AvaliaGeo: Sistema para Validação de Topônimos em Notícias
Abstract
Solutions to extracting geographic information problems from texts and documents often need labeled databases to carry out experiments or validate algorithms. However, many of these databases are not costless made available for use. This work aims to facilitate the generation of geographically labeled databases, using voluntary contributions to disambiguate toponyms present in the news. We propose using the Cronbach's Alpha coefficient to validate the contributions, considering each news item a questionnaire and each toponym candidate a questionnaire item. Preliminary experiments achieved 70% reliability in the disambiguation of toponyms for database generation.
References
Cronbach, L. J. (1951). Coefficient alpha and the internal structure of tests. psychometrika, 16(3):297–334.
Freitas, A. L. P. and Rodrigues, S. G. (2005). A avaliação da confiabilidade de questionários: uma análise utilizando o coeficiente alfa de cronbach. In Anais... XVII SIMPEP,.
Gritta, M., Pilehvar, M. T., Limsopatham, N., and Collier, N. (2018). What’s missing in geographical parsing? Language Resources and Evaluation, 52(2):603–623.
Larsen, N. (2010). Market segmentation a framework for determining the right target customers. Bachelor’s thesis, Aarhus School of Business, Aarhus BSS, Denmark.
Matthiensen, A. (2010). Uso do coeciente alfa de cronbach em avaliações por questionários. Embrapa Roraima-Documentos (INFOTECA-E).
Monteiro, B. R., Jr., C. A. D., and Fonseca, F. T. (2016). A survey on the geographic scope of textual documents. Computers & Geosciences, 96:23–34.
Streiner, D. L. (2003). Starting at the beginning: an introduction to coefficient alpha and internal consistency. Journal of personality assessment, 80(1):99–103.
