Methods and systems for clustering files

A method of clustering files, comprising, by a processing unit:

• obtaining a plurality of data (Dsignal, 1, . . . , Dsignal, Z) representative of a plurality of files (Dfile, 1, . . . , Dfile, Z) to be clustered,
• building a clustering structure comprising a plurality of nodes arranged in hierarchical levels Li, with i from 1 to N, wherein each node is representative of a category of files, wherein said category is representative of files sharing similarities,
• wherein said building comprises, based on said plurality of data (Dsignal, 1, . . . , Dsignal, Z):
• obtaining one or mode nodes of level L1, wherein each node is associated with an acceptance threshold and a centroid representative of files belonging to a category represented by said node,
• performing repetitively, for i=2 to N:
• building one or more nodes of level Li, wherein each node of level Li is linked to a parent node of level Li−1, with i from 2 to N, thereby indicating that each file belonging to a category represented by said node also belongs to a category represented by said parent node,
• wherein each node is associated with at least one acceptance threshold and at least one centroid representative of files belonging to a category represented by said node,
• wherein, for each node, said acceptance threshold and said centroid are usable for defining which file belongs to category represented by said node,
• wherein, during said building, at least part of said plurality of data (Dsignal, 1, . . . , Dsignal, Z) or of said plurality of files (Dfile, 1, . . . , Dfile, Z) is each associated with one or more of the nodes of the clustering structure, thereby reflecting that said node is representative of a category of files and facilitating future identification of a category of a file based on said clustering structure.

A system for clustering files, comprising, by a processing unit:

• obtain a clustering structure comprising a plurality of nodes arranged in hierarchical levels Li, with i from 1 to N,
• wherein each node is representative of a category of files, wherein said category is representative of files sharing similarities,
• wherein each node of level Li is linked to a parent node of level Li−1, with i from 2 to N, thereby indicating that each data belonging to a category represented by said node also belongs to a category represented by said parent node,
• wherein each node is associated with at least one acceptance threshold, wherein each node is associated with at least one centroid representative of files belonging to a category represented by said node,
• obtain at least one data (Dsignal) representative of a file (Dfile) to be assigned to a category;
• (O1) compare said data to each centroid of each node of the first level,
• (O2) if said comparison matches the acceptance threshold of one or more nodes, select a node among these nodes,
• (O3) compare Dsignal to each centroid of each node of a next level which is linked to said selected node,
• (O4) if said comparison matches the acceptance threshold of one or more nodes, select a node among these nodes,
• repeat O3 and O4 until a stopping condition is met, thereby indicating that said data Dsignal or said file Dfile belongs to a category of files represented by said selected node.

A system for clustering files, comprising, by a processing unit:

• obtain a plurality of data (Dsignal, 1, . . . , Dsignal, Z) representative of a plurality of files (Dfile, 1, . . . , Dfile, Z) to be clustered,
• build a clustering structure comprising a plurality of nodes arranged in hierarchical levels Li, with i from 1 to N, wherein each node is representative of a category of files, wherein said category is representative of files sharing similarities,
• wherein said building comprises, based on said plurality of data (Dsignal, 1, . . . , Dsignal, Z):
• obtaining one or mode nodes of level L1, wherein each node is associated with an acceptance threshold and a centroid representative of files belonging to a category represented by said node,
• performing repetitively, for i=2 to N:
• building one or more nodes of level Li, wherein each node of level Li is linked to a parent node of level Li−1, with i from 2 to N, thereby indicating that each file belonging to a category represented by said node also belongs to a category represented by said parent node,
• wherein each node is associated with at least one acceptance threshold and at least one centroid representative of files belonging to a category represented by said node,
• wherein, for each node, said acceptance threshold and said centroid are usable for defining which file belongs to the category represented by said node,
• wherein, during said building, at least part of said plurality of data (Dsignal, 1, . . . , Dsignal, Z), or of said plurality of files (Dfile, 1, . . . , Dfile, Z) is each associated with one or more of the nodes of the clustering structure, thereby reflecting that said node is representative of a category of files and facilitating future identification of a category of a file based on said clustering structure.

A non-transitory storage device readable by a machine, tangibly embodying a program of instructions executable by the machine to perform a method of clustering files comprising:

• obtaining a clustering structure comprising a plurality of nodes arranged in hierarchical levels Li, with i from 1 to N,
• wherein each node is representative of a category of files, wherein said category is representative of files sharing similarities,
• wherein each node of level Li is linked to a parent node of level Li−1, with i from 2 to N, thereby indicating that each data belonging to a category represented by said node also belongs to a category represented by said parent node,
• wherein each node is associated with at least one acceptance threshold, wherein each node associated with at least one centroid representative of files belonging to a category represented by said node,
• obtaining at least one data (Dsignal) representative of a file (Dfile) to be assigned to a category;
• (O1) comparing said data to each centroid of each node of the first level,
• (O2) if said comparison matches the acceptance threshold of one or more nodes, selecting a node among these nodes,
• (O3) comparing Dsignal to each centroid of each node of a next level which is linked to said selected node,
• (O4) if said comparison matches the acceptance threshold of one or more nodes, selecting a node among these nodes,
• repeating O3 and O4 until a stopping condition is met, thereby indicating that said data Dsignal or said file Dfile belongs to a category of files represented by said selected node.

A non-transitory storage device readable by a machine, tangibly embodying a program of instructions executable by the machine to perform a method of clustering files comprising:

• obtaining a plurality of data (Dsignal, 1, . . . , Dsignal, Z) representative of a plurality of files (Dfile, 1, . . . , Dfile, Z) to be clustered,
• building a clustering structure comprising a plurality of nodes arranged in hierarchical levels Li, with i from 1 to N, wherein each node is representative of a category of files, wherein said category is representative of files sharing similarities,
• wherein said building comprises, based on said plurality of data (Dsignal, 1, . . . , Dsignal, Z):
• obtaining one or mode nodes of level L1, wherein each node is associated with an acceptance threshold and a centroid representative of files belonging to a category represented by said node,
• performing repetitively, for i=2 to N:
• building one or more nodes of level Li, wherein each node of level Li is linked to a parent node of level Li−1, with i from 2 to N, thereby indicating that each file belonging to a category represented by said node also belongs to a category represented by said parent node, wherein each node is associated with at least one acceptance threshold and at least one centroid representative of files belonging to a category represented by said node,
• wherein, for each node, said acceptance threshold and said centroid are usable for defining which file belongs to category represented by said node,
• wherein, during said building, at least part of said plurality of data (Dsignal, 1, . . . , Dsignal, Z) or of said plurality of files (Dfile, 1, . . . , Dfile, Z) is each associated with one or more of the nodes of the clustering structure, thereby reflecting that said node is representative of a category of files and facilitating future identification of a category of a file based on said clustering structure.