The knot group is the fundamental group of the knot complement in $S^{3} $. The Dehn presentation of the knot group is a particular group presentation obtained by looking at the regions and crossings of a knot (see http://ncatlab.org/nlab/show/knot%20group). Is it possible to look at an arbitrary group presentation and know whether or not it is a Dehn presentation of a knot group?

Let's call your candidate knot group $G$. Also, I will take "look at" to mean find easily computable invariants and or obstructions. There are obvious obstructions that you might consider to help determine if $G$ could be a knot group. Many of these obstructions are more apparent from the Wirtinger presentation, which is described below the Dehn presentation in your reference. The Wirtinger presentation is generated by conjugate elements or rather normally generated by one element. Since knot groups are torsion free, the abelianization of $G$ is $Z$. Furthermore, using the Wirtinger presentation, adding the relation that one generator is trivial makes the whole group trivial. Those are two obvious necessary and often easily observable criterion for presentations involving few generators and short relations. However, if this group is showing up as the fundamental group of a 3manifold with torus boundary more can be said. Refine our assumptions so $G$ be the fundamental group of a manifold $M$ with a single torus boundary component $T$ and let $\mu \in \pi_1(T)$. By Perelman's solution to the Poincaré conjecture and a simple argument using Dehn filling (see Rolfsen's Knots and Links for background on Dehn filling), $G/<<\mu>>$ is trivial if and only if $M$ is a knot exterior in $S^3$ (i.e. $M \cong S^3  n(K)$, where $n(K)$ denotes the open neighborhood of an embedded knot). Checking that $\mu \in \pi_1(T)$ is necessary here. The key point here is that there are lots of groups with abelianization $Z$ that are normally generated by one element that are not knot groups. A large family of examples of 3manifold groups with this property can be constructed by taking 0/1 surgery along a (homologically framed) knot in $S^3$. With more work, one can find examples of such $\pi_1(M)$ where $M$ is a manifold with torus boundary having this property. Using Snappy, we can see the manifold 'm008' in the cusped census of hyperbolic 3manifolds by Callahan, Hildebrand and Weeks has $\pi_1(m008)=\langle a,b  a^{2}b^{2}a^{1}ba^{1}b^{2}\rangle$. Introducing the relation, $a=1$ makes the whole group trivial. In this case, $a$ does not correspond to an element in $\pi_1(\partial M)$ by trace considerations. To complete the argument, Snappy does not identify 'm008' as a CensusKnot, and so 'm008' has a fundamental group normally generated by one element such that $H_1(m008,Z)\cong Z$, but m008 is not a knot complement. 

