| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 986.77 KB | Adobe PDF |
Advisor(s)
Abstract(s)
Economic and Food Safety Authority receives on a daily basis reports and complaints regarding infractions, delicts and possible food and economic crimes. These reports and complaints can be in different forms, such as e-mails, online forms, letters, phone calls and complaint books present in every establishment. This paper aims to apply text mining and classification algorithms to textual data extracted from these reports and complains in order to help identify if the responsible entity to analyze the content is, in fact, the Economic and Food Safety Authority. The paper describes text preprocessing and feature extraction procedures applied to Portuguese text data. Supervised multi-class classification methods such as Naïve Bayes and Support Vector Machine Classifiers are employed in the task. We show that a non-semantical text mining approach can achieve good results, scoring around 70% of accuracy.
Description
Keywords
Text mining Economic and food safety Natural language processing Text classification Multi-class classification
Citation
Magalhães, G., Faria, B. M., & Cardoso, L. P. R. and H. L. (2019). Text mining applications to facilitate economic and food safety law enforcement (A. P. Abraham & J. Roth, Eds.; pp. 199–203). https://www.iadisportal.org/digital-library/text-mining-applications-to-facilitate-economic-and-food-safety-law-enforcement