Systems and methods for identifying trustworthiness of data
Systems and methods are described for determining trustworthiness. The systems and methods may perform obtaining a plurality of nodes associated with a first entity, wherein the plurality of nodes correspond to one or more additional entities, each of the one or more additional entities being defined by a trust metric, a relationship indication, and at least one activation function, generating, for each of the plurality of nodes, an output by executing each activation function according to a set of predefined rules defined for the plurality of nodes, wherein each activation function uses a respective trust metric defined for the one or more additional entities, and generating a model for determining trustworthiness of the first entity based on each relationship indication and the output for each of the plurality of nodes.
1 . A computer-implemented method for determining trustworthiness, the method comprising:
obtaining a plurality of nodes associated with a first entity, wherein the plurality of nodes correspond to one or more additional entities, each of the one or more additional entities being defined by a trust metric, a relationship indication, and at least one activation function;
generating, for each of the plurality of nodes, an output by executing each activation function according to a set of predefined rules defined for the plurality of nodes, wherein each activation function uses a respective trust metric defined for the one or more additional entities, wherein the set of predefined rules is applied by a neuroevolutionary model in which populations of neural networks compete to achieve a trust identification goal such that each activation function is evolved based on the competition between the populations of neural networks; and
generating a model for determining trustworthiness of the first entity based at least in part on each relationship indication and the output for each of the plurality of nodes, wherein generating the model comprises receiving observer-based input from at least one entity acting as an observer and injecting an uncertainty influence derived from the observer-based input directly into one or more of the plurality of nodes such that the uncertainty influence is reflected in the respective trust metrics and aggregated trust output associated with the one or more plurality of nodes.
2 . The computer-implemented method of claim 1 , further comprising:
receiving a request to determine trustworthiness of the first entity, the request including at least one parameter;
generating, using the model and the at least one parameter, an aggregated trust metric for the first entity; and
generating a graphical view of the model based on the aggregated trust metric.
3 . The computer-implemented method of claim 1 , wherein:
each node is a statement or entity associated with the first entity; and
the model is a trust network comprising a plurality of neural networks configured to execute, in parallel, each activation function to generate an aggregated trust metric for the first entity based on the trust metric and relationship indication for each statement or entity.
4 . The computer-implemented method of claim 3 , wherein the trust network represents a network in which each of the one or more entities associated with one or more of the plurality of nodes is defined by an uncertainty metric, wherein the uncertainty metric is determined as one minus the trust metric of the respective entity, the uncertainty metric having a value between zero and one, wherein zero represents no uncertainty and one represents full uncertainty, and wherein the uncertainty metric of each entity influences the aggregated trust metric generated by the trust network.
5 . The computer-implemented method of claim 1 , further comprising:
receiving an additional node from a second entity, the additional node being defined by at least one trust indicator and a relationship to the first entity;
generating, using the model, an aggregated trust metric for the first entity;
biasing the aggregated trust metric according to the at least one trust indicator; and
generating a graphical view of the model, the graphical view depicting an influence of the at least one trust indicator.
6 . The computer-implemented method of claim 5 , wherein:
the aggregated trust metric represents a probability of the first entity being trustworthy; and
the at least one trust indicator modifies the probability.
7 . The computer-implemented method of claim 1 , further comprising:
in response to receiving one or more updated trust metrics associated with one or more of the plurality of nodes associated with the first entity:
generating an updated model based on the one or more updated trust metrics and the respective activation functions associated with additional entities having an updated trust metric; and
generating, using the updated model, an updated aggregated trust metric and generating a graphical user interface of the model based on the updated aggregated trust metric.
8 . The computer-implemented method of claim 1 , wherein:
the plurality of nodes are conditioned on at least one node context; and
the generating of the output for each of the plurality of nodes is based on feeding the at least one node context into at least one of the activation functions.
9 . The computer-implemented method of claim 1 , wherein:
the set of predefined rules comprises a plurality of learning rules that when used during execution of each activation function:
modifies at least one trust metric of at least one of the plurality of nodes; and
generates, using the model and the at least one modified trust metric, an updated aggregated trust metric.
10 . A system comprising:
at least one processing device; and
memory storing instructions that when executed cause the processing device to perform operations comprising:
obtaining a plurality of nodes associated with a first entity, wherein the plurality of nodes correspond to one or more additional entities, each of the one or more additional entities being defined by a trust metric, a relationship indication, and at least one activation function;
generating, for each of the plurality of nodes, an output by executing each activation function according to a set of predefined rules defined for the plurality of nodes, wherein each activation function uses a respective trust metric defined for the one or more additional entities, wherein the set of predefined rules is applied by a neuroevolutionary model in which populations of neural networks compete to achieve a trust identification goal such that each activation function is evolved based on the competition between the populations of neural networks; and
generating a model for determining trustworthiness of the first entity based at least in part on each relationship indication and the output for each of the plurality of nodes, wherein generating the model comprises receiving observer-based input from at least one entity acting as an observer and injecting an uncertainty influence derived from the observer-based input directly into one or more of the plurality of nodes such that the uncertainty influence is reflected in the respective trust metrics and aggregated trust output associated with the one or more plurality of nodes.
11 . The system of claim 10 , wherein the operations further comprise:
receiving a request to determine trustworthiness of the first entity, the request including at least one parameter;
generating, using the model and the at least one parameter, an aggregated trust metric for the first entity; and
generating a graphical view of the model based on the aggregated trust metric.
12 . The system of claim 10 , wherein:
each node is a statement or entity associated with the first entity; and
the model is a trust network comprising a plurality of neural networks configured to execute, in parallel, each activation function to generate an aggregated trust metric for the first entity based on the trust metric and relationship indication for each statement or entity.
13 . The system of claim 12 , wherein the trust network represents a network influenced by one or more entities associated with one or more of the plurality of nodes, wherein each of the one or more entities is defined by an uncertainty metric having a value between zero and one determined as one minus the respective trust metric.
14 . The system of claim 10 , wherein the operations further comprise:
receiving an additional node from a second entity, the additional node being defined by at least one trust indicator and a relationship to the first entity;
generating, using the model, an aggregated trust metric for the first entity;
biasing the aggregated trust metric according to the at least one trust indicator; and
generating a graphical view of the model, the graphical view depicting an influence of the at least one trust indicator.
15 . The system of claim 14 , wherein:
the aggregated trust metric represents a probability of the first entity being trustworthy; and
the at least one trust indicator modifies the probability.
16 . The system of claim 10 , wherein the operations further comprise:
in response to receiving one or more updated trust metrics associated with one or more of the plurality of nodes associated with the first entity:
generating an updated model based on the one or more updated trust metrics and the respective activation functions associated with additional entities having an updated trust metric; and
generating, using the updated model, an updated aggregated trust metric and generating a graphical user interface of the model based on the updated aggregated trust metric.
17 . The system of claim 10 , wherein:
the plurality of nodes are conditioned on at least one node context; and
the generating of the output for each of the plurality of nodes is based on feeding the at least one node context into at least one of the activation functions.
18 . The system of claim 10 , wherein:
the set of predefined rules comprises a plurality of learning rules that when used during execution of each activation function:
modifies at least one trust metric of at least one of the plurality of nodes; and
generates, using the model and the at least one modified trust metric, an updated aggregated trust metric.
19 . A non-transitory computer-readable medium comprising:
at least one processor; and
a memory storing instructions that, when executed by the at least one processor, cause the at least one processor to perform operations comprising:
obtaining a plurality of nodes associated with a first entity, wherein the plurality of nodes correspond to one or more additional entities, each of the one or more additional entities being defined by a trust metric, a relationship indication, and at least one activation function;
generating, for each of the plurality of nodes, an output by executing each activation function according to a set of predefined rules defined for the plurality of nodes, wherein each activation function uses a respective trust metric defined for the one or more additional entities, wherein the set of predefined rules is applied by a neuroevolutionary model in which populations of neural networks compete to achieve a trust identification goal such that each activation function is evolved based on the competition between the populations of neural networks; and
generating a model for determining trustworthiness of the first entity based at least in part on each relationship indication and the output for each of the plurality of nodes, wherein generating the model comprises receiving observer-based input from at least one entity acting as an observer and injecting an uncertainty influence derived from the observer-based input directly into one or more of the plurality of nodes such that the uncertainty influence is reflected in the respective trust metrics and aggregated trust output associated with the one or more plurality of nodes.
20 . The non-transitory computer-readable medium of claim 19 , wherein the operations further comprise:
receiving a request to determine trustworthiness of the first entity, the request including at least one parameter;
generating, using the model and the at least one parameter, an aggregated trust metric for the first entity; and
generating a graphical view of the model based on the aggregated trust metric.
21 . The non-transitory computer-readable medium of claim 19 , wherein:
each node is a statement or entity associated with the first entity; and
the model is a trust network comprising a plurality of neural networks configured to execute, in parallel, each activation function to generate an aggregated trust metric for the first entity based on the trust metric and relationship indication for each statement or entity.
22 . The non-transitory computer-readable medium of claim 21 , wherein the trust network represents a network influenced by one or more entities associated with one or more of the plurality of nodes, wherein each of the one or more entities is defined by an uncertainty metric having a value between zero and one determined as one minus the respective trust metric.
23 . The non-transitory computer-readable medium of claim 19 , wherein the operations further comprise:
receiving an additional node from a second entity, the additional node being defined by at least one trust indicator and a relationship to the first entity;
generating, using the model, an aggregated trust metric for the first entity;
biasing the aggregated trust metric according to the at least one trust indicator; and
generating a graphical view of the model, the graphical view depicting an influence of the at least one trust indicator.
24 . The non-transitory computer-readable medium of claim 23 , wherein:
the aggregated trust metric represents a probability of the first entity being trustworthy; and
the at least one trust indicator modifies the probability.
25 . The non-transitory computer-readable medium of claim 19 , wherein the operations further comprise:
in response to receiving one or more updated trust metrics associated with one or more of the plurality of nodes associated with the first entity:
generating an updated model based on the one or more updated trust metrics and the respective activation functions associated with additional entities having an updated trust metric; and
generating, using the updated model, an updated aggregated trust metric and generating a graphical user interface of the model based on the updated aggregated trust metric.
26 . The non-transitory computer-readable medium of claim 19 , wherein:
the plurality of nodes are conditioned on at least one node context; and
the generating of the output for each of the plurality of nodes is based on feeding the at least one node context into at least one of the activation functions.
27 . The non-transitory computer-readable medium of claim 19 , wherein:
the set of predefined rules comprises a plurality of learning rules that when used during execution of each activation function:
modifies at least one trust metric of at least one of the plurality of nodes; and
generates, using the model and the at least one modified trust metric, an updated aggregated trust metric.
28 . A computer-implemented method for determining trustworthiness, the method comprising:
obtaining a plurality of nodes associated with a first entity, wherein the plurality of nodes correspond to one or more additional entities, each of the one or more additional entities being defined by a trust metric, a relationship indication, and at least one activation function;
generating, for each of the plurality of nodes, an output by executing each activation function according to a set of predefined rules defined for the plurality of nodes using a parallel computational framework comprising a plurality of neural networks configured to execute each activation function concurrently, wherein each neural network corresponds to a respective socket type that defines an algorithm to execute upon input data received at the respective node and wherein each activation function uses a respective trust metric defined for the one or more additional entities; and
generating a model for determining trustworthiness of the first entity based at least in part on each relationship indication and the output for each of the plurality of nodes, wherein the model is a trust network generated by the parallel computational framework in which each socket type independently processes its received input data according to the algorithm defined for that socket type and collectively produces an aggregated trust metric for the first entity by summing the outputs across the plurality of neural networks and applying an arctangent function to the sum.