System and method for dynamic segmentation of network nodes to manage workflow data synchronization

1 . A computer-implemented method for dynamically generating a recommendation, the method comprising:

(a) maintaining one or more timeseries databases configured to store:

i) customer profiles associated with a plurality of customers, wherein the customer profiles comprise customer portfolios each comprising timeseries data stored in the one or more timeseries databases;

ii) product traits associated with a plurality of products, wherein the product traits comprise product interest data stored in the one or more timeseries databases associated with timestamps, and wherein the product interest data is dynamically modifiable by a decay function based at least in part on the timestamps; and

iii) advisor profiles associated with a plurality of advisors, wherein an advisor profile comprises an advisor score generated by a first model based on an engagement data of the associated advisor, and wherein the engagement data of the associated advisor is stored in the one or more timeseries databases associated with timestamps, and wherein the advisor score is weighted based at least in part on the timestamps;

(b) training a model on a training data set by:

i) constructing a graph data structure comprising nodes each representing one of a plurality of customers, advisors, or products from historical data, and edges connecting pairs of nodes, wherein an edge connecting an advisor node to a customer node is weighted by a value derived from the product traits associated with the product involved in the advisor-customer relationship represented by that edge,

ii) identifying, within the graph data structure, closed-loop triangular subgraphs each comprising one customer node, one advisor node, and one product node, wherein each identified closed-loop triangular subgraph indicates a historical three-way match among the represented customer, advisor, and product, and associating a matched label with the customer-advisor-product combination represented by each identified closed-loop triangular subgraph,

iii) adjusting parameters of the model by comparing predicted match outputs of the model for candidate customer-advisor-product combinations against the matched and not-matched labels, and updating the parameters based on the comparison, including based on whether a customer accepted, rejected, or selected a different product than a previously recommended product;

(c) receiving an event, wherein the event is indicative of one of a plurality of triggers to adjust a customer portfolio associated with a customer, wherein the event is stored in the one or more timeseries databases with a timestamp and wherein the customer portfolio comprises at least a risk tolerance score and a financial goal associated with the customer;

(d) upon adjusting the customer portfolio, modifying at least the product interest data of the product traits by applying the decay function and taking as input the product traits comprising the modified product interest data, the adjusted customer profile, and the advisor profiles, by the model trained in (b), to generate an output indicative of a match among i) the customer, ii) an advisor from the plurality of the advisors and iii) a product from the plurality of products; and

(e) transmitting the output in (d) to a user device associated with the customer or a user device associated with the advisor.

2 . The method of claim 1 , wherein the model generates an alignment score between a customer and an advisor for the product, and if the alignment score exceeds a pre-defined threshold, the model generates the output indicative of a match between the customer and the advisor for the product.

3 . The method of claim 1 , wherein the event comprises receiving a new product and product traits associated with the new product, and wherein the selected product is the new product.

4 . The method of claim 1 , wherein the event comprises a market change.

5 . The method of claim 1 , wherein the advisor profiles comprise advisor bandwidths and advisor expertise areas.

6 . The method of claim 1 , further comprising:

transmitting an advisor notification to the user device associated with the advisor, wherein the advisor notification comprises a customer profile associated with the customer, and product traits associated with the product.

7 . The method of claim 6 , wherein the advisor notification further comprises a set of suggestions indicative of communication channel, communication frequency, and communication time frame with the customer.

8 . The method of claim 1 , further comprising:

transmitting a customer notification to the user device associated with the customer, wherein the customer notification comprises the advisor profile associated with the advisor.

9 . A computer-implemented method for dynamically generating a recommendation, the method comprising:

(a) maintaining one or more timeseries databases configured to store:

i) customer profiles associated with a plurality of customers, wherein the customer profiles comprise customer financial goals and customer portfolio and wherein a customer portfolio comprises timeseries data stored in the one or more timeseries databases and a risk tolerance score associated with a respective customer;

ii) product traits associated with a plurality of products, wherein the product traits comprise product interest data stored in the one or more timeseries databases associated with timestamps, and wherein the product interest data is dynamically modifiable by a decay function based at least in part on the timestamps; and

iii) advisor profiles associated with a plurality of advisors, wherein an advisor profile comprises an advisor score generated by a first model based on an engagement data of the associated advisor, and wherein the first model is trained by a machine learning algorithm using training data comprising a set of similar advisors' engagement data and associated advisor scores and wherein a feedback from a customer is fed back to the machine learning algorithm to update the first model;

(b) training a model on a training data set by:

i) constructing a graph data structure comprising nodes each representing one of a plurality of customers, advisors, or products from historical data, and edges connecting pairs of nodes, wherein an edge connecting an advisor node to a customer node is weighted by a value derived from the product traits associated with the product involved in the advisor-customer relationship represented by that edge,

ii) identifying, within the graph data structure, closed-loop triangular subgraphs each comprising one customer node, one advisor node, and one product node, wherein each identified closed-loop triangular subgraph indicates a historical three-way match among the represented customer, advisor, and product, and associating a matched label with the customer-advisor-product combination represented by each identified closed-loop triangular subgraph,

iii) adjusting parameters of the model by comparing predicted match outputs of the model for candidate customer-advisor-product combinations against the matched label and not-matched labels, and updating the parameters based on the comparison, including based on whether a customer accepted, rejected, or selected a different product than a previously recommended product;

(c) receiving an event indicative of a market change and detecting a change in the product traits associated with one or more of the products based at least in part on the market change;

(d) upon detecting the change in the product traits of the one or more products, performing a balancing action to re-align with the customer financial goals of the associated customers using a forecast model while the change is determined to be within the risk tolerance score of the associated customers;

(e) taking as input the adjusted product traits, the customer profiles associated with the plurality of customers, and the advisor profiles, by the model trained in (b), to generate an output indicative of a match among i) one of the plurality of customers, ii) an advisor from the plurality of advisors and iii) a product from the plurality of products; and

(f) transmitting the output of the model to a user device associated with the customer or a user device associated with the advisor.

10 . The method of claim 9 , wherein the advisor profiles comprise advisor bandwidths and advisor expertise areas.

11 . The method of claim 9 , further comprising:

transmitting an advisor notification to the user device associated with the advisor, wherein the advisor notification comprises a customer profile associated with the customer, and product traits associated with the product.

12 . The method of claim 11 , wherein the advisor notification further comprises a set of suggestions indicative of communication channel, communication frequency, and communication time frame with the customer.

13 . The method of claim 9 , further comprising:

transmitting a customer notification to the user device associated with the customer, wherein the customer notification comprises the advisor profile associated with the advisor.