How to Improve?

So I'm working on a project that solves the Kevin Bacon problem. The code works fine for a small file size, but I have to read in the entire IMDB movie database file, which contains thousands of movies. I have a graph where the vertices are the actors. I then create edges between the actors if they were in the same movie. However, I think my problem lies in creating the vertices because it is creating a string for each vertex, and so it quickly runs out of memory.

Code:

#ifndef GRAPHSTUFF_H_
#define GRAPHSTUFF_H_

#include <iostream>
#include <string>
#include <sstream>
#include <fstream>
#include <list>
#include <vector>

using namespace std;

struct Vertex;

struct Edge
{
                     // First vertex in edge is implicit
    Vertex  *dest;   // Second vertex in edge
    double   cost;   // Edge cost

    Edge( Vertex *d = 0, double c = 0.0 )
      : dest( d ), cost( c ) { }
};


// Structure stored in priority queue for Dijkstra's algorithm.
struct Path
{
    Vertex *dest;   // w
    double  cost;   // d(w)

    Path( Vertex *d = 0, double c = 0.0 )
      : dest( d ), cost( c ) { }

    bool operator> ( const Path & rhs ) const
      { return cost > rhs.cost; }
    bool operator< ( const Path & rhs ) const
      { return cost < rhs.cost; }
};

const int INFINITY = std::numeric_limits<int>::max();

// Basic info for each vertex.
struct Vertex
{
    std::string name;    // Vertex name
    vector<Edge>  adj;     // Adjacent vertices (and costs)
    double        dist;    // Cost
    Vertex       *prev;    // Previous vertex on shortest path
    int           scratch; // Extra variable used in algorithm

    Vertex( const std::string & nm ) : name( nm )
      { reset( ); }

    void reset( )
      { dist = INFINITY; prev = NULL; scratch = 0; }

  };

#endif /*GRAPHSTUFF_H_*/

Is there any way to improve this?

The question is what you really expect to optimize? I mean, you somewhere have to store the information and that space is limited (of course at some point the processor starts page swapping etc).

If you are sure that you're not creating a vertex twice for the same actor or that your program is allocating unnecessary space, then I can't see what you could optimize.

I hope you realize that you don't have to read "the entire imdb database", but simply if you start with Kevin bacon (for example) and build your graph from there, you will encounter a situation where you can't add any more actors because you've got them already.

Unfortunately for you, I heard some vague theory that it is possible to link any two actors together with a chain of only 4 intermediate nodes (with a working proof of concept on some website).