Building a graph index and searching a corresponding dataset

A method, comprising:

• identifying a first subset of points from a plurality of subsets of points from a dataset of points;
• generating an initial graph representative of the first subset of points, the initial graph including a first set of vertices associated with the subset of points;
• generating a first search graph for the first subset of points by iteratively applying a search algorithm for each vertex from the first set of vertices and updating the first search graph by adding out-neighbors between vertices of the first set of vertices to ensure that each vertex from the first set of vertices includes a navigable path from each vertex to a start vertex while reducing a number of hops between vertices of the first set of vertices;
• generating a combined search graph representative of the dataset of points by combining the first search graph with additional search graphs representative of additional subsets of points from the plurality of subsets of points;
• and
• storing the combined search graph on a storage of a computing device.

A system, comprising:

• one or more processors;
• a memory in electronic communication with the one or more processors;
• and
• instructions stored in the memory, the instructions being executable by the one or more processors to:
• generate a random graph representative of a dataset of points, the random graph including a set of vertices associated with the dataset of points and randomly assigned out-neighbor data for each vertex of the set of vertices;
• generate a first iteration of a search graph by iteratively applying a search algorithm for each vertex of the set of vertices and updating the random graph by modifying the randomly assigned out-neighbor data to ensure that each vertex of the set of vertices includes a navigable path from each vertex to a start vertex;
• generate an updated search graph by iteratively applying the search algorithm for each vertex of the set of vertices and refining out-neighbor data from the first iteration of the graph to reduce a number of hops between vertices of the set of vertices;
• and
• store the updated search graph for storage on a computing device.