TEKNIK REALTIME DATA PROCESSING UNTUK MONITORING TWEET MENGGUNAKAN ANALISIS SENTIMEN DENGAN ALGORITMA NAIVE BAYES STUDI KASUS TWEET YANG TERKAIT INSTITUSI UGM; REALTIME DATA PROCESSING TECHNIQUE TO MONITOR TWEETS USING SENTIMENT ANALYSIS WITH NAIVE BAYES ALGORITHM A CASE STUDY OF TWEETS THAT RELATED TO UGM

WIDIANTO, TEGUH PUJI, N

2016 | Skripsi | FMIPA

Twitter.com has aproximately 316 million active users montly and more than 500 million tweets are published every day. More specific in Indonesia, Country that have aproximately 255 Million citizen is country that has most active user in the world to use Twitter. Indonesia is expected more than 50 Million have Twitter account, so Twitter is most influence social media in indonesian people. Many cases become national effect because published in Twitter. Therefore, It is required spesific attention to minimalize impact from using Twitter. One of way to solve the problem is monitor the indonesian tweet that relevan with institution in realtime. To overcome these obstacles, the study aims to create monitoring tools that use technique of realtime data processing. The result of data processing tweets were classified with Sentiment analisys before tweets are visualized in web browser realtimely. The data processing is started when application filter tweets with specific keyword to get tweets that related with UGM institution. Storm library is used to process filtered tweet in realtime. The result of processing tweet is classified into tweets containing the positive sentiment, negative sentiment and neutral sentiment with Naive Bayes Algorithm. Before it used Naive Bayes algorithm, application need training data that save in Redis Database. Twitter4J, Apache Storm and Redis Database can be used to process data tweet in realtime. make data processing tool that can monitor tweet in realtime. Websocket protokol and NodeJs can visualize result of processing data in realtime with lower latency.

Kata Kunci : sentimen analysis, realtime processing, public relation monitoring