Tired of Topic Models? Clusters of Pretrained Word Embeddings Make for Fast and Good Topics too!

Topic models are a useful analysis tool to uncover the underlying themes\nwithin document collections. The dominant approach is to use probabilistic\ntopic models that posit a generative story, but in this paper we propose an\nalternative way to obtain topics: clustering pre-trained word embeddings while\nincorporating document information for weighted clustering and reranking top\nwords. We provide benchmarks for the combination of different word embeddings\nand clustering algorithms, and analyse their performance under dimensionality\nreduction with PCA. The best performing combination for our approach performs\nas well as classical topic models, but with lower runtime and computational\ncomplexity.\n