Method and system for the on-demand generation of graph-like models out of multidimensional observation data

A computer-implemented method for automatically detecting vertical relationships between computing entities in a distributed computing environment, comprising:

• receiving, by a computer processor, a first value for a candidate parent computing entity in the distributed computing environment;
• retrieving, by the computer processor, a first index entry corresponding to the candidate parent computing entity from a storage medium, such that the first index entry includes a plurality of data identifiers, where each data identifier is derived from a set of values describing two or more computing entities in the distributed computing environment and at least one computing entity in the two or more computing entities is the candidate parent computing entity;
• receiving, by the computer processor, a second value for a candidate child computing entity in the distributed computing environment;
• retrieving, by the computer processor, a second index entry corresponding to the candidate child computing entity from a storage medium, such that the second index entry includes a plurality of data identifiers, where each data identifier is derived from a set of values describing two or more computing entities in the distributed computing environment and at least one computing entity in the two or more computing entities is the candidate child computing entity;
• determining, by the computer processor, cardinality of data identifiers in the second index entry;
• determining, by the computer processor, cardinality of data identifiers in an intersection of the data identifiers in the first index entry and the data identifiers in the second index entry;
• dividing, by the computer processor, cardinality of intersection by cardinality of the second index entry to yield an overlap score;
• comparing, by the computer processor, the overlap score to a threshold;
• and
• identifying, by the computer processor, a parent-child relationship between the candidate parent computing entity and the candidate child computing entity in response to the overlap score exceeding the threshold.