eunyhwang/epit_sesotho

This repository is to document dataset, methods, and results of the NaLaA project. # Project Background NaLaA Project (Natural Language Processing for Event-based Surveillance with Africa CDC) aims to improve event-based surveillance by utilizing epitweetr, the R-based tool developed by European CDC, and applying NLP techniques for African CDC. Event-based surveillance[^1] is one of two main public health surveillances; event-based surveillance and indicator-based surveillance. Both approachs intend to detect potential disease events from sources, but event-based surveillance is based on unofficial and unstructured sources such as media or social media whereas indicator-based surveillance is based on official and structured sources such as reports from public health providers or medical examination. In this project, we focus on event-based surveillance and try to improve it by detecting public health threats on tweetr data. The European Centre for Disease Prevention and Control (ECDC) developed an amazing early warning tool, named epitweetr, using data from Twitter, which collects, geolocates and aggregates tweets generating signals and email alerts.[^2] Epitweetr already provides more than 100 languages to detect public health threats. As our target language, Sesotho, is not one of them, we trained our own word embedding, updated the tool, and created an API for those who also want to use a minor language with epitweetr. [^1]: cdc.gov [^2]: eurosurveillance.org # Prerequisites Before you run epitweetr, you need to get **a Twitter account** as well as below-mentioned softwares. Please be aware of that you need to install the correct version of each software. For the further information about getting a Twitter account and the softwares, check out …