From my point of view, there are two definitions I find "definitive" and helpful. I hasten to add that I am neither a particle physicist nor quantum field theorist; my interest in the former comes from an application of Lie groups (which I am interested in), in the latter from a professional career in optics where I have had to gather a sound working knowledge of quantum optics and the second quantised electromagnetic field. But since you don't seem to be either, I'm guessing you might find the answer of someone who has muddled through similar questions helpful.
1. Wigner's Postulate
For a quantum system, a particle is an irreducible unitary representation of the Poincaré group. This statement is essentially Eugene Wigner's postulate - an axiom if you like - but it is very intuitive and makes sense. I'm guessing from your question that you haven't seen representation of groups in quantum state space before, so I'll try to give you a simple rundown and hopefully in the process show you why Wigner's postulate is a reasonable notion of a particle.
The postulates of special relativity can be essentially and pithily restated as: there is no experiment that two observers in two frames of reference can do to tell the difference between their two frames if the two frames are related to one another by either (1) a Euclidean isometry such as a rotation or translation or (2) if one is a "boosted" version of the other, i.e. is moving at a uniform relative velocity vector. The set of transformations that relate all such pairs of reference frames is the Poincaré group: you can think of it as a certain "group" (i.e. set closed under the operations of "product" and "inverse") of matrices that represent the co-ordinate transformations between pairs of co-ordinate systems. When you shift from a first inertial frame to a second, and then from the second to the third, your nett transformation matrix is the product of the transformation matrices between frames 1 and 2 and 2 and 3. To get the transformation from frame B to frame A, you simply take the inverse of the transformation from frame A to frame B.
So far so good. However, if we have quantum systems of whatever, the transformations between quantum states have to be unitary, i.e. it must conserve the "norm" of the quantum state vector owing to Born's probability interpretation of QM: the system has to end up in some state and all the probabilities must naturally sum unity. Moreover, many if not most interesting quantum systems are infinite dimensional. Something that is infinite dimensional and unitary is very different from spacetime co-ordinates: the Poincaré group is neither unitary and it transforms vectors with ten components (specifying an observers 4-position in spacetime together with their orientation). So the question is, how does the quantum state space transform if the postulates of special relativity hold: otherwise put, how must the quantum state space transform under Poincaré transformations if the physics has to be the same.
The precise condition for this to happen is that if a reference frame undergoes a sequence of transformations by the Poincaré transformations $P_1,\, P_2,\,\cdots$ and the corresponding quantum space undergoes the corresponding sequence of unitary transformations $U_1,\,U_2,\,\cdots$, then $U_j$ is some function of $P_j$ such that the transformations are compatible, i.e. if $U(P)$ is the relevant mapping, then our physics is unchanged if
$$U(P_1\,P_2) = e^{i\,\phi_{1,2}}\,U(P_1)\,U(P_2)$$
because then the unitary state transformations compose in the same way as the Poincaré transformations - almost- we allow a phase $e^{i\,\phi_{1,2}}$ to show itself because a quantum state's physics is unaffected by a global (as opposed to phase difference between the state components) phase.
We call such a mapping between transformations a representation of the Poincaré group.
Now some representations map the Poincaré group to operators $U$ that transform the whole of the quantum state space. But sometimes the representation is "more structured": it leaves subspaces of the quantum subspace intact - it might churn the innards of the subspace around, but the subspace as a whole stays fixed. It might be that the representation can be essentially split up into a whole set of parallel, independent representations that each act on different subspaces of the quantum space that otherwise leave each other alone. You can keep on splitting representations up in this way up to a point: in the end you'll get irreducible representations: it is impossible to split them up any further.
Once you get to this point, you have "atomic" subspaces of your quantum state space that are always transformed as a whole when Poincaré transformations act. You can't split these states up any more. These state subspaces are the states of the particles - the states whoes transformation cannot be broken down into parallel, simpler transformations.
2. Excitation of a Quantum Field
Let's confine our discussion to photons, since these are the most wonted to me. When discussing light and photons, we need to uderstand that there is only one object that begets and shows all the optical behaviours we witness: the second-quantised electromagnetic field.
The only things that are believed to be real in modern physics are this field and other quantum fields like it. There are only a handful of them. When we witness physical phenomena we are seeing interactions between these quantum fields.
The second quantised electromagnetic field can be thought of as a infinite gathering of quantum simple harmonic oscillators, one for each classical plane wave mode of Maxwell's equations. The eigenstates of quantum simple harmonic oscillators are discrete and they are evenly spaced by an amount of energy $h\,\nu$, where $\nu$ is the frequency of the oscillator in question. So each oscillator can change its state only discontinuously, by taking up or shedding a whole number multiple of this basic energy "chunk" $h\,\nu$. So the interactions of the electromagnetic field with the other quantum fields in the world is by way of these discrete packets. I like to think of these packets not so much as billiard balls but more like discrete data packets that are swapped between networks on the Internet, thus giving being to "stuff that happens" on the Internet. The quantum fields of the World talk to each other in discrete, chunky, communications, thus giving being to everything that we see happenning around us. These communications are the particles of the theory.
Note that it sometimes is not even meaningful to talk about where these particles are: they are simply excitations of the quantum field as a whole. They are nowhere in particular and everywhere all at once! For the quantum fields I spoke of are the space around us. We don't need to deal with the mysterious concept of a "void" any more in physics: empty space is nothing more than what we see when the quantum oscillators of the quantum fields of the World are all in their ground states!