Risk identification engine and supply chain graph generator
The present invention relates to a computer-based system for identifying supply chain risks and generating supply chain graphs representing an interconnected network of entities. An industrial graph database application is configured to account for direct and indirect (transitive) supplier risk and importance, based on a weighted set of measures: criticality, replaceability, centrality and distance. A graph-based model serves as an interactive and visual supply chain risk and importance explorer. A supply network is induced from textual data by applying text mining techniques to news stories and used to populate the supply chain/graph database.
1. A computer-based system for identifying supply chain risks and generating supply chain graphs representing an interconnected network of entities, the system comprising: a computing device having a processor in electrical communication with a memory, the memory adapted to store data and instructions for executing by the processor; a graph database using graph structures for semantic queries and accessible by the computing device and having stored therein a first data set comprising supply relations data, a set of nodes, and a set of edges; wherein each node represents a company and comprises a set of attributes related to the company, the set of attributes including two or more of: business sector, credit risk score, risk score, company name, and closeness centrality score; wherein each edge represents a directed supply relationship pointing from a supplier to a customer and directly relates items stored in the graph database; wherein the supply relations data is at least in part derived from a set of source data in electronic form and representing textual content comprising potential relation and risk phrases and/or numeric data; a set of application programming interfaces (APIs), each API in the set of APIs adapted to execute an associated algorithm having a set of input parameters on the graph database and return a structured file of a type associated with the API, the structured file comprising a header and a set of block types adapted to store a set of results of the execution of the associated algorithm on the graph database; a risk scoring module adapted to access, via the set of APIs using at least one graph query language comprising a set of callable commands, the first data set data from the graph database as provided by the set of APIs in the structured file, generate a set of scores related to the first data set, and store the set of scores in the graph database, wherein the risk scoring module generates the set of scores using both direct and transitive risk propagation along a plurality of nodes; and a supply graph generator adapted to access the first data set and the set of scores stored in the graph database and generate for presentation at a remote user computing device a directed graph comprising a plurality of interconnected nodes and edges representing a network of supply chain related entities, wherein the closeness centrality score C for each node is determined by using the following equation:
C ( x )=( N− 1)/ Ey d ( y,x )
wherein here a given node(x) represents company x and the closeness centrality score C(x) for company x represents a measure of node(x)' s centrality in a graph G as the sum of the length of the shortest paths between the node(x) and all other nodes in the graph, with N being the total node count and d(x,y) being the distance between nodes x and y.
2. The system of claim 1 wherein the supply graph generator is adapted to generate for presentation at a remote user computing device at least one of: 1) a cyclic graph; and 2) a directed acyclic graph (DAG).
3. The system of claim 1 wherein the graph database is adapted for use with one of Cypher Query Language or SPARQL query language.
4. The system of claim 1 wherein the graph database is a Resource Description Framework (RDF) database and supply relations data is stored in a RDF triple format and is used to populate the graph database.
5. The system of claim 1 further comprising a supply relations search and analytics engine adapted to automatically identify and extract supply relations data from the set of source data and to store the extracted supply relations data in the graph database, the extracted supply relations data comprising supplier/customer pairs, and wherein the supply relations search and analytics engine is adapted to automatically update the first data set on a continuing basis as additional supply relations data is extracted from additional source data.
6. The system of claim 1 wherein the graph database is from the group consisting of: neo4j, Gephi, AllegroGraph, ArangoDB, Blazegraph, Cayley, DGraph, DataStax, SAP Hana, Oracle Spatial and Graph, OpenLink Virtuoso, Sparksee, Graphbase, gStore, InfiniteGraph, JanusGraph, MarkLogicSqrrl Enterprise, Teradata Aster, TigerGraph, Resource Description Framework (RDF) database, and Microsoft SQL Server.
7. The system of claim 1 further comprising:
an entity-risk relation classifier adapted to identify and extract entity-risk relations from the set of source data, the entity-risk relation classifier comprising:
a risk tagger adapted to identify in the set of source data a set of risk candidates based on the set of risk types; and
an entity tagger adapted to identify mentions of entity names in the set of source data;
wherein the entity-risk relation classifier maps the identified set of risk types to the identified entity names.
8. The system of claim 1 , wherein for a given customer c having an associated node c and having an identified set of suppliers s={s 0 , . . . s n } each having an associated node, the risk scoring module is further adapted to generate a set of risk scores R={r 0 , . . . r n } and a set of importance scores I={i 0 , . . . i n }.
9. The system of claim 8 , wherein each risk score r m ∈ R is based on a single attribute of a node in the graph G, the single attribute representing a credit risk associated with supplier m.
10. The system of claim 8 , wherein each importance score i m ∈ I is an aggregate of a plurality of measures including at least two of the following measures: criticality; replaceability; centrality; and distance;
wherein criticality is a function of a proportion of goods, materials, commodities or other things received in business sector q associated with target company c as supplied from suppliers s={s 0 , . . . s n } operating in identified business sector associated with such suppliers s;
wherein replaceability is a function of the sum of the number of suppliers s∈ S that operate in the same business sector as s m ;
wherein centrality is a closeness centrality score represents the importance i m of supplier s m as a function of the associated node's centrality in a graph G as the sum of the length of the shortest paths between the node(s m ) and all other nodes in the graph; and
wherein distance is a function of a distance between the node associated with each supplier s m and the node associated with company c.
11. The system of claim 10 , wherein the importance score i m ∈ I for a given supplier s m is an aggregate of criticality (a), replaceability (b), centrality (c), and distance (d) measures associated with that supplier as represented in the following equation:
i
m
=
(
a
+
b
+
c
+
d
4
)
/
max
∈
I
.
12. The system of claim 1 further comprising a supply relations search and analytics engine adapted to automatically identify and extract supply relations data from the set of source data and to store the extracted supply relations data in the graph database, and wherein the supply relations search and analytics engine comprises a machine learning module and is further adapted to automatically identify and extract supply relations data based on the following model: a) identify company names appearing in a document; b) parse documents into word sentences; c) identify and select candidate word sentences that contain two company names and a predefined relation-indicating pattern; and d) label company names appearing in the candidate word sentences as one of supplier, customer or neither.
13. The system of claim 12 , wherein the relation-indicating pattern is based on a set of indicative n-grams and variations of the indicative n-grams associated with relation-indicating words.
14. The system of claim 1 wherein the set of source data received comprises one or more of: an indexed search; a news archive; a news feed; structured data sets; unstructured data sets; social media content; regulatory filings Bills of lading; Customs forms/data; Procurement data bases; and Enterprise Resource Planning (ERP) systems.
15. A method for identifying supply chain risks and generating supply chain graphs representing an interconnected network of entities, the method comprising: storing data and executable instructions in a memory of a computing device having a processor in electrical communication with the memory for accessing and executing the stored data and instructions; storing in a graph database accessible by the computing device a set of graph structures and a first data set comprising supply relations data, a set of nodes, and a set of edges; receiving, by the computing device, a semantic query for execution; accessing and executing, by the computing device, the set of graph structures and processing the received semantic query using the set of graph structures; wherein each node represents a company and comprises a set of attributes related to the company, the set of attributes including two or more of: business sector, credit risk score, risk score, company name, and closeness centrality score; wherein each edge represents a directed supply relationship pointing from a supplier to a customer and directly relates items stored in the graph database; wherein the supply relations data is at least in part derived from a set of source data in electronic form and representing textual content comprising potential relation and risk phrases and/or numeric data; executing a set of application programming interfaces (APIs), each API in the set of APIs having an associated algorithm and set of input parameters on the graph database, and returning a structured file of a type associated with a given API, the structured file comprising a header and a set of block types adapted to store a set of results of the execution of the associated algorithm on the graph database; accessing, via the set of APIs using at least one graph query language comprising a set of callable commands, the first data set data from the graph database as provided by the set of APIs in the structured file, generating a set of scores related to the first data set, and storing the set of scores in the graph database, wherein the set of scores is generated using both direct and transitive risk propagation along a plurality of nodes; and accessing the first data set and the set of scores stored in the graph database and generating for presentation at a remote user computing device a directed graph comprising a plurality of interconnected nodes and edges representing a network of supply chain related entities; wherein the closeness centrality score C for each node is determined by using the following equation:
C ( x )=( N− 1)/ Ey d ( y,x )
wherein here a given node(x) represents company x and the closeness centrality score C(x) for company x represents a measure of node(x)' s centrality in a graph G as the sum of the length of the shortest paths between the node(x) and all other nodes in the graph, with N being the total node count and d(x,y) being the distance between nodes x and y.
16. The method of claim 15 further comprising by a supply relations search and analytics engine automatically identifying and extracting supply relations data from the set of source data and storing the extracted supply relations data in the graph database, the extracted supply relations data comprising supplier/customer pairs, and wherein the supply relations search and analytics engine is adapted to automatically update the first data set on a continuing basis as additional supply relations data is extracted from additional source data.
17. The method of claim 15 further comprising:
by an entity-risk relation classifier executed on the computing device, identifying and extracting entity-risk relations from the set of source data, including identifying in the set of source data a set of risk candidates based on the set of risk types; and identifying mentions of entity names in the set of source data; and mapping the identified set of risk types to the identified entity names.
18. The method of claim 15 , wherein for a given customer c having an associated node c and having an identified set of suppliers s={s 0 , . . . s n } each having an associated node, the risk scoring module is further adapted to generate a set of risk scores R={r 0 , . . . r n } and a set of importance scores I={i 0 , . . . i n }.
19. The method of claim 18 , wherein each risk score r m ∈ R is based on a single attribute of a node in the graph G, the single attribute representing a credit risk associated with supplier m.
20. The method of claim 18 , wherein each importance score i m ∈ I is an aggregate of a plurality of measures including at least two of the following measures: criticality; replaceability; centrality; and distance;
wherein criticality is a function of a proportion of goods, materials, commodities or other things received in business sector q associated with target company c as supplied from suppliers s={s 0 , . . . s n } operating in identified business sector associated with such suppliers s;
wherein replaceability is a function of the sum of the number of suppliers s∈ S that operate in the same business sector as s m ;
wherein centrality is a closeness centrality score represents the importance i m of supplier s m as a function of the associated node's centrality in a graph G as the sum of the length of the shortest paths between the node(s m ) and all other nodes in the graph; and
wherein distance is a function of a distance between the node associated with each supplier s m and the node associated with company c.
21. The method of claim 20 , wherein the importance score i m ∈ I for a given supplier s m is an aggregate of criticality (a), replaceability (b), centrality (c), and distance (d) measures associated with that supplier as represented in the following equation:
i
m
=
(
a
+
b
+
c
+
d
4
)
/
max
∈
I
.
22. The method of claim 15 further comprising automatically identifying and extracting supply relations data from the set of source data and to store the extracted supply relations data in the graph database, and automatically identifying and extracting supply relations data based on the following model: a) identify company names appearing in a document; b) parse documents into word sentences; c) identify and select candidate word sentences that contain two company names and a predefined relation-indicating pattern; and d) label company names appearing in the candidate word sentences as one of supplier, customer or neither.