Extraction of Event Sentence Information in the Covid-19 Distribution Location Detection System based on the Indonesian Language Corpus

The procedure for obtaining data on people affected by the COVID-19 disease is carried out through a manual data collection mechanism through health centers, government and private hospitals, health clinics spread throughout Indonesia, and rapid tests carried out at certain times and locations. Such a surveillance system requires time, a lot of health personnel, and is expensive. In addition, the geographical condition of the Indonesian state, which consists of many large and small islands and the vast territory of Indonesia requires another strategy to find out and make it easier to complete data on people affected by COVID-19, such as the use of information technology. The use of the Twitter dataset to detect the spread of disease in a region or country has been widely carried out by researchers. Sentence extraction is a process that must be done to facilitate the analysis of a short or very long news sentence to get the meaning and essence of the news contained in the sentence. The primary information can be identified based on keywords generated using extraction and abstraction techniques. The initial stage of the research focused on building a corpus of twitter data and a corpus of vocabulary. The first process that will be carried out is to collect natural language datasets from Indonesian-language Tweets on Twitter. Next, carry out the process of extracting incident sentence information with steps, namely making standard sentence formats, simplifying sentences, identifying essensial words, and determining input and target words in sentences.