Automatic Text Summarizing on Sports News Portals uses the Maximum Marginal Relevance method.

Abstract

News contains a fact or opinion that may hold some tendency. News can access various media such as newspapers, television, the internet, and others. The internet is the famous one used for accessing news. To find for the main information on the news may take time. It requires neat access and minimizes the time for reading. Therefore it is necessary to summarize the news so that the acquisition from the news is more efficient and effective. Research begins with five stages of text preprocessing: sentence solving, case folding, tokenizing, filtering, and stemming. The next process is calculating tf-idf-df weights, query relevance weights and similarity weights. Summaries are generated from sentence extraction using the maximum marginal relevance method. This method is used for reducing redundancy in ranking sentences on multiple documents. The effect of Maximum Marginal Relevance on the results of the system summary accuracy, namely, the test is taken from 5 online news news samples using lamda 0.7 which later the summary results are used to be compared with the summary of the system and the summary by experts includes an. Maya Rahma, and after that look for correct then look for wrong and missed, when these results are obtained then the precision, recall, f-measure of each respondent are sought and the test results will be obtained from looking for the average of precision, recall, f-measure, so that the lamda effect of 0.7 produces an average accuracy of 57.7% Precision. , Recall 48.5% and F - Measure 50.3%.