Figure 6.

A weighting example. Here is an example of a context (t₁,..., t₁₆). The term t is the name to be disambiguated. Figure illustrates the weights of the context words t₁,..., t₁₆ with two different parameter value pairs of λ and β. The weight values are represented as a continuous function, although they take discrete values. The parameter combination λ = 1.5 and β = 0.025 yields the best performance when using the linear kernel.