Method described on the paper Topic Modeling for Short Texts with Cooccurrence Frequency-based Expansion published in the Proceedings of Brazilian Conference on Intelligent Systems (BRACIS) 2016. Work developed together with Marcelo Pita, Paulo Bicalho, Anisio Lacerda and Gisele L. Pappa.
CoFE generates pseudo-documents from a given text collection which are larger than the original ones. It does so by creating a similarity graph, which is generated exploiting word cooccurrence information found in the dataset. Then, for each document, CoFE explores the graph to select good words to be appended to the document to generate its larger version.
Short texts are everywhere on the Web, including messages in social media, status messages, etc, and extracting semantically meaningful topics from these collections is an important and difficult task. Topic modeling methods, such as Latent Dirichlet Allocation, were designed for this purpose. However, discovering high quality topics in short text collections is a challenging task. This is because most topic modeling methods rely on information coming from the word co-occurrence distribution in the collection to extract topics. As in short text this information is scarce, topic modeling methods have difficulties in this scenario, and different strategies to tackle this problem have been proposed in the literature. In this direction, this paper introduces a method for topic modeling of short texts that creates pseudo-documents representations from the original documents. The method is simple, effective, and considers word co-occurrence to expand documents, which can be given as input to any topic modeling algorithm. Experiments were run in four datasets and compared against state-of-the-art methods for extracting topics from short text. Results of coherence, NPMI and clustering metrics showed to be statistically significantly better than the baselines in the majority of cases.