I've always assumed like the others. The vector is maintained, velocity and direction.
The issue is how useful that vector will be in the target question.
As a "generic" simple rule, the "maintain velocity, random direction" is a nice hand wavy game mechanic, that's playable and useful.
From the hard science guys, there's a question of relative vector relationships.
Specifically, a ship with a velocity V, really has a velocity relative, most likely, the local star or other large gravity well. When a ship is landed on a planet, it has a velocity of 0, but that's relative to the planet. The planet has a velocity of PV relative to the star, and the star has a velocity of SV relative to it's galaxy, and on and on.
I don't even know if there's some universal constant reference for velocity. I also really have no idea how different the velocities of stars are, relative to the galactic core, but more importantly for this discussion, relative to each other.
From a mechanics point of view, we don't really consider this problem at all (for good reason). But if the vector is truly maintained, then it actually becomes an issue. System A may be moving dramatically "faster" compared to System B. The concern is whether that velocity becomes an issue in terms of matching vectors for the ship once it arrives.
In practice, it likely doesn't matter. I did a little simulation, and when you have a constant thrust 2G drive, especially with unlimited acceleration, i.e. thruster plates, then the planetary velocities are ALMOST zero in the big picture. By the time a ship gets from Earth to Jupiter, Jupiter has literally "hardly moved".
But you can that if you have a velocity of "0" in System A, and you arrive in System B, that "0" may now be 100,000km/hr or whatever because of the relative differences between the two systems. And, especially for fuel limited games (i.e. TNE), that could be an interesting consideration since they'll have to compensate for that velocity difference either by accelerating in the source system or in the destination system. The work is the same, just a matter of where it's done.
Some quick google fu brings up
http://www.enchantedlearning.com/subjects/astronomy/planets/earth/Speeds.shtml
Our system is moving at 155 miles/sec vs the Galactic core, and the Galaxy is moving 185 miles/sec. Not insane velocities, but in TNE from "0" to 155 miles/sec is 14 G Turns of acceleration. So, there is room "adventure" here in fuel starved environments.
I imagine the relative velocities of systems is pretty minor. Just like the relative velocities of our planets are pretty minor, and the stars pretty much all go in a similar direction around the galaxy.
Still, fun thought experiment.