Diffusion-limited aggregation (DLA) is a model of fractal growth. Computer models can simulate the fast aggregation of millions of particles. We propose a modified version of DLA, called semantically modified DLA (SM-DLA), for visualizing large-scale networks. SM-DLA introduces similarity measures between particles so that instead of attaching to the nearest particle in the aggregation, a new particle is stochastically directed to attach to particles that are similar to it. The results of our initial experiment with a co-citation network using SM-DLA are encouraging, suggesting that the algorithm has the potential as an alternative paradigm for visualizing large-scale networks. Further studies in this direction are recommended.