Heracles: A framework for developing and evaluating text mining algorithms. (1st August 2019)

• Record Type: Journal Article
• Authors: Schouten, Kim; Frasincar, Flavius; Dekker, Rommert; Riezebos, Mark
• Is Part Of: Expert systems with applications. Volume 127(2019)
• Page Start: 68
• Page End: 84
• Publication Date: 2019-08-01
• Subjects: Text mining -- Algorithm evaluation -- Research and development -- Developers framework; Expert systems (Computer science) -- Periodicals; Systèmes experts (Informatique) -- Périodiques; Electronic journals; 006.33
• DOI: 10.1016/j.eswa.2019.03.005
• Languages: English
• ISSNs: 0957-4174

Abstract

Highlights: We propose a unified framework for developing text mining. It integrates packages for natural language processing and machine learning. It has built-in support for standard research evaluation practices. The design patterns and technical details are given and code is available at GitHub. The framework has successfully been used in various text mining projects. Abstract: Many of today's businesses are driven by data, and while traditionally only quantitative data is considered, the role of textual data in our digital world is rapidly increasing. Text mining allows to extract and aggregate numerical data from textual documents, which in turn can be used to improve key decision processes. In this paper, we propose Heracles, a framework for developing and evaluating text mining algorithms, with a broad range of applications in industry. In contrast to other frameworks, Heracles supports both the development and evaluation stages of text mining algorithms. A practical use case shows the versatility and ease-of-use of the proposed framework in the domain of aspect-based sentiment analysis.