Event visualization device for generating hierarchical directed acyclic graph and related computer program product

1. An event visualization device, configured to generate one or more directed acyclic graphs (DAGs) that can be used as a basis for diagnosing whether a target network system has been hacked according to a plurality of activities records, wherein the plurality of activities records pertain to an event cluster associated with a suspicious event category, and the event cluster is generated through a clustering operation of command lines received by a plurality of computing devices; and

the event visualization device comprises:

a storage circuit, configured to store a graph generating program;

a control circuit, coupled to the storage circuit and configured to execute the graph generating program and perform a graph generating operation on the plurality of activities records in a recursive manner to generate a hierarchical directed acyclic graph (HDAG), wherein the HDAG comprises a hierarchical structure formed by connecting a plurality of common nodes and a plurality of branch nodes to correspondingly represent similar features and differential features of the plurality of activities records; and

a display device, coupled to the control circuit and configured to display the HDAG as a basis for diagnosing whether there are intrusions or abnormalities in the target network system,

wherein the graph generating operation comprises:

interpreting an activities record into a target DAG, wherein the target DAG comprises one or more nodes, each node represents a token, and each token comprises one or more characters;

performing a hierarchical partial order alignment (HPOA) operation on the target DAG and a reference DAG to obtain a merging condition of each node;

merging the target DAG and the reference DAG into the HDAG according to the merging condition; and

using the HDAG as the reference DAG of a next HPOA operation.

2. The event visualization device according to claim 1 , wherein the HPOA operation comprises:

performing a first partial order alignment (POA) operation on the target DAG and the reference DAG with tokens as an alignment unit to find out all aligned node groups;

performing a second POA operation on each aligned node group with character orders as the alignment unit to find out an optimal character alignment path from each aligned node group;

performing a merging condition analysis operation on each aligned node group by using the optimal character alignment path, and determining whether each aligned node group meets the merging condition;

merging aligned node groups that meet the merging condition into a common node to replace the original aligned node groups;

marking each aligned node group that does not meet the merging condition into a branch node;

recombining and drawing all common nodes and branch nodes into the HDAG; and

drawing the common node in the HDAG into a directed acyclic subgraph with characters as a node unit according to the optimal character alignment path.

3. The event visualization device according to claim 2 , wherein the first POA operation comprises: calculating a replacement score of each possible path pair of the target DAG and the reference DAG, and finding out all corresponding aligned node groups according to a path pair with a highest replacement score; and

the operation of calculating the replacement score for a possible path pair comprises: substituting a ratio of a longest common subsequence (LCS) value of two paths in the possible path pairs to a root of a product of a number of characters into an arcsine function to calculate a normalized longest common subsequence (NLCS) value of the two paths.

4. The event visualization device according to claim 2 , wherein the second POA operation performed on an aligned node group comprises:

converting a first node and a second node aligned therewith in the aligned node group into a first directed acyclic subgraph and a second directed acyclic subgraph with characters as a node unit;

calculating a longest common subsequence (LCS) value of each possible path pair of the first directed acyclic subgraph and the second directed acyclic subgraph; and

determining a path pair with the greatest LCS value as the optimal character alignment path of the aligned node group.

5. The event visualization device according to claim 2 , wherein the merging condition analysis operation comprises:

calculating a normalized longest common subsequence (NLCS) value of the optimal character alignment path; and

in a case that the NLCS value is greater than or equal to a similarity threshold α, determining that an aligned node group meets the merging condition,

wherein the calculation of the NLCS value comprises: substituting a ratio of a longest common subsequence (LCS) value of two paths in the optimal character alignment path to a root of a product of the number of characters into an arcsine function to calculate the NLCS value of the two paths.

6. A computer program product embodied in a non-transitory computer-readable storage medium, the computer program product allowing an event visualization device to perform a hierarchical directed acyclic graph (HDAG) generating operation to generate one or more HDAGs that can be used as a basis for diagnosing whether a target network system has been hacked, wherein a plurality of activities records pertain to an event cluster associated with a suspicious event category, and the event cluster is generated through a clustering operation of command lines received by a plurality of computing devices; and the event visualization device comprises a control circuit and a display device, wherein the HDAG generating operation comprises:

performing, by using the control circuit, a graph generating operation on the plurality of activities records in a recursive manner to generate an HDAG, wherein the HDAG comprises a hierarchical structure formed by connecting a plurality of common nodes and a plurality of branch nodes to correspondingly represent similar features and differential features of the plurality of activities records; and

displaying, by using the display device, the one or more HDAGs as a basis for diagnosing whether there are intrusions or abnormalities in the target network system,

wherein the graph generating operation comprises:

interpreting an activities record into a target directed acyclic graph (DAG), wherein the target DAG comprises one or more nodes, each node represents a token, and each token comprises one or more characters;

performing a hierarchical partial order alignment (HPOA) operation on the target DAG and a reference DAG to obtain a merging condition of each node;

merging the target DAG and the reference DAG into the HDAG according to the merging condition; and

using the HDAG as the reference DAG of a next HPOA operation.

7. The computer program product according to claim 6 , wherein the HPOA operation comprises:

performing a first partial order alignment (POA) operation on the target DAG and the reference DAG with tokens as an alignment unit to find out all aligned node groups;

performing a second POA operation on each aligned node group with character orders as the alignment unit to find out an optimal character alignment path from each aligned node group;

performing a merging condition analysis operation on each aligned node group by using the optimal character alignment path, and determining whether each aligned node group meets the merging condition;

merging aligned node groups that meet the merging condition into a common node to replace original aligned node groups;

marking each aligned node group that does not meet the merging condition into a branch node;

recombining and drawing all common nodes and branch nodes into the HDAG; and

drawing the common node in the HDAG into a directed acyclic subgraph with characters as a node unit according to the optimal character alignment path.

8. The computer program product according to claim 7 , wherein the first POA operation comprises: calculating a replacement score of each possible path pair of the target DAG and the reference DAG, and finding out all corresponding aligned node groups according to a path pair with the highest replacement score,

wherein the operation of calculating the replacement score for a possible path pair comprises: substituting a ratio of a longest common subsequence (LCS) value of two paths in the possible path pairs to a root of a product of a number of characters into an arcsine function to calculate a normalized longest common subsequence (NLCS) value of the two paths.

9. The computer program product according to claim 7 , wherein the second POA operation performed on an aligned node group comprises:

converting a first node and a second node aligned therewith in the aligned node group into a first directed acyclic subgraph and a second directed acyclic subgraph with characters as a node unit;

calculating a longest common subsequence (LCS) value of each possible path pair of the first directed acyclic subgraph and the second directed acyclic subgraph; and

determining a path pair with a greatest LCS value as the optimal character alignment path of the aligned node group.

10. The computer program product according to claim 7 , wherein the merging condition analysis operation comprises:

calculating a normalized longest common subsequence (NLCS) value of the optimal character alignment path; and

in a case that the NLCS value is greater than or equal to a similarity threshold α, determining that the aligned node group meets the merging condition,

wherein the calculation of the NLCS value comprises: substituting a ratio of a longest common subsequence (LCS) value of two paths in the optimal character alignment path to a root of a product of a number of characters into an arcsine function to calculate the NLCS value of the two paths.