ASABE Technical Library - Abstract

Click on “Download PDF” for the PDF version or on the title for the HTML version.
If you are not an ASABE member or if your employer has not arranged for access to the full-text, view access and subscription options

DiagText: a text mining tool for agriculture

Published by the American Society of Agricultural and Biological Engineers, St. Joseph, Michigan www.asabe.org

Citation:  7th World Congress on Computers in Agriculture Conference Proceedings, 22-24 June 2009, Reno, Nevada (electronic only)  711P0409e.(doi:10.13031/2013.29107)
Authors:   Silvia M. F. S Massruhá, Kleber X. S Souza, Helano P. de Lima, Raphael Ricciotti, André D. L Zanchetta
Keywords:   knowledge acquisition, machine-learning, clustering, text-mining, plant diseases.

In this paper a tool called Diagtext is presented. This tool aims to help the process of extraction of information from textual documents by identifying groups of similar documents in such a way that an expert can decide which category a document would better fit in. There are two main approaches to extract information from textual data. The first uses the semantic analysis that is based on the functionality of words in the corpora and the second uses statistical analysis based on frequency of terms. These approaches can be used together to analyze data in what is called a hybrid approach. In the scope of this work was adopted the statistical approach. DiagText allows the extraction of information from unstructured data in plain text format(.txt) or in PDF format. From the text, groups were generated and the trees were compared with the trees generated by experts in the field. Although the initial validation is on agriculture, DiagText can be used in several domains, since it was developed to be language and subject independent. Finally, future work using an approach that incorporates fuzzy clustering techniques is also discussed in this paper.

(Download PDF)    (Export to EndNotes)