predicting a next item as a probability distribution, and using the next actual interacted item as a positive sample against a negative sample drawn from that distribution to compute a contrastive loss. It is a training-method claim rather than a specific application claim, so its practical reach depends on how narrowly the negative-sampling and contrastive-loss steps are drafted. Anyone building a similar training pipeline should review its claim language closely rather than assume it only covers retail or media use cases.
Disclaimer. This analysis is based on Patsnap Eureka data drawn from a limited snapshot of global patent records and is provided for general information and reference only. Patent data carries inherent limitations — recent filings are under-counted because of publication lag, counts may be on a record or family basis, classification and applicant-name data may contain errors or duplicates, and the underlying search query defines the scope shown — so the analysis may be incomplete or inaccurate and may not reflect the full technology landscape.
Nothing here is an exhaustive prior-art, novelty, freedom-to-operate or validity search, nor does it constitute legal, financial or professional advice, and it should not be relied upon as such. Verify independently and review with qualified patent and legal professionals before acting on it.
Method: Filing trend and technology composition. Derived from a Patsnap search on Sequential and Session Based Recommendation covering 2015–2026, data cut-off 2026-07-31. Counts reflect published records only and shift as new filings publish. Every share divides by all records in scope. Data: Patsnap Eureka. See the full landscape report.