Exploiting Knowledge Graph to Improve Text-based Prediction

As a special kind of "big data," text data can be regarded as data reported by human sensors. Since humans are far more intelligent than physical sensors, text data contains useful information and knowledge about the real world, making it possible to make predictions about real-world phenomena based on text. As all application domains involve humans, text-based prediction has widespread applications, especially for optimization of decision making. While the problem of text-based prediction resembles text classification when formulated as a supervised learning problem, it is more challenging because the variable to be predicted may not be directly derivable from the text and thus there is a semantic gap between the target variable and the surface features that are often used for representing text data in conventional approaches. In this paper, we propose to bridge this gap by using knowledge graph to construct more effective features for text representation. We propose a two-step filtering algorithm to enhance such a knowledge-aware text representation for a family of entity-centric text regression tasks where the response variable can be treated as an attribute of a group of central entities. We evaluate the proposed algorithm by using two revenue prediction tasks based on reviews. The results show that the proposed algorithm can effectively leverage knowledge graphs to construct interpretable features, leading to significant improvement of the prediction accuracy over traditional features.

I. Introduction

The rapid and continuous growth of online text data has created an unprecedented opportunity to mine big text data to discover knowledge and support many applications especially for optimization of decision making. Among many applications of text data mining, prediction of interesting variables based on relevant text data is especially interesting because it allows decision makers to "see" patterns behind the data that they would otherwise not be able to see or easily see, thus leading to better decision making. Since humans are involved in every application domain, they would generate text data in every application domain, which we can leverage for various prediction tasks, making text-based prediction useful in virtually all the application domains. Text-based prediction can also facilitate a variety of downstreaming applications [1], [2], [3].