Candidate Retrieval Patents: Who Leads, Where the Gaps Are 2026
Candidate Retrieval and Dual Encoder Model Patents
42.7% concentration. The top 5 assignees hold 70 of 164 records in scope — filing here is already concentrated at the top, not evenly spread. Filings climbed from zero in 2017 to a peak of 35 in 2022, then declined to 15 by 2024 — a 40% drop over that three-year span. Because publication typically lags filing by around 18 months, the 2025-2026 counts shown are still incomplete and should not be read as a continued decline.
- 1TATA CONSULTANCY SERVICES LTD24
- 2GOOGLE LLC15
- 3WALMART APOLLO LLC14
- 4DIGITAL REASONING SYSTEMS INC9
- 5MICROSOFT TECHNOLOGY LICENSING LLC8
See the full candidate retrieval and dual encoder models 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
Common questions about this landscape
Who holds the most patents in candidate retrieval and dual encoder models?
One assignee leads with 24 of the 164 records in this dataset, well ahead of the fifth-ranked company at 8 records and tenth place at just 4. The top 5 assignees combined hold 42.7% of all records, and the top 10 hold 59.8%, which marks this as a field led by a small group rather than evenly spread across many filers. Beyond that group, 53 companies appear in the ranking overall, most contributing only a handful of records each.
Is patent filing in this space growing or slowing down?
Filing activity rose sharply through the late 2010s, peaked at 35 records in 2022, and had fallen to 15 by 2024 — a 40% decline across the last three complete filing years. However, publication typically lags filing by around 18 months, so the 2025 and 2026 figures in any trend chart are still incomplete and should not be read as evidence of continued decline. The honest read is that the field cooled from its 2022 high, not that it is disappearing.
What patent classes cover candidate generation and dual encoder technology?
The two most common IPC subclasses are G06N (AI-based computing models), appearing on 60.4% of the 164 records, and G06F (electric digital data processing), on 55.5%. G06Q (business and commerce data processing) appears on 35.4% of records, reflecting the commercial recommendation use case. Speech-related (G10L) and video-related (H04N) classes each appear on under 2% of records, marking them as comparatively open branches within this claim space.
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 Candidate Retrieval and Dual Encoder Models 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.