IP Library › Granted Patent US 12,597,065
Granted Patent B2
US 12,597,065 · App. 18/609,576 · Granted Apr 7, 2026

Probabilistic account linking

Inventors: Sahana Chattopadhyay Debnath (San Jose, CA); Thomas Schweiger (Seattle, WA)
Assignee: eBay Inc.
G06Q40/024G06Q20/4016
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 12,597,065
App. No.
18/609,576
Granted
Apr 7, 2026
Kind
B2
Abstract

Some aspects relate to technologies for probabilistic account linking, for instance, to perform fraud detection on online transaction platforms. In accordance with some configurations, linking strategies are defined for linking accounts based on account attributes. An average linking probability is generated for each linking strategy using account data for accounts on an online transaction platform, and the average linking probabilities are stored. To determine whether to link two accounts, linking strategies shared by the two accounts are identified, an account linking probability for the two accounts is generated using the average linking probabilities for the linking strategies shared by the two accounts, and the account linking probability is compared against a threshold. If the account linking probability satisfies the threshold, the accounts are linked and an action is taken based on the account linking.

Claims (63)

1 . One or more computer storage media storing computer-usable instructions that, when used by one or more computing devices, cause the one or more computing devices to perform operations, the operations comprising:

generating, by at least one of one or more servers of an online transaction platform, an average linking probability for each linking strategy from a plurality of linking strategies to provide a plurality of average linking probabilities, each linking strategy identifying a different combination of one or more attributes for assessing whether to link pairs of accounts on the online transaction platform based on whether the pairs of accounts share an attribute value for each of the one or more attributes, and wherein each average linking probability is generated by analyzing account data for a plurality of accounts on the online transaction platform to determine, for each linking strategy, a likelihood that accounts sharing attribute values correspond to the same entity based on observed associations among accounts in the account data;

storing, by at least one of the one or more servers, the plurality of average linking probabilities in a data store;

determining, by at least one of the one or more servers, one or more linking strategies shared by two accounts based on attribute values of attributes shared by the two accounts;

retrieving, from the data store by at least one of the one or more servers, the average linking probability for each of the one or more linking strategies shared by the two accounts;

generating, by at least one of the one or more servers, an account linking probability for the two accounts using the average linking probability for each of the one or more linking strategies shared by the two accounts;

performing, by at least one of the one or more servers, a comparison of the account linking probability for the two accounts to a threshold to determine whether to link the two accounts; and

responsive to determining to link the two accounts based on the comparison, causing, by at least one of the one or more servers, an action to be performed on the online transaction platform, the action comprising at least one of generating and providing a notification regarding linking of the two accounts, blocking a prospective transaction, reporting a previous transaction, or closing at least one of the two accounts.

2 . The one or more computer storage media of claim 1 , wherein the plurality of average linking probabilities includes a first set of average linking probabilities for a first linking strategy, each average linking probability from the first set of average linking probabilities for the first linking strategy corresponding to a different number of accounts sharing a same attribute value for the first linking strategy.

3 . The one or more computer storage media of claim 2 , wherein the two accounts share the first linking strategy, and wherein retrieving the average linking probability for each of the one or more linking strategies shared by the two accounts comprises:

retrieving a first average linking probability from the first set of average linking probabilities for the first linking strategy based on a total number of accounts sharing an attribute value of the first linking strategy with the two accounts.

4 . The one or more computer storage media of claim 1 , wherein generating the average linking probability for a first linking strategy from the plurality of linking strategies comprises:

determining, for each attribute value of a plurality of attribute values for the first linking strategy in the account data, a linking probability to provide a plurality of linking probabilities for the first linking strategy; and

generating the average linking probability for the first linking strategy using the plurality of linking probabilities for the first linking strategy.

5 . The one or more computer storage media of claim 4 , wherein determining the linking probability for a first attribute value of the plurality of attribute values for the first linking strategy comprises:

identifying a subset of accounts having the first attribute value for the first linking strategy;

accessing account cluster information for the subset of accounts; and

determining the linking probability for the first attribute value of the first linking strategy as a function of a number of accounts in each account cluster in the account cluster information and a total number of accounts in the subset of accounts.

6 . The one or more computer storage media of claim 1 , wherein the two accounts share two or more linking strategies, and wherein generating the account linking probability for the two accounts comprises:

selecting a subset of the two or more linking strategies with non-overlapping account attributes; and

generating the account linking probability for the two accounts as a function of the subset of the two or more linking strategies.

7 . The one or more computer storage media of claim 6 , wherein selecting the subset of the two or more linking strategies with non-overlapping account attributes comprises:

sorting the two or more linking strategies in descending order of average linking probabilities; and

iteratively evaluating the two or more linking strategies in the descending order of average linking probabilities to determine whether to include the linking strategy at each iteration in the subset of the two or more linking strategies, wherein a linking strategy evaluated at an iteration is included in the subset of the two or more linking strategies if the linking strategy does not correspond with an account attribute of a linking strategy previously selected for the subset of the two or more linking strategies.

8 . The one or more computer storage media of claim 1 , wherein causing the action to be performed responsive to determining to link the two accounts comprises blocking a prospective transaction by preventing an electronic transaction between the two accounts on the online transaction platform.

9 . The one or more computer storage media of claim 1 , wherein causing the action to be performed responsive to determining to link the two accounts comprises providing a notification regarding linking of the two accounts by communicating, over a network to a remote computing device, a user interface that presents an indication that the two accounts have been linked.

10 . A computer-implemented method comprising:

generating, by at least one of one or more servers of an online transaction platform, an average linking probability for each linking strategy from a plurality of linking strategies using account data for a plurality of accounts on an online transaction platform, each linking strategy identifying a different combination of one or more attributes for assessing whether to link pairs of accounts on the online transaction platform based on whether the pairs of accounts share an attribute value for each of the one or more attributes, wherein a first average linking probability for a first linking strategy is generated by:

determining a linking probability for each attribute value from a plurality of attribute values for an account attribute corresponding to the first linking strategy using account cluster information identifying account clusters, wherein a same number of accounts have each attribute value from the plurality of attribute values for the account attribute corresponding to the first linking strategy, and

generating the first average linking probability for the first linking strategy as a function for the linking probabilities for the plurality of attributes values for the account attribute corresponding to the first linking strategy;

storing, by at least one of the one or more servers, the average linking probabilities for the plurality of linking strategies in a data store, wherein each average linking probability is stored in association with a linking strategy identifier and an indication of a number of accounts sharing a same linking strategy attribute value;

generating, by at least one of the one or more servers, an account linking probability for two accounts on the online transaction platform using one or more of the average linking probabilities in the data store; and

responsive to determining to link the two accounts based on the account linking probability for the two accounts, causing, by at least one of the one or more servers, an action to be performed on the online transaction platform, the action comprising at least one of generating and providing a notification regarding linking of the two accounts, blocking a prospective transaction, reporting a previous transaction, or closing at least one of the two accounts.

11 . The computer-implemented method of claim 10 , wherein determining the linking probability for a first attribute value of the plurality of attribute values for the account attribute corresponding to the first linking strategy comprises:

identifying a subset of accounts having the first attribute value for the account attribute corresponding to the first linking strategy;

determining, from the account cluster information, an account cluster for each account in the subset of accounts; and

determining the linking probability for the first attribute value for the account attribute corresponding to the first linking strategy as a function of a number of accounts in each account cluster and a total number of accounts in the subset of accounts.

12 . The computer-implemented method of claim 10 , wherein

the two accounts are linked based on a comparison of the account linking probability for the two accounts against a threshold.

13 . The computer-implemented method of claim 10 , wherein generating the account linking probability for the two accounts comprises:

identifying one or more linking strategies shared by the two accounts and a total number of accounts sharing each of the one or more linking strategies;

retrieving, from the data store, the average linking probability for each of the one or more linking strategies based on the total number of accounts sharing each of the one or more linking strategies; and

generating the account linking probability for the two accounts as a function of the average linking probability for each of the one or more linking strategies.

14 . The computer-implemented method of claim 13 , wherein the two accounts share two or more linking strategies, and wherein generating the account linking probability for the two accounts comprises:

selecting a subset of the two or more linking strategies having the highest average linking probabilities without overlapping account attributes.

15 . The computer-implemented method of claim 13 , wherein the two accounts share two or more linking strategies, and wherein the account linking probability is generated for the two accounts using a recursive function.

16 . The computer-implemented method of claim 10 , wherein causing the action to be performed comprises blocking a prospective transaction by preventing an electronic transaction between the two accounts on the online transaction platform.

17 . A computer system comprising:

one or more processors; and

one or more computer storage medium storing computer-usable instructions that, when used by the one or more processors, causes the computer system to perform operations comprising:

generating, by at least one server of one or more servers of an online transaction platform, an account linking probability for two accounts on the online transaction platform by:

identifying, from a plurality of linking strategies, one or more linking strategies shared by the two accounts, each linking strategy identifying a different combination of one or more attributes for assessing whether to link pairs of accounts on the online transaction platform based on whether the pairs of accounts share an attribute value for each of the one or more attributes,

retrieving, from a data store storing an average linking probability for each linking strategy from the plurality of linking strategies, an average linking probability corresponding to each linking strategy from the one or more linking strategies shared by the two accounts based on a total number of accounts sharing each linking strategy with the two accounts, wherein the average linking probabilities for the plurality of linking strategies were generated using account cluster information for a plurality of accounts on the online transaction platform, and

generating the account linking probability for the two accounts using the average linking probability corresponding to each linking strategy from the one or more linking strategies shared by the two accounts;

linking the two accounts based on the account linking probability; and

causing, by at least one of the one or more servers, one or more actions to be performed on the online transaction platform responsive to linking the two accounts, the one or more actions comprising at least one of generating and providing a notification regarding linking of the two accounts, blocking a prospective transaction, reporting a previous transaction, or closing at least one of the two accounts.

18 . The computer system of claim 17 , wherein the account linking probability is generated in response to a request for a transaction between the two accounts; and wherein the one or more actions comprise not allowing the transaction.

19 . The computer system of claim 17 , wherein the two accounts share two or more linking strategies; and wherein generating the account linking probability for the two accounts further comprises:

selecting a subset of the two or more linking strategies with non-overlapping account attributes; and

generating the account linking probability for the two accounts as a function of the subset of the two or more linking strategies.

20 . The computer system of claim 19 , wherein selecting the subset of the two or more linking strategies with non-overlapping account attributes comprises:

sorting the two or more linking strategies in descending order of average linking probabilities; and

iteratively evaluating the two or more linking strategies in the descending order of average linking probabilities to determine whether to include the linking strategy at each iteration in the subset of the two or more linking strategies, wherein a linking strategy evaluated at an iteration is included in the subset of the two or more linking strategies if the linking strategy does not correspond with an account attribute of a linking strategy previously selected for the subset of the two or more linking strategies.

Assignments (1)
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Mar 19, 2024
From: CHATTOPADHYAY DEBNATH, SAHANA; SCHWEIGER, THOMAS
To: EBAY INC.
Reel/Frame 066828/0555 →
Continuity (1)
Related Publication 20250299194A1 · Sep 25, 2025
References Cited (21)
US 8606631B2 · Blagg · 2013 [cited by applicant]
US 9691058B2 · Epler et al. · 2017 [cited by applicant]
US 10621492B2 · Oberhofer et al. · 2020 [cited by applicant]
US 20140330845A1 · Feldschuh · 2014 [cited by examiner]
US 20170286539A1 · Smith et al. · 2017 [cited by applicant]
US 20180113928A1 · Oberhofer · 2018 [cited by examiner]
US 20190295089A1 · Jia · 2019 [cited by examiner]
US 20220253406A1 · Abu-Abed et al. · 2022 [cited by applicant]
CN 109150974B · 2020 [cited by applicant]
Ahmad et al., “Missing Link Prediction using Common Neighbor and Centrality based Parameterized Algorithm”, Scientific Reports—Natureresearch, vol. 10, No. 364, 2020, 9 pages. [cited by applicant]
Brown et al., “Estimating parameters for probabilistic linkage of privacy-preserved datasets”, BMC Medical Research Methodology, vol. 17, No. 95, Jul. 2017, 11 pages. [cited by applicant]
Dasylva et al., “Revisiting the probabilistic method of record linkage”, Available at : <arXiv:1911.01874v1 [stat.ME] Nov. 5, 2019>, Nov. 5, 2019, 96 pages. [cited by applicant]
Farley et al., “A Bayesian Approach to Linking Data Without Unique Identifiers”, Journal of Statistical Software, vol. 10, No. 2, Available at : <https://arxiv.org/pdf/2012.00601.pdf>, Sep. 27, 2021, pp. 1-26. [cited by applicant]
Goga, O., “Matching user accounts across online social networks :methods and applications”, HAL Open Science, Jun. 18, 2015. 151 pages. [cited by applicant]
Sayers et al., “Probabilistic record linkage”, International Journal of Epidemiology, vol. 45, No. 3, 2016, pp. 954-964. [cited by applicant]
Seidl et al., “CC-MR-Finding Connected Components in Huge Graphs with MapReduce”, Data Management and Data Exploration Group, pp. 1-16. [cited by applicant]
Talburt, J., “Entity Resolution and Information Quality”, Google Books—Elsevier, 2011, 1 page. [cited by applicant]
Veigh et al., “Scaling Bayesian Probabilistic Record Linkage with Post-Hoc Blocking: An Application to the California Great Registers”,Available at : <arXiv:1905.05337v2 [stat.ME] Jan. 7, 2020>, Jan. 8, 2020. 69 pages. [cited by applicant]
Yang et al., “A user identity matching method based on integrating account attributes”, Journal of Physics: Conference Series, 2018, 8 pages. [cited by applicant]
Chang et al., “AI-URG: Account Identity-Based Uncertain Graph Framework for Fraud Detection”, IEEE Transactions on Computational Social Systems, vol. 11, No. 3, Nov. 2, 2023, pp. 3706-3728. [cited by applicant]
Extended European Search Report received for European Patent Application No. 25164022.3, mailed on Jul. 30, 2025, 9 pages. [cited by applicant]