ハリケーン・マシューのこと何て言ってる?英語のツイートが伝えるのは状況の60%だけ
人々の声を正確かつ十分に活かすには、英語だけでなく、どの言語のツイートもすべて分析しなければなりません。
ツイッターでハリケーンの話題を追跡するには、対象地域で使われる言語のハリケーンを意味する言葉で多言語検索し、それらのツイートをそれぞれの記述言語で分析する必要があります。
El Sentimiento Real sobre Huracán Matthew
October 7, 2016
What are people saying about Hurricane Matthew? English tweets only give you 60% of the picture.
As the winds of Hurricane Matthew march up the Florida coast, real-time social media monitoring is the only reporting that matches it for speed. Armed with the power of smartphones, the public has increasingly become the front line of real-time, updates on Matthew’s progress, direction and destruction.

Gather the data…ALL the data
To accurately and thoroughly harness the value of the voice of the public, your text analysis must handle the diverse languages represented within it.
The bulk of the storm’s destruction so far has been in the Caribbean, where there are six official languages. Even in Florida, nearly 20% of the metropolitan Orlando population speaks primarily or only Spanish at home, and in the Jacksonville area, that number is closer to 40%. Searching only for “Hurricane” on Twitter will leave out a wealth of firsthand accounts tweeting about “huracán,” “orkaan,” or “siklòn—that would be picked up by a multilingual search.
Analyze Spanish sentiments in Spanish for the best results
Access to all the data is worthless without machine-speed analysis. In the past, machine translating multilingual text to English and then performing automated linguistic analysis was that only way, but that often stripped any nuances. Today, cutting-edge tools perform linguistic analysis on news, reviews, blogs, tweets, and posts in the language of the original text. With this technology, every word is analyzed in its native context, delivering analysis un-skewed by subtle errors in slang, syntax, and spelling.
Officials have advised nearly 2 million residents of Florida, South Carolina, and Georgia to evacuate their homes. However, not all of these evacuations are mandatory (just ask rapper Vanilla Ice). By tracking publically accessible social media posts about the storm in real time government officials and meteorologists can more accurately track and predict Matthew’s movement and more quickly revise evacuation plans and recommendations. Likewise, data gleaned from social media posts can also be crucial in the aftermath.
Using Rosette API’s English and Spanish sentiment analysis on tweets from residents of affected areas, the government, news agencies, and aid organizations could track where need is most acute…
…which regions are in good shape…
…and which devastated areas are now safe to send in aid workers.
Is your text analytics up for the job?
Whether you work in an enterprise, government, or an NGO, we guarantee that there are issues being discussed on publicly accessible social media that affect your work. To get the whole picture, you have to be able to analyze all the data.
What is it like to access all relevant data in all the languages? Check it out. We’ve created an open source Rosette plugin for analyzing sentiment on Twitter and Facebook. Sign up for an API account at developer.rosette.com—our default plan gives you 10,000 calls a month for free. Download the plugin here. If you come up with any interesting results, let us know! We love to feature user stories on our blog.