Identifying Drone Web Sites in Multiple Countries and Languages with a Single Model

A text-based, bag-of-words, model was developed to identify drone company websites for multiple European countries in different languages. collection of Spanish and non-drone used initial development. Various classification methods were compared. Supervised logistic regression (L2-norm) performed best with an accuracy 87% on the unseen test set. The later improved 88% when it trained texts which all words translated into English. Retraining typical words, such as names cities regions, indicative specific periods time, months year days week, removed did not affect overall performance made more generally applicable. Applying cleaned, completely English word-based, a Irish Italian revealed, after manual inspection, that able detect those 82 86%, respectively. required creation translation list 1560 word-based features their analogs. Because had very high recall, 93, 100, 97% Spanish, respectively, particularly well suited select potential large collections websites.