Mixed Graph of Terms: Beyond the Bags of Words Representation of a Text

摘要

The main purpose of text mining techniques is to identify common patterns through the observation of vectors of features and then to use such patterns to make predictions. Vectors of features are usually made up of weighted words, as well as those used in the text retrieval field, which are obtained thanks to the assumption that considers a document as a "bag of words". However, in this paper we demonstrate that, to obtain more accuracy in the analysis and revelation of common patterns, we could employ (observe) more complex features than simple weighted words. The proposed vector of features considers a hierarchical structure, named a mixed Graph of Terms, composed of a directed and an undirected sub-graph of words, that can be automatically constructed from a small set of documents through the probabilistic Topic Model. The graph has demonstrated its efficiency in a classic "ad-hoc" text retrieval problem. Here we consider expanding the initial query with this new structured vector of features.